0% found this document useful (0 votes)

47 views586 pages

State of The Art in Partial Least Squares Structural Equation Modeling (PLS-SEM)

Uploaded by

yassinekaa10

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

47 views586 pages

State of The Art in Partial Least Squares Structural Equation Modeling (PLS-SEM)

Uploaded by

yassinekaa10

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 586

Springer Proceedings in Business and Economics

Lăcrămioara Radomir ·
Raluca Ciornea · Huiwen Wang ·
Yide Liu · Christian M. Ringle ·
Marko Sarstedt Editors

State of the Art

in Partial Least
Squares Structural
Equation Modeling
(PLS-SEM)
Methodological Extensions and
Applications in the Social Sciences and
Beyond
Springer Proceedings in Business and Economics
Springer Proceedings in Business and Economics brings the most current research
presented at conferences and workshops to a global readership. The series features
volumes (in electronic and print formats) of selected contributions from conferences
in all areas of economics, business, management, and finance. In addition to an
overall evaluation by the publisher of the topical interest, scientific quality, and
timeliness of each volume, each contribution is refereed to standards comparable to
those of leading journals, resulting in authoritative contributions to the respective
fields. Springer’s production and distribution infrastructure ensures rapid publication
and wide circulation of the latest developments in the most compelling and prom-
ising areas of research today.
The editorial development of volumes may be managed using Springer Nature’s
innovative EquinOCS, a proven online conference proceedings submission, man-
agement and review system. This system is designed to ensure an efficient timeline
for your publication, making Springer Proceedings in Business and Economics the
premier series to publish your workshop or conference volume.
This book series is indexed in SCOPUS.
Lăcrămioara Radomir • Raluca Ciornea •
Huiwen Wang • Yide Liu • Christian M. Ringle •
Marko Sarstedt
Editors

State of the Art in Partial

Least Squares Structural
Equation Modeling
(PLS-SEM)
Methodological Extensions and Applications
in the Social Sciences and Beyond
Editors
Lăcrămioara Radomir Raluca Ciornea
Faculty of Economics and Business Faculty of Economics and Business
Administration Administration
Babeș-Bolyai University Babeș-Bolyai University
Cluj-Napoca, Romania Cluj-Napoca, Romania

Huiwen Wang Yide Liu

School of Economics and Management School of Business
Beihang University Macau University of Science and Technology
Beijing, China Taipa, Macao

Christian M. Ringle Marko Sarstedt

Department of Management Sciences and Ludwig-Maximilians-University, Munich,
Technology Germany
Hamburg University of Technology Babeș-Bolyai University
Hamburg, Germany Cluj-Napoca, Romania

ISSN 2198-7246 ISSN 2198-7254 (electronic)

Springer Proceedings in Business and Economics
ISBN 978-3-031-34588-3 ISBN 978-3-031-34589-0 (eBook)
https://doi.org/10.1007/978-3-031-34589-0

© Springer Nature Switzerland AG 2023

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

The use of partial least squares structural equation modeling (PLS-SEM) has gained
enormous momentum in the past decade in various business research fields such as
accounting, information systems, marketing, strategic management, tourism but also in
other non-business disciplines such as computer sciences, engineering, environmental
sciences, medical sciences, political sciences, and psychology. The widespread adoption
of the method is attributed to the confluence of various factors. PLS-SEM facilitates the
estimation of models with many constructs and complex inter-relationships, while also
accommodating advanced modeling such as higher-order constructs, nonlinear relation-
ships, and conditional process models. Other strengths of PLS-SEM include the ability
to support model comparisons and to test the predictive power of models. Especially the
predictive assessment of the results obtained by PLS-SEM allows researchers to sub-
stantiate their findings and managerial recommendations, which are predictive in nature.
Continued efforts by researchers to refine and improve the method have yielded
rich resources (e.g., textbooks and edited volumes on the method, methodological
articles, and review papers on the method’s use), which have further fueled the
method’s dissemination by creating an understanding how PLS-SEM can support
researchers in accomplishing the study goals. In addition, access to software with
user-friendly graphical user interfaces and to freely available PLS-SEM packages in
the R Statistical Environment have encouraged the use of the method among
non-technical researchers and among the researchers who are mindful of costs.
This proceedings book includes a collection of manuscripts presented during the
2022 International Conference on Partial Least Squares Structural Equation Model-
ing Conference (PLS2022) that was held September 6–9, 2022 at the Faculty of
Economics and Business Administration of the Babeș-Bolyai University in Cluj-
Napoca, Romania. The conference has been designed to cater the needs of
researchers and practitioners who empirically apply and methodologically advance
the PLS-SEM method. It is part of a tradition of great conferences in the context of
the PLS-SEM method such as the PLS2005 in Barcelona, Spain, the PLS2015 in
Seville, Spain, and the PLS2017 in Macau, China. As with the previous conferences,
the PLS2022 served as a vehicle to share and discuss new ideas, help each other, and
explore new ideas in a friendly environment among friends.

v
vi Preface

Photograph by Mumtaz Ali Memon, NUST Business School (NBS), Pakistan

The diverse scientiﬁc inquiries addressed in this collection of manuscripts testify

both the widespread adoption of the method and researchers’ interest to further
address methodological issues in PLS-SEM. This book also supplements the
resources available for the PLS-SEM community, which, we hope, will expand
knowledge and foster novel research in various ﬁelds of inquiry.
We would like to seize this opportunity to thank all the authors and the numerous
colleagues who have devoted their time to reviewing the submissions, thereby
helping further improving the manuscripts. Without your consolidated effort, this
edited volume wouldn’t have been possible. Thank you!

Cluj-Napoca, Romania Lăcrămioara Radomir

Cluj-Napoca, Romania Raluca Ciornea
Beijing, China Huiwen Wang
Taipa, Macau Yide Liu
Hamburg, Germany Christian M. Ringle
Munich, Germany Marko Sarstedt
Contents

Part I Methodology
Empirical Validation of the 10-Times Rule for SEM . . . . . . . . . . . . . . . . 3
Ralf Wagner and Malek Simon Grimm
Missing Values in RGCCA: Algorithms and Comparisons . . . . . . . . . . . 9
Caroline Peltier, Laurent Le Brusquet, François-Xavier Lejeune,
Ivan Moszer, and Arthur Tenenhaus
Comparing Local vs Global Clustering with FIMIX-PLS:
Application to Marketing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Sophie Dominique, Mohamed Hanaﬁ, Fabien Llobell, Jean-Marc Ferrandi,
and Véronique Cariou
Partial Least Squares Structural Equation Modeling-Based Discrete
Choice Modeling: An Illustration in Modeling Hospital Choice with
Latent Class Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Andreas Fischer, Marcel Lichters, and Siegfried P. Gudergan
The Use of PLS-SEM in Engineering: A Tool to Apply the Design
Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Ari Melo-Mariano and Ana Bárbara Plá
Discovering Issues in Cross-Cultural Adaptation of Questionnaire
Through PLS-SEM Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Fariha Reza and Huma Amir
Use of PLS-SEM Approach in the Construction Management
Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Sachin Batra

vii
viii Contents

Part II Consumer Behavior and Marketing

Understanding the Role of Consumer Psychological Motives in Smart
Connected Objects Appropriation: A Higher-Order PLS-SEM
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Zeling Zhong and Christine Balagué
Advertisements that Follow Users Online and Their Effect on Consumers’
Satisfaction and Expectation Conﬁrmation: Evidence from the Tourism
Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Jordi López-Sintas, Giuseppe Lamberti, and Haitham Alghanayem
Similarities in Factors Affecting Online Shopping Intention in Ecuador
and Peru: A Multigroup Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Vasilica-Maria Margalina and Alberto Magno Cutipa Limache
The Impact of Subjective Norms, Perceived Behavioral Control,
and Purchase Intention on Purchase Behavior of Eco-Friendly Food
Packaging Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Julen Castillo-Apraiz, Jesús Manuel Palma-Ruiz, and Mauricio Iván
García-Montes
Sustainability of Food Waste in the Agro-industry of Michoacan,
Mexico . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Jesús Sigifredo Gastélum-Valdez, Irma Cristina Espitia-Moreno,
and Betzabé Ruiz-Morales
Predicting Sustainable Consumption Behavior of Europeans Using
the CVPAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Lena Frömbling, Svenja Damberg, Ulla A. Saari, and Christian M. Ringle

Part III Hospitality and Tourism

The Relationship Among Perceived Value, Tourist Satisfaction, and
Citizenship Behaviors: The Difference Between Overnight Tourists
and Non-overnight Tourists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Yangpeng Lin, Di Zhang, and Hongfeng Zhang
Nonlinear Relationship Between Intellectual Capital and Hotel
Performance: Quadratic Effect in PLS-SEM . . . . . . . . . . . . . . . . . . . . . 129
Lorena Ruiz-Fernández, Bartolomé Marco-Lajara, Pedro Seva-Larrosa,
and Javier Martínez-Falcó
The U-Shape Inﬂuence of Family Involvement in Hotel Chain:
Examining Dynamic Capabilities in PLS-SEM . . . . . . . . . . . . . . . . . . . . 133
Lorena Ruiz-Fernández, Laura Rienda, and Rosario Andreu
Contents ix

Part IV Innovation Management and Entrepreneurship

Two-Way Business Innovation in Central Eastern Europe: Analysing
Innovative Enterprises Using the PLS-SEM Method . . . . . . . . . . . . . . . 139
Márton Gosztonyi
Institutional Context and Entrepreneurship Typologies: The Moderator
Role of Environmental Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Serap Korkarer, Mahmut Hızıroğlu, and Joseph F. Hair Jr.
PLS-Based Evaluation of a Digital Transformation Adoption Model
for Biopharma Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Frederick K. Johnson and Chinazunwa C. Uwaoma
Entrepreneurial Culture, Management, and Innovation of Dairy
Industries in Greece, in a Bureaucracy Environment . . . . . . . . . . . . . . . 193
Athanasios Falaras and Odysseas Moschidis
The Inﬂuence of Facebook on Value Cocreation: Evidence from
the Moroccan Fast-Food Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Mohammed Hassouni, Abdellatif Chakor, and Siham Mourad

Part V Human Resource Management and Organizational Design

and Behavior
Research on Interaction Pattern of Virtual Team in Crisis Situation . . . 229
Shen Yuanyanhang, Jiale Wu, and Xiaodan Yu
Predicting the Performance of New Hires: The Role of Humility,
Interpersonal Understanding, Self-Conﬁdence, and Flexibility . . . . . . . . 239
Debolina Dutta, Chaitali Vedak, and Varghees Joseph
Mindful Leadership Under Fire: A Validation Study of a Hierarchical
Component Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Stephanie Dygico Gapud and Joseph F. Hair Jr.
When Navigating Uncertainty Lead Mindfully . . . . . . . . . . . . . . . . . . . . 277
Stephanie Dygico Gapud and Joseph F. Hair Jr.
The Interplay Between Push Factors and Transformational Leadership
in Inﬂuencing Interorganizational Labor Mobility in Public Sector . . . . 315
Rosemary Massae, Deusdedit A. Rwehumbiza, and John J. Sanga
Integrated Leadership: Assessing an Integrated Principal Leadership
Practices Construct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Ahmed Mohamed, Ahmad Zabidi Abdul Razak, and Zuraidah Abdullah
Exploring the Interrelationship Among Management Accounting
Systems, Decentralization, and Organizational Performance . . . . . . . . . 345
Elsa Pedroso and Carlos F. Gomes
x Contents

The Impact of Ethical Leadership on Employee Intrapreneurship,

Work–Life Balance, and Psychological Empowerment: A PLS-SEM
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
Huma Bashir, Mumtaz Ali Memon, Naukhez Sarwar, Asﬁa Obaid,
and Muhammad Zeeshan Mirza

Part VI Education
PLS and Educational Research: Epistemological and Methodological
Interpretations in Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Hongfeng Zhang, Haoqun Yan, and Johnny F. I. Lam
Public Higher Education Organizational Climate’s Structural Model . . . 383
Joel Bonales-Valencia
An Investigation of Predictive Relationships Between University
Students’ Online Learning Power and Learning Outcomes in
a Blended Course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Yue Zhu, Ming Hua Li, Lu Li, Rong Wei Huang, and Jia Hua Zhang
The Effects of Marketing Orientation on the Performance of Higher
Learning Institutions in Tanzania: Staff and Students’ Perceptions . . . . 409
Francis Muya and Hawa Tundui

Part VII Health

Participative Leadership Is the Discriminating Factor for Country’s
Performance During the COVID-19 Pandemic . . . . . . . . . . . . . . . . . . . . 437
Stephanie Dygico Gapud and George Faint
A PLS-SEM Analysis of Consumer Health Literacy and Intention to
Use Complementary and Alternative Medicine in the COVID-19
Pandemic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Michael Christian, Henilia Yulita, Guan Nan, Suryo Wibowo,
Eko Retno Indriyarti, Sunarno Sunarno, and Rima Melati
Patient Satisfaction of Brazilian Military Health-Care System:
An Exploratory Study by Multivariate Analysis . . . . . . . . . . . . . . . . . . . 475
Ari Melo-Mariano, Simone Borges Simão Monteiro,
Ronaldo Rodrigues Pacheco, and Marcelo Ladeira

Part VIII Other Fields of Business Research and Economics

FinTech Loan Continuance Intention: How Far Can Self-Efﬁcacies
Go? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
Samuel Danilola, Adewumi Odeniran, and Adewumi Otonne
Does Coercive Pressure Matter on the Practices and Performance
of the Public Procurement? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
Gerald Zachary Paga Tinali
Contents xi

Determinants of Behavioral Intention to Use E-Procurement System in

Developing Countries: Suppliers’ Perception from Tanzania . . . . . . . . . 537
Deus N. Shatta
Receptivity of Eastern and Southern African English-Speaking Countries’
Executives to Use, Diffuse and Adopt Humanitarian Logistics Digital
Business Ecosystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
Benjamin Ohene Kwapong Baffoe and Wenping Luo
A PLS-SEM Model on Auditing and GDPR: The Mediating Role
of Internal Audit in Business Management . . . . . . . . . . . . . . . . . . . . . . . 589
Stratos Moschidis, Evrikleia Chatzipetrou, George Drogalas,
and Paraskevi Zisiopoulou
Part I
Methodology
Empirical Validation of the 10-Times Rule
for SEM

Ralf Wagner and Malek Simon Grimm

1 Introduction

Structural equation modeling has become indispensable in scholarly empirical

research due to its ability to investigate latent constructs. However, researchers
always face the central dilemma of determining a sufficient sample size to ensure a
converting model, unbiased estimates, and sufficient statistical power. Various rules
of thumb have been established to determine the optimal sample size, such as a
sample of at least 200 participants (Boomsma 1983), 5–10 observations per esti-
mated parameter (Bentler and Chou 1987), and 10 observations per variable
(Nunnally 1975). Monte Carlo simulation and the inverse square root method are
alternative and more complex methods that require a deeper understanding of
statistical power analysis (Ranatunga et al. 2020).
For decades, researchers have cited the 10-times rule as representing a sufficient
sample size in structural equation models (SEMs) (Thompson et al. 1995). The
10-times rule suggests that the minimum sample size should be 10 times the
maximum number of arrowheads pointing at a latent variable anywhere in the partial
least squares (PLS) path model (Hair et al. 2021). According to Google Scholar
(October 2022), Thompson et al.’s (1995) initial paper by has been cited over 7000
times. However, an empirical validation of this rule of thumb is still lacking. This
paper takes a simulation approach to validate the 10-times rule. Only a few studies
have investigated the sample size requirements for SEMs (cf. Wolf et al. 2013; Stone
and Sobel 1990), but no extant study has specifically attempted to validate the
prominent 10-times rule.

R. Wagner (✉) · M. S. Grimm

University of Kassel, Chair for Sustainable Marketing, Kassel, Germany
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 3

L. Radomir et al. (eds.), State of the Art in Partial Least Squares Structural Equation
Modeling (PLS-SEM), Springer Proceedings in Business and Economics,
https://doi.org/10.1007/978-3-031-34589-0_1
4 R. Wagner and M. S. Grimm

To pursue the research question, we drew upon the European Customer Satisfac-
tion Index (ESCI) data set provided by SmartPLS 3.0 and the respective model,
which contains 24 independent variables. We used a random algorithm to induce
randomly missing cases (MCs) into the data set to decrease the sample size ran-
domly. A 60-times trait (N = 240) was tested against the original trait (N = 250) and
traits with 50 times (N = 200), 40 times (N = 160), 30 times (N = 120), 20 times
(N = 80), and 10 times (N = 40). Each trait was tested by an aggregated simulation
of 30 data sets per trait, so that each data set had a unique pattern of MCs. We tested
model stability by investigating the model ﬁt indices SRMR, d_ULS, d_G, X2, and
NFI. It was found that the 60-times trait was nearly as precise as the original trait,
whereas the remaining traits suffered signiﬁcantly in measurement quality.

2 Methodology

The simulation data were evaluated using SmartPLS 4.0 and the associated ESCI
data set and model containing 24 independent variables. All the available model ﬁt
indices (SRMR, d_ULS, d_G, X2, and NFI) as well as the adjusted R2 values were
investigated. To address the research question, we induced MCs into the original
data set, which contained 250 cases without any missing values (MVs). The char-
acteristics of the MCs can be described by the missing completely at random
(MCAR) condition.
The MCAR condition requires that the probability of the MVs (and, in this case,
MCs) in a given variable is unrelated to all other measures (Grimm and Wagner
2020; Parwoll and Wagner 2012). This condition was ensured by using a code that
generated randomly MCs within the spectrum of N with respect to the desired
number of MCs. Seven traits were investigated. To prevent occasionally occurring
patterns, 30 data sets with unique MCs patterns were generated for each trait (except
the original trait). Accordingly, the results depict the average of 30 data sets for the
test traits.

3 Results

Table 1 illustrates the results of the model ﬁt indices as well as the respective
differences from the original trait; Table 2 illustrates the results of the adjusted R2
values.
The diverse values shown in Tables 1 and 2 imply that the 60-times trait is the
most stable in comparison to the original trait. Biases arise if the number of MCs
increases. Because the adjusted R2 will vary according to the variable and from
survey to survey, no general conclusion can be derived with this assessment.
Notably, the values are increasingly biased as the number of MCs increases. The
Empirical Validation of the 10-Times Rule for SEM 5

Table 1 Estimation results and differences in percentage of the model ﬁt indices

Model ﬁt indices
Trait Estimates SRMR D_ULS d_G X2 NFI
Original trait Estimate 0.13 5.21 0.79 993.22 0.67
Δ 0% 0% 0% 0% 0%
60× trait Estimate 0.13 5.21 0.81 968.59 0.66
Δ 1% 0% 2% - 2% -1%
50× trait Estimate 0.13 5.36 0.88 875.43 0.65
Δ 3% 3% 12% -12% -4%
40× trait Estimate 0.13 5.44 0.98 772.31 0.62
Δ 3% 4% 25% -22% -7%
30× trait Estimate 0.13 5.37 1.17 672.99 0.6
Δ 3% 3% 48% -32% -10%
20× trait Estimate 0.14 6.33 1.67 597.36 0.52
Δ 10% 21% 108% -38% -11%
10× trait Estimate 0.17 8.67 3.88 546.42 0.40
Δ 27% 62% 339% -38% -38%

Table 2 Estimation results and differences in percentage of the adjusted R2 values

R2 Values
Trait Estimates Complaints Expectation Loyalty Quality Satisfaction Value
Original Estimate 0.28 0.25 0.45 0.31 0.68 0.34
trait Δ 0% 0% 0% 0% 0% 0%
60× trait Estimate 0.27 0.25 0.68 0.44 0.31 0.34
Δ -3% 0% 50% 43% -55% -1%
50× trait Estimate 0.28 0.26 0.68 0.44 0.31 0.34
Δ -1% 4% 51% 43% -54% 1%
40× trait Estimate 0.28 0.27 0.67 0.45 0.32 0.34
Δ -2% 7% 48% 44% 54% 1%
30× trait Estimate 0.29 0.30 0.69 0.47 0.35 0.35
Δ 2% 19% 52% 53% -48% 4%
20× trait Estimate 0.28 0.28 0.68 0.47 0.33 0.37
Δ 11% 11% 0% 4% 10% -4%
10× trait Estimate 0.28 0.25 0.45 0.31 0.68 0.34
Δ 0% 0% 0% 0% 0% 0%

results in Table 1 also imply that a bias is induced and increases when the number of
MCs increases.
One-sample t-tests were computed to test the statistical signiﬁcance with the
original trait (Fig. 1), with each single trait being tested against the respective
original trait. The t-tests were run with the original values of the respective model
ﬁt indices. To provide a better comparison, Fig. 1 includes percentage values that
indicate the percentage of difference from the original value.
6 R. Wagner and M. S. Grimm

Fig. 1 Direct comparison of the model ﬁt indices with statistical signiﬁcance test. Notes:
*p < 0.05; **p < 0.01; ***p < 0.001

The results suggest that d_G, NFI, and X2 in particular are substantially and
signiﬁcantly biased and that d_ULS and SRMR are biased if the number of MCs
increases. Overall, the 60-times trait is the most stable in comparison to the original
trait, even though the models are slightly skewed with respect to the indices d_G,
NFI, and X2. Because the NFI measure is calculated on the basis of the X2 value, the
assessment of biases is related within that measure. The most relevant characteristic
is the elbow that becomes apparent at the 30-times trait.

4 Conclusion

Challenging the common assumption that the PLS algorithm is robust against MVs
and small sample sizes, this simulation study yields important implications for the
theory and practice of recommending a minimum sample size. The results indicate
that the 10-times rule is a misleading heuristic. With a decrease in sample size, the
likelihood of biases increases. The 60-times trait performed best in this study and in
comparison to the original trait. An elbow in the bias plot suggests that researchers
are well advised to use at least a 30-times trait. Further research should adopt
different models as test instances and should address the facets of statistical power
(inverse square root of error variance) in relation to the ﬁt statistics
Empirical Validation of the 10-Times Rule for SEM 7

(heteroskedasticity), an overﬁtting assessment (outer sample validation routine),

endogeneity tests, and non-MCAR MVs.

References

Bentler PM, Chou C-P (1987) Practical issues in structural modeling. Sociol Methods Res 16(1):
78–117
Boomsma A (1983) On the robustness of LISREL (maximum likelihood estimation) against small
sample size and non-normality. Sociometric Research Foundation, Amsterdam
Google Scholar (2022) https://scholar.google.com/ Accessed Oct 2022
Grimm MS, Wagner R (2020) The impact of missing values on PLS, ML and FIML model ﬁt.
Advance online publication. Arch Data Sci Ser A 6(1):1–17
Hair JF, Hult GTM, Ringle CM, Sarstedt M, Danks NP, Ray S (2021) Partial least squares structural
equation modeling (PLS-SEM) using R. Springer, Cham
Nunnally JC (1975) Psychometric theory, 11th edn. McGraw-Hill, New York
Parwoll M, Wagner R (2012) The impact of missing values on PLS model ﬁtting. In: Gaul WA,
Geyer-Schulz A, Schmidt-Thieme L, Kunze J (eds) Challenges at the interface of data analysis,
computer science, and optimization. Springer, Berlin, pp 537–544
Ranatunga RVSPK, Priyanath HMS, Megama RGN (2020) Methods and rule-of-thumbs in the
determination of minimum sample size when applying structural equation modelling: a review. J
Social Sci Res 15:102–109
Stone CA, Sobel ME (1990) The robustness of estimates of total indirect effects in covariance
structure models estimated by maximum. Psychometrika 55(2):337–352
Thompson R, Barclay D, Higgins CA (1995) The partial least squares approach to causal modeling:
personal computer adoption and use as an illustration. Technology Studies: Special Issues on
Research Methodology 2:284–324
Wolf EJ, Harrington KM, Clark SL, Miller MW (2013) Sample size requirements for structural
equation models: An evaluation of power, bias, and solution propriety. Educ Psychol Meas
76(6):913–934
Missing Values in RGCCA: Algorithms
and Comparisons

Caroline Peltier, Laurent Le Brusquet, François-Xavier Lejeune,

Ivan Moszer, and Arthur Tenenhaus

C. Peltier (✉)
iCONICS core facility, Institut du Cerveau et de la Moelle Epinière, Inserm U1127, CNRS
UMR 7225, Sorbonne Université, Paris, France
Centre des Sciences du Goût et de l’Alimentation, CNRS, INRAE, Institut Agro, University of
Bourgogne Franche-Comté, Dijon, France
CNRS, INRAE, PROBE Research Infrastructure, ChemoSens Facility, Dijon, France
INRAE, Dijon, France
e-mail: [email protected]
L. Le Brusquet
Laboratoire des Signaux et Systèmes (L2S), CentraleSupélec, Université Paris-Saclay, Gif-Sur-
Yvette, France
CentraleSupélec, Gif-sur-Yvette, France
e-mail: [email protected]
F.-X. Lejeune · I. Moszer
iCONICS core facility, Institut du Cerveau et de la Moelle Epinière, Inserm U1127, CNRS
UMR 7225, Sorbonne Université, Paris, France
Institut du Cerveau, Hôpital Pitié, Paris, France
e-mail: [email protected]
A. Tenenhaus
iCONICS core facility, Institut du Cerveau et de la Moelle Epinière, Inserm U1127, CNRS
UMR 7225, Sorbonne Université, Paris, France
Laboratoire des Signaux et Systèmes (L2S), CentraleSupélec, Université Paris-Saclay, Gif-Sur-
Yvette, France
CentraleSupélec, Gif-sur-Yvette, France
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 9

1 Introduction

Multidisciplinary approaches are now common in scientiﬁc research and provide

multiple and heterogeneous sources of measures of a given phenomenon. These
sources or blocks can be viewed as a collection of interconnected datasets, and
dedicated algorithms are mandatory for providing relevant information from
multiblock data.
Regularized generalized canonical correlation analysis (RGCCA) is a general
statistical framework for multiblock data analysis which gathers many multiblock
methods (PCA, PLS, CCA, consensus PCA, PLS-PM, etc.) through a single and
very simple iterative algorithm. The global convergence of the algorithm was
demonstrated in Tenenhaus et al. (2017).
However, multiblock data often have missing structure, i.e., data in one or more
blocks may be completely unobserved for a sample (block-wise structure) or par-
tially unobserved (random structure). The probability to observe missing data
increases with the number of blocks. It is therefore mandatory to properly handle
these missing structures within the framework of RGCCA.
In this work, several solutions were investigated and compared on simulations.
An R package, “RGCCA,” implementing all the methods is under development and
is available on GitHub (https://github.com/rgcca-factory/RGCCA).

2 Background
2.1 Regularized Generalized Canonical Correlation Analysis

Let X1, . . ., XJ be J blocks of variables, each block representing a set of pj variables

observed on n individuals. The number and the nature of the variables usually differ
from one block to another, but the individuals must be the same across blocks.
RGCCA is based on the following optimization problem (Tenenhaus et al. 2017).

J
Maximize c g
j,k = 1 jk
cov Xj wj , Xk wk s:t:wtj M j wj = 1, 8j 2 f1, . . . , J g

where g(x) is any continuously differentiable convex function.

The connection matrix C = (cjk) is a symmetric J*J matrix of nonnegative
elements describing the network of connections between blocks that the user
wants to take into account. Usually, cjk = 1 for two connected blocks and
0 otherwise.
Mj is a positive deﬁnite matrix, such as Mj = τj I þ 1 - τj Xtj Xj with 0 ≤ τj ≤ 1.
In the context of RGCCA, τj is called shrinkage constant. τj varies between 0 and
1 and interpolates smoothly between maximizing the covariance and maximizing the
correlation.
Missing Values in RGCCA: Algorithms and Comparisons 11

In this work, we focused on the situation where τ j = 1 and all blocks are
connected or connected to a superblock, deﬁned as the concatenation of the indi-
viduals blocks.

2.2 Missing Data Literature

Pigott (2001) reviewed different types of strategies to deal with missing data:
complete data technique, technique based on available data, imputation methods
(including expectation-maximization algorithms), and multi-imputation methods.

2.3 Methodology

2.3.1 Complete Data Technique (Complete)

The complete approach consists in computing RGCCA only on the complete

individuals. This is the simplest method and is the only one available until now.

2.3.2 Technique Based on Available Data (Passive)

This approach (the so-called passive) follows the approach presented in PLS-PM
(Tenenhaus et al. 2005). RGCCA is closely related to the nonlinear estimation by
iterative partial least squares algorithm (NIPALS; Wold 1966) as the components,
and the axes are estimated alternatively. As in Tenenhaus et al. (2005), means and
standard deviations of the variables are computed on all the available data; covari-
ance matrices are computed using all the pairwise available data. This pairwise
deletion procedure shows the drawback of possibly computing covariances of
different sample sizes and/or different individuals.

2.3.3 Imputation Method (Iterative)

The imputation method (Fig. 1) is iterative and based on the alternated two steps
(i) calculating the RGCCA weight vectors wj and components yj = Xjwj and
(ii) imputation of the missing values. This imputation was based on the hypothesis
that each column of Xj can be estimated by a vector proportional to the RGCCA
block component yj. This estimation is especially relevant when each block is
unidimensional. Thus, the used reconstruction formulae for imputation was Xj = yj
γTj where γj j is a column vector of size pj containing the regression coefﬁcients of yj
in the regression of Xj on yj.
12 C. Peltier et al.

Fig. 1 Iterative algorithm for RGCCA with missing values

This method is directly inspired by the ﬁxed effect model used in Josse and
Husson (2012) for PCA with missing data. However, when the missing structure is
blockwise, this approach gives the same results for all missing individuals, even if
they have very different (non-missing) values in the other blocks.
To overcome this issue, we propose to use a “superblock” strategy. In the
multiblock literature, a superblock XJ + 1 = [X1, . . ., XJ] is deﬁned as the concate-
nation of all blocks. In this framework, each block is connected to the superblock
(that is, cj, J + 1 = 1 for j = 1, . . ., J and 0 otherwise), and optimization problem
(1) reduces to

J
Maximize j=1
g cov Xj wj , XJþ1 wJþ1 s:t:wtj MJ wj = 1, 8j 2 f1, . . . , J þ 1g

This optimization problem is fully described in Tenenhaus et al. (2017). When a

superblock XJ + 1 is considered, the imputation is obtained by using the following
reconstruction formulae XJ + 1 = yJ + 1 γTJþ1 , where γJ + 1 is a column vector of size
p = Σpj containing the regression coefﬁcients of yJ + 1 in the regression of XJ + 1 on
yJ + 1.with an algorithm similar to Algorithm 1.

2.3.4 Dataset Presentation

These methods were tested on the Russett dataset, available within the RGCCA
package. The Russett dataset (Russett 1964) are studied in Giﬁ (1990) and aims to
study the relationships between agricultural inequality, industrial development, and
Missing Values in RGCCA: Algorithms and Comparisons 13

political instability. Three blocks of variables were deﬁned accordingly (3, 2, and
2 variables for agriculture, industry, and political, respectively) for 47 countries.

2.3.5 Comparison Method

From the Russett dataset, 20 datasets with 5, 10, 15, 20, and 25% of random missing
datasets were simulated. For each block, the norm of the difference between the ﬁrst
weight vector obtained from the full case wj and the one from the missing case wj is
calculated. Consequently, the closer to 0 this norm is, the better the method is. We
used this method to compare complete, available, iterative, and superblock
approach.

3 Results and Discussion

From our simulations, it appears that the iterative algorithm (with and without
superblock) converges monotonically. Furthermore, the simulations showed that
the implemented methods outperformed the complete approach on Russett data
(Fig. 2) especially for superblock or iterative method.

Fig. 2 The norm of the difference (mean and standard error) between the ﬁrst axis of RGCCA
based on the complete dataset and the ﬁrst axis of the different “missing” methods according to the
proportion of missing values in the dataset
14 C. Peltier et al.

4 Conclusions and Implications for Theory and Practice

This work presents different methods for taking missing values into account in
RGCCA. The iterative method presented in this chapter gave good results on the
Russett dataset. The convergence of this algorithm was observed but is still to be
studied. Multiple imputations for visualizing the variability induced by the imputa-
tion could be implemented according to the model used. Furthermore, the strong
hypothesis of unidimensionality of blocks used in this chapter could be relaxed by
considering more than one component per block, including deﬂation steps in the
iterative algorithm. Finally, this work was illustrated on the Russett dataset with a
small number of variables, blocks, and samples, but aims to be also tested on more
datasets.

References

Giﬁ A (1990) Nonlinear multivariate analysis. Wiley, Chichester

Josse J, Husson F (2012) Handling missing values in exploratory multivariate data analysis
methods. Journal de la Société Française de Statistique 153(2):79–99
Pigott TD (2001) A review of methods for missing data. Educ Res Eval 7(4):353–383
Russett BM (1964) Inequality and instability: The relation of land tenure to politics. World Polit
16(3):442–454
Tenenhaus M, Esposito Vinzi V, Chatelin YM, Lauro C (2005) PLS path modeling. Comput Stat
Data Anal 48(1):159–205
Tenenhaus M, Tenenhaus A, Groenen PJF (2017) Regularized generalized canonical correlation
analysis: a framework for sequential multiblock component methods. Psychometrika 82(3):
737–777
Wold H (1966) Estimation of principal components and related models by iterative least squares. In:
Krishnajah PR (ed) Multivariate analysis. Academic Press, New York, pp 391–420
https://github.com/rgcca-factory/RGCCA
Comparing Local vs Global Clustering
with FIMIX-PLS: Application to Marketing

Sophie Dominique, Mohamed Hanaﬁ, Fabien Llobell, Jean-Marc Ferrandi,

and Véronique Cariou

1 Introduction

The use of Partial Least Squares Structural Equation Modeling (PLS-SEM) to

estimate model parameters with interrelationships between observed concepts and
their corresponding latent variables has gained popularity in marketing research
since past decades (Sarstedt et al. 2022). When applied to a set of observations,
the determination of a single set of parameters, encompassing the latent variables
scores and their associated path coefﬁcients estimations, implicitly assumes that
heterogeneity is negligible. In practice, this assumption is not realistic in the social
sciences, and there is generally unobserved heterogeneity that can taint the validity
of PLS-SEM results. To uncover this potential heterogeneity in the data, one can
apply a clustering strategy that aims to identify homogeneous segments of observa-
tions sharing the same pattern relationships (i.e., the same path coefﬁcients within a
segment).

S. Dominique (✉)
Oniris, INRAE, STATSC, Nantes, France
Lumivero, XLSTAT, Paris, France
e-mail: [email protected]
M. Hanafi · V. Cariou
Oniris, INRAE, STATSC, Nantes, France
e-mail: mohamed.hanafi@oniris-nantes.fr; [email protected]
F. Llobell
Lumivero, XLSTAT, Paris, France
e-mail: fl[email protected]
J.-M. Ferrandi
LEMNA, Oniris, Nantes, France
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 15

For the last two decades, several clustering techniques have been developed
within the PLS-SEM framework, such as FIMIX-PLS (Hahn et al. 2002; Ringle
et al. 2010), REBUS (Esposito Vinzi et al. 2008), PLS-POS (Becker et al. 2013),
GAS (Ringle et al. 2014), PLS-IRRS (Schlittgen et al. 2016), or more recently
PLS-SEM KMEANS (Fordellone and Vichi 2020). Undoubtedly, the finite mixture
PLS method (FIMIX-PLS) remains the most popular of them. In FIMIX-PLS, the
measurement model is considered common to all observations, while observations
can differ from each other only at the structural level. This approach is well suited to
the context of social science because it leads to the rational assumption that the
measurement scales are identical from one segment to another and observation to
observation.
Going further, in some situations, heterogeneity can be concentrated only in a part
of the structural model and not affect the entire set of path coefficients parameters.
This leads to determining part of the path coefficients as common to all the
observations and the other part as parameters reflecting heterogeneity, which
depends on the observation segments. A marketing expert’s knowledge is required
to identify the constant and variable parts of the model for all observations. Our work
aims to adapt FIMIX-PLS procedure so that the segmentation is only performed on a
subpart of the structural model. Subsequently, local partitioning is introduced as a
moderating variable in the PLS-SEM applied to all the observations. We advocate
such a rationale to provide clusters, which aim to be more stable and interpretable.
This strategy is illustrated using a case study pertaining to marketing. Local vs global
partitioning with FIMIX-PLS are compared both in terms of interpretability and
model quality.

2 Methodology

This analysis was carried out on a sample of 315 French people who are members of
a CSA (Community Supported Agriculture) in Nantes in 2011 (Dufeu and Ferrandi
2013). CSA belongs to the family of short supply chains, with products sold directly
from producers to consumers. Their members are generally opposed to other modes
of retailing, particularly mass distribution. Mainly based on the principle of mutual
commitment of both parts, CSA seeks to create strong links and proximity between
producers and consumers.
The objective of this study on the Nantes CSAs was to model the link between the
proximity perceived by members toward their CSA and their trust, satisfaction, and
commitment in this CSA that are the key variables of relational marketing.
In marketing the concept of proximity is composed of several dimensions.
Beyond the geographical proximity, the direct and repeated exchanges between all
the actors (relational proximity), the sharing of values (identity proximity), and the
knowledge on the production and distribution process (process proximity) are the
dimensions evaluated in our work (Bergadaà and Del Bucchia 2009). They can all
constitute antecedents of trust, satisfaction, and loyalty. Figure 1 represents the path
Comparing Local vs Global Clustering with FIMIX-PLS: Application to Marketing 17

Fig. 1 Studied model

18 S. Dominique et al.

diagram associated with the six latent concepts detailed above. Trust, satisfaction,
and commitment are the founding concepts of relational marketing and are therefore
common to all segments that can be defined.
The first subpart of the model was chosen to segment the respondents. Indeed,
proximity is considered here as an antecedent of the relationship marketing chain;
thus, we assume that heterogeneity lies mainly on the relationships between prox-
imity and trust. A first partitioning algorithm was carried out on this subpart of the
model leading to a so-called local partition.
In parallel, a global partition was determined on the basis of the complete
structural model. To compare the two strategies, parameters associated with the
complete structural model were estimated independently for each segment of the
local partition. Finally, the two partitions obtained (local vs global) were compared
in terms of path coefficients and variance accounted for (R2). Both partitions were
obtained with FIMIX-PLS algorithm implemented in the SmartPLS software.

3 Results and Discussion

The partition obtained from the sub-model consisted of two segments. In contrast,
the segmentation on the complete model led to retain identiﬁed three segments of
observations. This number of segments was chosen with the BIC and CAIC indices
(Sarstedt et al. 2011).

3.1 Global Partition

The global partition consists of one main segment (N1 = 201) and two smaller ones
(N2 = 80 and N3 = 34). Table 1 presents the results of the model associated with
each cluster.
The first segment is characterized by a valuation of process and identity proxim-
ities (path coefficient, respectively, equals to 0.630 and 0.252). By feeling close to
these dimensions, consumers are confident in their CSA which in turn leads to
increased satisfaction and commitment. The R2 values for each latent variable are
greater than 0.6, indicating that they are well explained by the variables directly
related to them and therefore that the structure model associated with segment
1 performs well.
Only the identity dimension of proximity explains the trust of segment 2 con-
sumers (path coefficient = 0.347). Notwithstanding, proximity alone explains the
trust dimension poorly (R2 = 0.147).
Finally, the last segment is characterized by a strong valuation of identity
proximity associated with a lower valuation of relational proximity. Moreover, the
R2 values are very close to 1; this indicates that the model is fully explained by the
relationships defined between the latent variables, which is difficult to achieve in
Comparing Local vs Global Clustering with FIMIX-PLS: Application to Marketing 19

Table 1 Results of the global partition models

Segment 1 (N = 201) Segment 2 (N = 80) Segment 3 (N = 34)
Path coefﬁcient
Relational prox. → Trust 0.074 -0.149 0.130*
Identity prox. → Trust 0.252*** 0.347* 0.914***
Process prox. → Trust 0.630*** -0.055 -0.012
Trust → Satisfaction 0.911*** 0.648*** 0.957***
Satisfaction → 0.811*** 0.798*** 0.999***
Commitment
R2
Trust 0.774 0.147 0.919
Satisfaction 0.831 0.420 0.916
Commitment 0.658 0.636 0.998
Notes: *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001

practice. Moreover, one emphasizes the small size of the segment (N3 = 34).
Focusing on the optimization of the structural model, the FIMIX-PLS strategy has
led to a partition with a very speciﬁc cluster corresponding to an overﬁtted model
with R2 close to 1.

3.2 Local Partition

The local partitioning strategy, that is, to say on the ﬁrst part of the structural model
as shown in Fig. 1, led to consider a partition with two clusters whose respective size
are N1 = 228 and N2 = 87. The results associated with the two models are ﬁgured out
in Table 2.
In the largest segment, all three dimensions of proximity are antecedents to the
concept of trust, with the proximity of the process being particularly important to the

Table 2 Results of the local partition models

Segment 1 (N = 228) Segment 2 (N = 87)
Path coefﬁcient
Relational prox. → Trust 0.154*** -0.245*
Identity prox. → Trust 0.234*** 0.425***
Process prox. → Trust 0.634*** -0.204*
Trust → Satisfaction 0.869*** 0.746***
Satisfaction → Commitment 0.844*** 0.804***
R2
Trust 0.857 0.250
Satisfaction 0.755 0.556
Commitment 0.712 0.646
Notes: *p-value < 0.05; **p-value < 0.01; ***p-value < 0.001
20 S. Dominique et al.

trust that consumers place in their CSA (path coefficient = 0.634). The quality of the
structural model related to this segment is satisfactory, with R2 values greater than
0.7.
The smallest segment values only identity proximity (path coefficient = 0.425).
The negative sign of the path coefficients linking process and relational proximity to
trust may reveal that consumers have more trust if they share the company’s identity
values, but less if the process and relational proximity dimensions are present.

3.3 Local Partition and Global Partition

By comparing the sizes of both partitions, in Table 3, we notice that segments 1 and
2 of each partition group have the same observations. Moreover, the Adjusted Rand
Indices (ARI) and Normalized Mutual Information (NMI) are satisfactory, and thus
the two partitions are close (ARI = 0.32, NMI = 0.23).
From Table 3, it can also be seen that the third segment obtained in the global
partition is included into the first segment of the local partition. The average R2 of
each segment of the local partition is greater than that obtained with the global
partition (segment 1: R2local1 = 0:775 vs R2global1 = 0:754; segment 2: R2local2 = 0:484 vs
R2global2 = 0:401). Furthermore, we achieve a larger value for the local partition by
computing the average of the R2 weighted by the number of observations on all
clusters of each partition (weighted R2local = 0:694 vs weighted R2global = 0:685 ).
Therefore, the local partition that groups segment 3 into segment 1 performs better.
Thus, this tends to confirm our hypothesis that this third cluster is probably an
artifact. The strategy of performing a partition on a subpart of the structural model
improves both the quality of the segmentation and its interpretability.
To conclude, local partitioning aims at bringing a prior knowledge on the source
of the heterogeneity by imposing constraints on the structure for the determination of
the path coefficients. These constraints consist in imposing some of the path
coefficients to be common for all segments. The underlying constraints aim at
(1) improving the quality of the partition and (2) facilitating the interpretation of
the clusters obtained.
Future works are needed to directly take into account this a priori knowledge to be
integrated as constraints within the FIMIX-PLS clustering criterion.

Table 3 Cross-tabulation of the size of the two partitions

Global partition
Local partition Segment 1 Segment 2 Segment 3 Total
Segment 1 177 21 30 228
Segment 2 24 59 4 87
Total 201 80 34 315
Comparing Local vs Global Clustering with FIMIX-PLS: Application to Marketing 21

References

Becker J-M, Rai A, Ringle CM, Völckner F (2013) Discovering unobserved heterogeneity in
structural equation models to avert validity threats. MIS Q 37(3):665–694
Bergadaà M, Del Bucchia C (2009) La recherche de proximité par le client dans le secteur de la
grande consommation alimentaire. Revue Management et Avenir 21:121–135
Dufeu I, Ferrandi J-M (2013) Les ressorts de l’engagement des consommateurs dans une forme
particulière de consommation collaborative: les AMAP. Décis Market 72:157–178
Esposito Vinzi V, Trinchera L, Squillacciotti S, Tenenhaus M (2008) REBUS-PLS: a response-
based procedure for detecting unit segments in PLS path modelling. Appl Stoch Model Bus Ind
24(5):439–458
Fordellone M, Vichi M (2020) Finding groups in structural equation modeling through the partial
least squares algorithm. Comput Stat Data Anal 147:106957
Hahn C, Johnson MD, Herrmann A, Huber F (2002) Capturing customer heterogeneity using a
ﬁnite mixture PLS approach. Schmalenbach Bus Rev 54(3):243–269
Ringle CM, Sarstedt M, Mooi EA (2010) Response-based segmentation using ﬁnite mixture partial
least squares. Theoretical foundations and an application to American customer satisfaction
index data. In: Stahlbock R, Crone S, Lessmann S (eds) Data mining. Annals of information
systems, vol 8. Springer, Boston, pp 19–49
Ringle CM, Sarstedt M, Schlittgen R (2014) Genetic algorithm segmentation in partial least squares
structural equation modeling. OR Spectr 36(1):251–276
Sarstedt M, Becker J-M, Ringle CM, Schwaiger M (2011) Uncovering and treating unobserved
heterogeneity with FIMIX-PLS: which model selection criterion provides an appropriate num-
ber of segments? Schmalenbach Bus Rev 63(1):34–62
Sarstedt M, Hair JF, Pick M, Liengaard BD, Radomir L, Ringle CM (2022) Progress in partial least
squares structural equation modeling use in marketing research in the last decade. Psychol Mark
39(5):1035–1064
Schlittgen R, Ringle CM, Sarstedt M, Becker J-M (2016) Segmentation of PLS path models by
iterative reweighted regressions. J Bus Res 69(10):4583–4592
Partial Least Squares Structural Equation
Modeling-Based Discrete Choice Modeling:
An Illustration in Modeling Hospital Choice
with Latent Class Segmentation

Andreas Fischer, Marcel Lichters, and Siegfried P. Gudergan

1 Introduction

Understanding and predicting individual choices—such as those between different

means of transport, hospitals, or retailers—is extremely important in business
research (Louviere et al. 2008). One commonly used method to explain and predict
consumer choices is discrete choice modeling (DCM), ﬁrst introduced by Luce and
Tukey (1964), formalized by McFadden (1974), and introduced to marketing as
choice-based conjoint (CBC) analysis by Louviere and Woodworth (1983). This
method allows researchers to estimate the relative likelihood of choosing one option
from a set of alternatives. This allows for estimating the utility values (i.e., prefer-
ence of an attribute level over others, usually zero-centered within each attribute) and
determining every attribute’s relative importance weight (i.e., the importance of
every attribute as a whole in the consumer’s decision process) (Train 2009).

A. Fischer (✉)
Institute of Human Resource Management and Organizations, Hamburg University of
Technology, Hamburg, Germany
e-mail: andreas.ﬁ[email protected]
M. Lichters
Faculty of Economics and Business Administration, Chemnitz University of Technology,
Chemnitz, Germany
e-mail: [email protected]
S. P. Gudergan
College of Business, Law & Governance, James Cook University, Douglas, QLD, Australia
Department of Global Business and Trade, Vienna University of Economics and Business,
Vienna, Austria
Aalto University School of Business, Aalto University, Helsinki, Finland
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 23

However, since DCM provides results on an aggregated data level, researchers often
use the latent class analysis (LCA) introduced by Greene and Hensher (2003) or
hierarchical Bayes analysis (Lenk et al. 1996) to uncover distinct segments of
individuals in the dataset. As an advantageous alternative, Hair et al. (2019a) suggest
using partial least squares structural equation modeling (PLS-SEM) (Hair et al.
2016a, 2019b) for the estimation of individuals’ preference functions drawing on
discrete choice experiment (DCE) data. This method permits estimating the coeffi-
cients and the importance of both the attribute levels but also the entire attributes. In
this research, we substantiate that LCA using PLS-SEM produces similar results
compared to those generated using conventional DCM. An advantage of using
PLS-SEM draws on its capability to reveal segments in consideration of the relation-
ships between attributes as a whole (rather than attribute levels) and the choice
variable. Thereby, the segmentation approach allows for uncovering segments with
segment-specific differences related to the entire attribute rather than certain attribute
levels when using conventional DCM. Specifying segments based on whole attri-
butes is beneficial for describing and characterizing segments, interpreting results,
and drawing conclusions.

2 Methodology

Our illustrative application of PLS-SEM in LCA utilizes DCE data from the
healthcare sector. The model focuses on the choice of patience between different
hospitals in Germany. Schuldt et al. (2017) estimated patient choices concerning
hospitals, using a sample of 590 randomly selected participants in three different
German cities of the federal state Saxony-Anhalt, which responded to the DCE study
in a “paper-and-pen” questionnaire. Within this study, each participant provided
answers to eight choice tasks, comprising two options each. Four attributes described
each option (distance to hospital, information, number of treatments, and complica-
tion rate). Each of these attributes had two levels (distance 1 km or 20 km to hospital,
high or low level of information for treatment, high or low number of treatments per
year, high or low rate of complications). The authors used Sawtooth Software
Lighthouse Studio Sawtooth Software (2019) to create individualized choice designs
according to a balanced overlap strategy (Johnson et al. 2013; Orme and Chrzan
2017).

3 Results

For the DCM analysis, we revert to the results presented by Schuldt et al. (2017).
These authors used a mixed effect logit model with a random effects intercept to
estimate attribute level utilities and the relative importance of each attribute. The
PLS-SEM results reported here draw on the use of SmartPLS 3 (Ringle et al. 2015).
Partial Least Squares Structural Equation Modeling-Based Discrete. . . 25

60%

50% 48%
45%

40%

30%
23% 24%
19% 18%
20%
13%
10%
10%

0%
Distance Information Number of Complication Rate
Treatments

Schuldt et al. (2017) PLS

Fig. 1 Comparison of importance weights

Table 1 Part worth utilities and segment sizes PLS-SEM

Attribute Attribute level 1 2 3 4
Part worth Distance 1 km 0.054 0.182 0.019 0.00
utilities 20 km -0.054 -0.182 -0.019 0.00
Information on Brief 0.151 0.161 0.642 0.00
treatment Detailed -0.151 -0.161 -0.642 0.00
Number of More than 0.127 0.254 0.106 0.00
treatments 100 times
Less than -0.127 -0.254 -0.106 0.00
50 times
Complication rate Low 0.659 0.227 0.200 1.00
High -0.659 -0.227 -0.200 -1.00
Segment size 46.8% 36.4% 14.1% 2.7%

Figure 1 shows both the importance weights of our estimates and those of the
estimates of Schuldt et al. (2017). The differences are very small and play no role
in the interpretation of results. The complication rate is the most important attribute,
followed by information and the number of treatments, and then distance.
A LCA allows for uncovering distinct segments of individuals with different
preferences. Based on the latent variable scores we run a LCA by using the statistical
software R (R Core Team 2019) and the ﬂexmix package (Grün and Leisch 2007).
Table 1 shows the results of the four-segment solution, Table 2 presents the results
from Schuldt et al. (2017), which rely on LCA model proposed by DeSarbo et al.
(1995) and implemented in Sawtooth Software Lighthouse Studio (Sawtooth Soft-
ware 2019). In contrast to the traditional LCA, our method reveals segments at the
level of path coefﬁcients in the PLS-SEM and the level of attributes in the DCM.
26 A. Fischer et al.

Table 2 Part worth utilities and segment sizes (Schuldt et al. 2017)
Attribute Attribute level 1 2 3 4
Part worth Distance 1 km 0.511 -0.084 1.103 1.170
utilities 20 km -0.511 0.084 -1.103 -1.170
Information on Brief 1.173 0.913 0.882 0.218
treatment Detailed -1.173 -0.913 -0.882 -0.218
Number of More than 0.906 0.466 2.505 -0.173
treatments 100 times
Less than -0.906 -0.466 -2.505 0.173
50 times
Complication rate Low 3.103 0.616 1.885 0.263
High -3.103 -0.616 -1.885 -0.263
Segment size 60.1% 20.3% 11.0% 8.7%

70% 66%

60% 54%
50%

40%

30%
21%
20% 15% 16%
13%
9%
10% 5%

0%
Distance Information Number of Complication rate
Treatments

Schuldt et al. (2017) PLS

Fig. 2 Importance weights Segment 1

Figure 2 shows that the estimated relative importance scores for the largest
segment are almost identical when comparing both methods. The complication
rate is the most important attribute, after that information on treatment, then the
number of treatments, and subsequently distance. The differences between the two
methods play again no role in the interpretation. The results for the other segments
are, however, different. While the coefﬁcients have consistent signs, they differ in
magnitude. The segment sizes are also different, but with both methods the largest
two segments cover more than 80% of the individuals.
A closer look at the smallest segment—based on PLS-SEM estimation—offers,
however, interesting insights. For each respondent, in this segment, each individual
choice set throughout the DCM study presented one option with a high and a low
complication rate alternative and the respondents choose systematically the alterna-
tive with the lower complication rate. Thus, our PLS segmentation successfully
Partial Least Squares Structural Equation Modeling-Based Discrete. . . 27

Table 3 Fit indices for a one- to four-segment solution

Number of segments
Criteria 1 2 3 4
AIC 10559.642 9932.747 -7810.799 -7982.517
BIC 10595.405 10011.427 -7689.203 -7818.005
EIC Na 0.815 0.793 0.781

identiﬁed decision-makers with non-compensatory decision-making behavior.

Schuldt et al. (2017) did not reveal this segment.
To determine the optimal number of segments, Table 3 shows the Akaike
information criterion (AIC; Akaike 1974), Bayesian information criterion (BIC;
Schwarz 1978), and entropic information criterion (EIC; Ramaswamy et al. 1993)
for a one- to a four-segment solution obtained by PLS-SEM and ﬂexmix. In line with
the original study, also our segmentation method would suggest four distinct seg-
ments of respondents.
For the selection of the number of segments, a deeper understanding of the
information criteria is, however, important (Oliveira-Brochado and Vitorino Martins
2014). Information criteria are not always useful to determine the optimal numbers
of segments, because they don’t take into account, how well separated the segments
are (Hair et al. 2016b). Further research is needed to understand the performance of
the information criteria in the use of PLS-SEM in DCM.

4 Discussion

With this chapter, we substantiate the suitability of PLS-SEM to estimate DCMs

drawing on DCE data. The DCM and PLS-SEM results are comparable to other
estimation methods; in this case, a multilevel mixed-effects logit model and a LCA,
drawing on Sawtooth Software Lighthouse Studio. Using PLS-SEM is promising,
because it reduces computational runtimes as compared to hierarchical Bayes esti-
mation, while at the same time, it reduces complexity by focusing on the attribute
and not on the attribute level. Furthermore, the use of PLS-SEM enables the
researcher to model latent benefit dimensions, by combining multiple attributes
into a higher-order construct, when appropriate. More importantly, we show that
PLS-SEM in combination with the LCA permits to uncover meaningful and distinct
segments in the DCE data and allows uncovering of unobserved heterogeneity.
Unobserved heterogeneity is a validity threat in statistical analysis (Becker et al.
2013). The estimation of DCM with PLS-SEM enables the researcher to conduct
respondent-specific analysis (e.g., calculating individual R2 values to uncover indi-
vidual choice behavior) to develop a deeper understanding of the choices.
These findings are important on a practical level, as they enable companies, for
example, to target individuals belonging to different segments. On a theoretical
level, they enable researchers to understand more about the choice process
28 A. Fischer et al.

(non-compensatory decision-making) and the advantages and disadvantages of

different estimation methods. Nonetheless, some limitations pertain to the results
that we are presenting here. First, we need to further analyze the differences between
traditional LCA and our method, and we need to more fully understand, which data
structures are better uncovered with our or the traditional method if the data structure
conditions the suitability of a particular method. For instance, does the decision-
making approach (compensatory versus non-compensatory) affect the suitability of
employing PLS-SEM for the estimation of choices?

Acknowledgments We also would like to acknowledge the contributions provided by Professor

Christian M. Ringle.

References

Akaike H (1974) A new look at the statistical model identification. In: Parzen E, Tanabe EK,
Kitagawa G (eds) Selected papers of Hirotugu Akaike. Springer series in statistics. Springer,
New York, pp 215–222
Becker J-M, Rai A, Ringle CM, Völckner F (2013) Discovering unobserved heterogeneity in
structural equation models to avert validity threats. MIS Q 37(3):665–694
DeSarbo WS, Ramaswamy V, Cohen SH (1995) Market segmentation with choice-based conjoint
analysis. Mark Lett 6(2):137–147
Greene WH, Hensher DA (2003) A latent class model for discrete choice analysis: contrasts with
mixed logit. Transp Res B Methodol 37(8):681–698
Grün B, Leisch F (2007) Fitting finite mixtures of generalized linear regressions in R. Comput Stat
Data Anal 51(11):5247–5252
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2016a) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Los Angeles
Hair JF, Sarstedt M, Matthews LM, Ringle CM (2016b) Identifying and treating unobserved
heterogeneity with FIMIX-PLS: part I–method. Eur Bus Rev 28(1):63–76
Hair JF, Ringle CM, Gudergan SP, Fischer A, Nitzl C, Menictas C (2019a) Partial least squares
structural equation modeling-based discrete choice modeling: an illustration in modeling retailer
choice. Bus Res 12(1):115–142
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019b) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Johnson FR, Lancsar E, Marshall D, Kilambi V, Mühlbacher A, Regier DA et al (2013)
Constructing experimental designs for discrete-choice experiments: report of the ISPOR con-
joint analysis experimental design good research practices task force. Value Health 16(1):3–13
Lenk PJ, DeSarbo WS, Green PE, Young MR (1996) Hierarchical Bayes conjoint analysis:
recovery of partworth heterogeneity from reduced experimental designs. Mark Sci 15(2):
173–191
Louviere JJ, Woodworth G (1983) Design and analysis of simulated consumer choice or allocation
experiments: an approach based on aggregate data. J Mark Res 20(4):350–367
Louviere JJ, Street D, Burgess L, Wasi N, Islam T, Marley AA (2008) Modeling the choices of
individual decision-makers by combining efficient choice experiment designs with extra pref-
erence information. J Choice Model 1(1):128–164
Luce RD, Tukey JW (1964) Simultaneous conjoint measurement: a new type of fundamental
measurement. J Math Psychol 1(1):1–27
McFadden D (1974) Conditional logit analysis of qualitative choice behavior. In: Zarembka P
(ed) Frontiers in econometrics. Academic Press, New York, pp 105–139
Partial Least Squares Structural Equation Modeling-Based Discrete. . . 29

Oliveira-Brochado A, Vitorino Martins F (2014) Identifying small market segments with mixture
regression models. Int J Latest Trends Fin Econ Sci 4(4):812–820
Orme BK, Chrzan K (2017) Becoming an expert in conjoint analysis: choice modelling for Pros:
Sawtooth Software
R Core Team (2019) R: a language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna
Ramaswamy V, DeSarbo WS, Reibstein DJ, Robinson WT (1993) An empirical pooling approach
for estimating marketing mix elasticities with PIMS data. Mark Sci 12(1):103–124
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. SmartPLS GmbH, Boenningstedt
Sawtooth Software (2019) Lighthouse Studio 9 (Version 9.6.1). Sequim, WA: Sawtooth Software,
Inc. https://www.sawtoothsoftware.com/products/online-surveys Accessed 17 Jan 2022
Schuldt J, Doktor A, Lichters M, Vogt B, Robra B-P (2017) Insurees’ preferences in hospital
choice—a population-based study. Health Policy 121(10):1040–1046
Schwarz G (1978) Estimating the dimension of a model. Ann Stat 6(2):461–464
Train KE (2009) Discrete choice methods with simulation. Cambridge University Press, London
The Use of PLS-SEM in Engineering:
A Tool to Apply the Design Science

Ari Melo-Mariano and Ana Bárbara Plá

1 Introduction

The growth in using PLS-SEM in the social sciences is a fact. Hair et al. (2019a)
explain that since 2015 there has been a significant advance in the number of
published papers in the social sciences area, which evidences the growth of the
method’s importance. However, PLS-SEM use has not been limited to the social
sciences; it has demonstrated its use’s versatility in distinct areas such as Knowledge
Management (Cepeda-Carrion et al. 2019), Hospitality and Tourism (Usakli and
Kucukergin 2018), Information Systems Research (e.g., Hair et al. 2017), Psychol-
ogy (e.g., Willaby et al. 2015), Medicine (e.g., Menni et al. 2018), and Engineering
(Aibinu and Al-Lawati 2010; Durdyev et al. 2018).
The adaptability of the PLS-SEM method to different contexts demonstrates its
flexibility in different knowledge areas, positioning it as a “border” tool once it can
consolidate different science fields, becoming applicable in contexts where the limits
between one science field and another are not well known, as Industry 4.0, or even in
the large data volume, as Big Data.
According to Schwab (2019), to think about Industry 4.0 is to comprehend the
increasing harmonization and integration in the different areas’ discoveries making
the fourth industrial revolution unique, the fruit of innovation that resulted from
these technologies collaborations from distinct areas that are already real and are
transforming the society.

A. Melo-Mariano (✉) · A. B. Plá

University of Brasília, Production Engineering Department, DataLab, University Campus Darcy
Ribeiro, Brasília-DF, Brazil
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 31

Industry 4.0 has been a significant engineering concern, mainly as regards the
interconnection with other knowledge areas and the need for a greater comprehen-
sion of the production context.
The same interaction inquiries between the areas occur in the data context
currently receiving the title of Big Data. For example, Shmueli (2017) explains
that one of the Big Data challenges is in the integration of different data sources
(with different volumes, variety, and velocity), many times unknown for the Engi-
neering, that is used to work with inanimate data, which can generate erroneous
analyses when not considering behavioral aspects.
These Engineering challenges favor the search for models and methodologies that
dialogue well with different knowledge areas, maximizing results for the research in
the area. Engineering has been using a few decades of distinguished research
approaches compared to the natural sciences, the Design Science Research. Dresch
et al. (2015) explain that traditional scientific research presents descriptions and
explanations about existing phenomena, but Engineering is not limited to existing
phenomena and systems. It also includes the project and study of phenomena and
objects that do not exist yet but will be created by the engineer for the common good,
which shows the need for a distinguished research approach. Thus, Engineering
research assumes a position based on an artifact that should be developed, validated,
and applied, generating an impact on society.
The new challenges faced by Engineering and its research assumptions based on
design approximate the knowledge area to the current understanding of PLS-SEM
from the composites’ perspective (Henseler 2017; Schuberth et al. 2018), by
adopting the view of artificial sciences, advocated by Simon (1978), separating
what is natural from what is artificial.
This research seeks to answer the question: How can the PLS-SEM be inserted in
the scientific research context in Engineering?
The existing Engineering context is marked by Industry 4.0 and a significant data
background (known as the Big Data era), which sometimes favors exploratory
research, sometimes confirmatory research, with new constructs, experimental
models, or almost experimental, and mainly the scarcity of consolidated literature.
These premises can be well developed in the PLS-SEM context. This way is possible
to comprehend how the PLS-SEM can be a useful Engineering tool that can
collaborate with a better understanding and management of the borders between
the different knowledge areas and kinds, being a connection point and favoring in a
certain way the areas in which Engineering will be connected with to overcome the
new challenges.
Thus, this chapter aims to realize the methodological connections between the
PLS-SEM and Engineering through Design Science.
This chapter is divided into one chapter of background, followed by methodol-
ogy, results, and discussions, and finally, the conclusions and implications for theory
and practice.
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 33

2 Background

2.1 Design Science

According to Dresch et al. (2015), the concept of Design Science (or project science)
was born from observations of the Nobel Prize winner Herbert Simon, from what he
interprets as Artificial Science, a book launched in 1969. In this work, he detaches
what is natural from what is artificial regarding the research object.
In this new context, the artificial is everything that is idealized/designed by man,
as machines, organizations, and economy, and in this way, the research engages in
the study of the creation and projection of new artifacts, or yet, to support the actual
problem solving that is not sustained by the paradigm of the natural sciences (Dresch
et al. 2015).
Aken (2004) describes the design science concept, which includes engineering,
medical sciences, and modern psychotherapy, and has as its research mission:
develop valid and reliable knowledge to design solutions to problems. It is empha-
sized that the goal is not the action itself but the knowledge that will be generated and
could be used to design new solutions, and then, it will be made an action from that
point.
Design Science is a scientific paradigm that emerges to help researchers with a
goal, a prescription, and, consequently, the creation of knowledge about how to
project (Cauchick et al. 2019). It is an approach alternative that promotes investiga-
tions through artifacts that will contribute to the creation of new systems or in the
improvement of the existing ones, aiming for better results in the actual problem
solving, differentiating from the natural sciences that aim the study of known
phenomena and problems, and abandoning the concept of new.
This way, artificial science sees the actions involving man as artifacts, a connec-
tion between the internal environment (artifact organization) and external environ-
ment (operating conditions), projected to answer a particular purpose that should be
validated and given a satisfactory solution.
This construction character favors Engineering research, mainly artifacts in the
solution creation context. This way to think about models created as ingredients of a
construct is to think in artifacts and, at the same time, in composite models of free
constructions, typical of Engineering use.
Design Science as a research approach operationalizes its stages in a context
called Design Science Research, which is essential to comprehend the difference
in the use of both terminologies (one as a method and the other as steps to the
method).
34 A. Melo-Mariano and A. B. Plá

2.2 PLS-SEM (Partial Least Square-Structured Equation

Model)

The technology increment and the need to improve the understanding of data to
decision-making has caused a search for methods to project and deliver relevant
results. In this context emerges the partial least square structural equation
modeling—PLS-SEM, a union of two critical areas, the econometrics, utilizing
predictive models, and the psychometrics, using behavioral models (Hair et al.
2019a).
Structural equation modeling (SEM) is a technique that allows the analysis of
multiple relations between factors that are not observable and are challenging to
measure through variables that function as indicators. Equations are used to evaluate
the relationship between these variables using a statistical toolset.
The variables represent constructs, which are the unobservable factors that will be
analyzed. The SEM then enables modeling and estimating parameters of the relation
between the constructs.
Henseler (2017) defines construct as “constructions that are theoretically justi-
fied.” In other words, it is a term that describes an event that generates theoretical
interest. Usually, the constructs are formed by measures made by observation or
indicators that can represent the concepts.
The SEM realized through the partial least squares (PLS) method becomes a
technique that can estimate the cause-and-effect relationship between the constructs
(Hair et al. 2012). The PLS cannot represent visual concepts in complex models.
However, according to Shmueli et al. (2016), the PLS can “produce parameter
estimates of complex models without many of the distributional and other con-
straints of traditional parametric methods.”
The PLS-SEM is more attractive primarily when the research goal focuses on
predicting and explaining the variance of the primary constructs through the differ-
ent explanation constructs (Hair et al. 2012).
Usually, the constructs that study behavior are treated via common factors, but in
recent years, composite models have been discovered. Behavioral constructs are,
generally, latent variables that can be understood and ontological entities, that is, of
the science of being, as people’s attributes. However, the design constructs are
created as the fruit of thinking (composite constructions) (Henseler 2017), the
same as the artifact in Design Science.
Similar to the Design Science concepts, the PLS-SEM has its models anchored in
artifacts that should be identified in the literature, developed, validated, and applied,
delivering valuable results.
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 35

3 Methodology

This was exploratory research with a qualitative-quantitative approach. It was

conducted in three steps: first, describing the principles of Design Science and
PLS-SEM; second, presenting the principles of Design Science and PLS-SEM to
15 professionals and professors from different engineering fields to opine about the
principles’ adhesion; and third, conducting a systematic review through an adapta-
tion of the theory of consolidated meta-analytical approach from Mariano and
Rocha (2017).
In step 1, many principles were searched in article databases, getting to books
(that had to be acquired by the researchers) that explained the concepts of Design
Science and PLS-SEM in a more didactic way, which was very important to step
2, since the experts and professors evaluations depended of the understanding of
both approaches. In step 2, professionals and professors from different Engineering
areas were searched in person and sometimes online. The information contained in
the Background section of this article and the Design Science stages was presented to
them, named Design Science Research. Some of the interviewees requested addi-
tional material, such as Henseler’s articles. These results showed opinions about the
possible connections between Design Science and PLS-SEM in a text presented in
the results section.
Finally, step 3 is through a systematic review of Web of Science (WoS) and
Scopus databases. The review was made using the strings presented by Khan et al.
(2019) (“partial least squares structural equation modeling” or “partial least squares
structural equation modeling” or “partial least squares path modeling” or “partial
least squares path modeling” or “PLS path modeling” or “PLS path modeling” or
“path model with latent variables” or “PLS-SEM” or “PLS path model” or “PLS-
PM” or “SmartPLS” or “PLSgraph” or “PLS-Graph” or “XLSTAT” or “semPLS” or
“matrixpls” or “ADANCO” or “PLSgui” or “PLS-GUI” or “LVPLS”).
All the papers in Engineering areas in Scopus and WoS were filled with the search
endings referring to PLS-SEM to understand the evolution and use of the method in
Engineering after that. They analyzed the authors through the number of quotes to
have a cluster of the most influential authors (Fig. 1).
The four clusters were analyzed to understand the ecosystem in which all the use
of PLS-SEM in Engineering was organized.
Finally, the union between Design Science and the terms that refer to PLS-SEM
was sought to understand what had already been done from this perspective
(although it was presented last, it was one of the first steps in this chapter, which
motivated the work due to the insufficient number of indexed articles).
To analyze the quote network between authors, it was used the VOS viewer. The
VOS viewer is free software available to create and visualize bibliographic maps.
According to Van Eck and Waltman (2009), the VOS viewer allows the biblio-
graphic map to be shown in different ways, reinforcing an aspect that would prefer.
This allows analysis with more options. In general, it indicates a large amount of data
uses, but its use is also beneficial in cases of little data.
36 A. Melo-Mariano and A. B. Plá

Identification of studies via databases and registers

Records removed before screening:

("Partial least squares structural equation modeling" OR "Partial
least squares structural equation modeling" OR "Partial least
squares path modeling" OR "Partial least squares path
Records identified from*: modeling" OR "PLS path modeling" OR "PLS path modeling"
Identification

Databases Wos (n =13,678,951) OR "path model with latent variables" OR "PLS-SEM" OR "PLS
path model" OR "PLS-PM" OR "SmartPLS" OR "PLSgraph" OR
Databases Scopus (n =8,431,638) "PLS-Graph" OR "XLSTAT" OR "semPLS" OR "matrixpls" OR
"ADANCO" OR "PLSgui" OR "PLS-GUI" OR "LVPLS" AND
Engineering
Records removed WoS(n =8,431304)
Records removed Scopus(n =13,678,510)

Reports sought for retrieval

Screening

Databases Wos (n =334) Records removed

Databases Scopus (n =441) Out of Co-citaton and Coupling
Included

4 clusters WoS
(n = 16)
4 Clusters Scopus
(n =20)

Fig. 1 Research PRISMA process adapted

4 Results and Discussion

4.1 Similarities Between Design Science and PLS-SEM

(Partial Least Square-Structured Equation Model)

A first impression is that Design Science is defined through an organization from the
bottom up. This organization begins with artifacts following constructs, models,
methods, instances, and design propositions (Cauchick et al. 2019):
• Construct: Concepts used to describe problems or specify solutions.
• Models: Set of elements and relations that represent the general structure of
reality.
• Method: Set of logical steps needed to effectuate a specific activity.
• Instances: Execution of the artifacts in their natural environment, highlighting the
viability and efficiency of the artifacts.
• Design proposition: Technological rules or project rules are considered theoret-
ical contributions of Design Science.
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 37

The research with the Design Science approach is oriented to problem-solving,

being of practical nature. With the creation of artifacts support, these solutions aim to
benefit society increasingly. This organization resembles the modeling phases of the
PLS-SEM.
A second impression about Design Science is that the solution it develops doesn’t
have to be a great solution but a satisfactory one. In other words, it means that the
main goal is to solve the problem, not necessarily that it is made in the better way it
can, which could have impediments to its application in the real world. In the same
way, the PLS-SEM is flexible and can present different solutions in different
contexts of the research, such as confirmatory, exploratory, predictive, explanatory,
and descriptive (Henseler 2018).
According to Dresch et al. (2015), the result is considered satisfactory to Design
Science when there is a consensus that there is a progress in the solution achieved
with the new artifact, in comparison with previous solutions, when exists solutions
exist.
Another factor to be seen is related to Design Science Research, the stages of
Design Science. From the comprehension of the Design Science concept, which is a
paradigm, it is possible to understand Design Science Research, a research method-
ology that applies Design Science. Therefore, it is a suggested structure when one
wishes to prescribe solutions or to develop and/or analyze artifacts (Aken 2004).
In Fig. 2, it is possible to verify the steps needed to execute the Design Science
Research. The methodology starts with problem identification, which consists in
identifying, structuring, and comprehending the situation which it is intended to
present a solution. Problem understanding is essential to creating a valuable and
proper solution. Then it must be done a systematic review of the literature, to
justifying all the research that will be done, in other words, to ground the ideas
that will be presented in the paper.
The next step is to identify options for artifacts to be developed to solve the
problem previously detected. This step is usually creative and subjective. After
analyzing the options, a proper artifact must be proposed. Finally, with the artifact
defined, the project to develop the artifact must be created. In Design Science
Research, the researcher also has the role of an artifact designer.
According to the interview, both steps occur accordingly to what occurs in the
PLS-SEM because the relations between the model’s variables must be guided by
the scientific literature or through specialists that can ensure the relation (Hair et al.
2019a).
The next step consists of the execution of the project that will result in the
development or construction of the artifact chosen in its functional state. The
researcher constructs the internal environment of the artifact through algorithms,
graphic models, mockups, and other available tools.
In order to validate the artifact, the next step is to evaluate the artifact. After
defining the requirements that should be validated and which is the desirable
performance of the artifact, the artifact should be evaluated to see if it meets what
is expected of it. Finally, the explicit learnings must be executed after the evaluation
38 A. Melo-Mariano and A. B. Plá

Fig. 2 DSR Methodology. Source: Cauchick et al. (2019)

since it is through it that the improvements needed can be perceived, avoiding

problems in the future use of the artifact.
It was noticed that this same process of validation occurs inside the PLS-SEM,
through the reliability (individual, internal), validity (convergent, discriminating),
and multicollinearity, to all the A mode and B mode, with the validity in a construct
level and indicator level (Hair et al. 2019a, b).
To start the conclusion, all the things that occurred during the research should be
reported. After that, the generalization to a problem class should be made. It must be
said that the technological rule quoted by Aken (2004) is defined as “a general
knowledge that connects an artifact with a result desired in an application field.” He
says that a technological rule is a primary product of Design Science, which is the
same as the generalization of a problem class. In other words, this stage is about the
fact that the solution found is not necessarily just a specific solution to a problem but
the knowledge that could be applied to a problem class similar to the problem faced.
Lastly, the results obtained must be reported. Worth it to say that the paper must
be accessible and must have a description of what has been done and of all the results
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 39

found, including problems and limitations faced, since this information will help
new research and new pieces of knowledge.
The comparison analysis noticed that besides the creation of the artifact as a result
of the Design Science Research to solve problems, there is also the generation of
prescriptive knowledge that meets the same goal of the artifact lined up with the
PLS-SEM context.
This way, it can be perceived that both contexts are connected through its
application stages, making the PLS-SEM a method that incorporates the Design
Science principles.
Once the comparisons were finished, a vision was made of the PLS-SEM use over
the years in Engineering and its evolution of publications and quotes between
authors in Web of Science and Scopus, finishing with the use of the PLS-SEM in
Engineering through Design Science.
The results demonstrate that the oldest record is the Web of Science database in
the Engineering field, using PLS-SEM in the article “Partial Least-Squares Path
Modeling with Latent-Variables” by Gerlach et al. (1979). In this work, the authors
explain that the partial least squares model is a solution to evaluate complex models,
ensuring an evaluation through many sources. This way, it can be perceived that the
PLS-SEM can offer solutions to complex models, being very useful to the current
challenges, like Industry 4.0 or Big Data, because in both situations, complex
models are treated, making the PLS-SEM become a border tool, consolidating
different kinds of knowledge.
The oldest paper about PLS-SEM + Engineering in the Scopus database is
“Transnational terrorism: Prospectus for a causal modeling approach” by Hopple
(1982), where the author proposes a causal model to comprehend the causes of
transnational terrorism. The author explains that the difficulty of this type of model
needs a softer modeling, ensuring flexibility to its calculus. Furthermore, the creation
of models is a human artifact, considered an artifact or artificial object. These
concepts agglutinate the proximity between Engineering through Design Science
and the PLS-SEM.
The most quoted record in the Web of Science database in the Engineering field
was the paper “Using PLS path modeling in new technology research: updated
guidelines” (490 quotes) by Henseler et al. (2016), where authors ratify the increas-
ing use of the tool in the technology field. In the paper, the authors discuss the recent
developments in the structural equations via variance area and the possibility of
working with composites and factors, being characterized as a formidable statistical
tool. It can be perceived that working with composites shows up again as a
preponderant factor in the use of techniques in Engineering.
In the Scopus database, the most quoted record is the book Modelling Transport
(1524 quotes) by de Dios Ortúzar and Willumsen (2011), where the authors present
different kinds of modeling to private and public sectors, providing depth in each
topic, pedagogically, with the discussion of the role of theory, data, model specifi-
cation, estimative, validation, and application. This way, it can be noticed that the
models always followed different Engineering areas, translating the recreation of the
40 A. Melo-Mariano and A. B. Plá

Fig. 3 Publication of articles per year in Scopus and WOS

real in models (artiﬁcial design) comprehensible and feasible to be manipulated on a

smaller scale, ensuring simulations and observations.
The publications with the use of PLS-SEM in both databases present growth in
Engineering, getting to 73 in Web of Science in 2019 and 68 in Scopus (Fig. 3).
The Web of Science database offers the distinction between the different areas of
Engineering. This way was possible to count that from the 334 records, 114 papers
belonged to Industrial Engineering, 80 to Electrical Engineering, 51 to Civil Engi-
neering, and 45 to Environmental Engineering, and the rest was divided between the
other areas.
This way, it can be noticed that the use of PLS, at least in the WoS, is diffuse
between the different areas of Engineering. Therefore, it generated two network
maps to ﬁnd the relationship between authors and discover the central quote core
(Fig. 4).
The ﬁrst cluster is revealed by the papers of Ooi and followers, with the approach
of the effects of new technologies and the way of accepting and using via structural
equations. Gunasekaran forms a second cluster, where evaluated questions related to
manufacturing, such as Lean and green product development. Fifty percent of the
author’s papers are made in Brazil, with the collaboration of Brazilian authors. The
Chong nucleus approach papers focus on innovation in the supply chain. Finally, the
Henseler kernel, in which composite-type structural equation models adhering to
Design Science Theory is shown. The Henseler cluster seen in zoom also reveals the
participation of Ringle presents in his paper “Gain more insight from your PLS-SEM
results The importance-performance map analysis,” the IPMA tool, inside the
SmartPLS software, transforming the results in a business application by revealing
the variables or indicators that must be prioritized, ratifying once again the use of
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 41

Fig. 4 Network of WoS quote relation. Source: Extracted from Vosviewer

Fig. 5 Network of Scopus quote relation. Source: Extracted from Vosviewer

PLS-SEM in Engineering, once the Design Science always defends ﬁnal results of
their studies, applicable.
In the same way, an analysis of the network of authors about the quote dynamic in
the Scopus database was made. It found four different nuclei (Fig. 5).
It can be noticed that Wamba and Gunasekaran formed the nucleus one. Wamba
approaches analysis via PLS-SEM to the business performance in different strategic
contexts, such as Big Data analysis, senior management commitment to
42 A. Melo-Mariano and A. B. Plá

environmental issues, and quality dynamics. Gunasekaran repeats the work identi-
fied in WoS and additionally is a coauthor of the paper about PLS-SEM and Big
Data. Lai brings studies related to China and mainly in the operations area. Muller
works with the study of the supply chains, mainly studying providers.
Bamgbade is shown in a cluster studying external and internal factors linked to
sustainability in construction. Moreover, finally, Henseler, shows up with most of
the papers found in WoS, but with the addition of one editorial about the use of
PLS-SEM in the industrial ambit, published in Industrial Management and Data
Systems (see Henseler 2016). Ringle and Sarstedt are shown together in articles
related to the use of IPMA, just like in WoS.
Both figures were made zoom on the name of Henseler because he is a seminal
author when considering the PLS-SEM context since the composite perspective, the
approach that is directly associated with the context of the artificial design present in
the Design Science is used in Engineering.
The same associations raised in this study were made before by Venable and
Baskerville (2012), where the authors used Design Science Research to analyze the
use of PLS-SEM, associating their stages, but in this occasion, the suggestion is the
use of the PLS-SEM faces to operationalize the Design Science Research.
In 2018 is already shown a study about a model to predict the use of services in
mobile health to bring attendance to regions with few resources (Mburu and Oboko
2018). This model is oriented by Design Science and uses PLS-SEM to
operationalize the steps of Design Science Research.
This way, it can be perceived that the PLS-SEM is increasingly used in Engi-
neering. The development of PLS-SEM and the new perspectives established by
Henseler (Henseler 2017; Schuberth et al. 2018) was significant to establish a more
extensive contact between PLS-SEM and its use in Engineering. Some studies have
already begun to make associations between PLS-SEM and Design Science
Research (Venable and Baskerville 2012) and the use of PLS-SEM to operationalize
Design Science (Mburu and Oboko 2018). This way to call the attention to the use of
PLS-SEM to the challenges faced by Engineering is to establish new possibilities of
research in an interdisciplinary context such as Big Data and Industry 4.0, making
the PLS-SEM a border tool.

5 Conclusion and Implications for Theory and Practice

It can be noticed that the use of PLS-SEM in Engineering has been growing each
year and that the propositions made by Henseler (Henseler 2017; Schuberth et al.
2018) about composite models approximate the PLS-SEM to the Design Science
Research approach. This way, the PLS-SEM is a tool that can contribute in a relevant
way to the interaction challenges between distinct areas, as in the Big Data context,
computer sciences, information systems, software engineering, statistic, and math-
ematics in a more generalist way, or speciﬁc contexts, as philosophy, marketing,
psychology and anthropology (in artiﬁcial intelligence development), and biology
The Use of PLS-SEM in Engineering: A Tool to Apply the Design Science 43

and medicine (bioinformatic), among others. In such a rich context of information

for research modeling, a soft tool, such as PLS-SEM, will help in results that will
converge in the different points of view of the different areas.
In the same way, it can be noticed that to the Industry 4.0 context, the uniﬁcation
of considered complex approaches, in an isolated way, in research and application
(Internet of things, cyber-physics systems, additive manufacture, among others),
potentialized in the integration to create the fourth industrial revolution. This way,
the PLS-SEM in Engineering, following the Design Science approach, can help
create applied models that will help in this knowledge frontier, helping a better
interaction between parts.
This way, the research problem, which was how the PLS-SEM could be inserted
in the scientiﬁc research context in Engineering, was solved. As a suggestion for
future research, we advise a research PLS-SEM following the Design Science
Research steps.

References

Aibinu AA, Al-Lawati AM (2010) Using PLS-SEM technique to model construction organizations’
willingness to participate in e-bidding. Autom Constr 19(6):714–724
Aken JEV (2004) Management research based on the paradigm of the design sciences: the quest for
ﬁeld-tested and grounded technological rules. J Manag Stud 41(2):219–246
Cauchick P, Lacerda D, Abackerli A, Carvalho M, Costa S, Lima E et al (2019) Metodologia
cientíﬁca para engenharia. Elsevier, Brasil
Cepeda-Carrion G, Cegarra-Navarro JG, Cillo V (2019) Tips to use partial least squares structural
equation modelling (PLS-SEM) in knowledge management. J Knowl Manag 23(1):67–89
de Dios Ortúzar J, Willumsen LG (2011) Modelling transport, 4th edn. Wiley, New York
Dresch A, Lacerda DP, Autunes Júnior JAV (2015) Design science research: método de pesquisa
para avanço da ciência e tecnologia. Bookman, Sao Paolo
Durdyev S, Zavadskas EK, Thurnell D, Banaitis A, Ihtiyar A (2018) Sustainable construction
industry in Cambodia: awareness, drivers and barriers. Sustainability 10(2):392
Gerlach RW, Kowalski BR, Wold HO (1979) Partial least-squares path modelling with latent
variables. Anal Chim Acta 112(4):417–421
Hair JF, Sarstedt M, Pieper TM, Ringle CM (2012) The use of partial least squares structural
equation modeling in strategic management research: a review of past practices and recommen-
dations for future applications. Long Range Plan 45(5-6):320–340
Hair J, Hollingsworth CL, Randolph AB, Chong AYL (2017) An updated and expanded assessment
of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hair JF, Hult GTM, Ringle CM, Sarstedt M, Castillo-Apraiz J, Cepeda Carrión G, Roldán JL
(2019a) Manual de partial least squares structural equation modeling (PLS-SEM), Spanish edn.
Sage, Barcelona
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019b) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Henseler J (2016) Guest editorial. Ind Manag Data Syst 116(9):1842–1848
Henseler J (2017) Bridging design and behavioral research with variance-based structural equation
modeling. J Advert 46(1):178–192
Henseler J (2018) Partial least squares path modeling: Quo vadis? Qual Quan Int J Methodol 52(1):
1–8
44 A. Melo-Mariano and A. B. Plá

Henseler J, Hubona G, Ray PA (2016) Using PLS path modeling in new technology research:
updated guidelines. Ind Manag Data Syst 116(1):2–20
Hopple GW (1982) Transnational terrorism: prospectus for a causal modeling approach. Stud Conf
Terrorism 6(1-2):73–100
Khan GF, Sarstedt M, Shiau WL, Hair JF, Ringle CM, Fritze MP (2019) Methodological research
on partial least squares structural equation modeling (PLS-SEM) an analysis based on social
network approaches. Internet Res 29(3):407–429
Mariano AM, Rocha MS (2017) Revisão da Literatura: Apresentação de uma Abordagem
Integradora. In: XXVI Congreso Internacional de la Academia Europea de Dirección y
Economía de la Empresa (AEDEM), Reggio Calabria, Italy, pp 427–443
Mburu S, Oboko R (2018) A model for predicting utilization of mHealth interventions in
low-resource settings: case of maternal and newborn care in Kenya. BMC Med Inform Decis
Mak 18(1):67
Menni C, Lin C, Cecelja M, Mangino M, Matey-Hernandez ML, Keehn L et al (2018) Gut
microbial diversity is associated with lower arterial stiffness in women. Eur Heart J 39(25):
2390–2397
Schuberth F, Henseler J, Dijkstra TK (2018) Conﬁrmatory composite analysis. Front Psychol 9:
2541
Schwab K (2019) A quarta revolução industrial. Edipro, Sao Paolo
Shmueli G (2017) Research dilemmas with behavioral big data. Big Data 5(2):98–119
Shmueli G, Ray S, Estrada JMV, Chatla S, B. (2016) The elephant in the room: predictive
performance of PLS models. J Bus Res 69(10):4552–4564
Simon HA (1978) Rationality as process and as product of thought. Am Econ Rev 68:1–16
Usakli A, Kucukergin KG (2018) Using partial least squares structural equation modeling in
hospitality and tourism: do researchers follow practical guidelines? Int J Contemp Hosp
Manag 30(11):3462–3512
Van Eck N, Waltman L (2009) Software survey: VOSviewer, a computer program for bibliometric
mapping. Scientometrics 84(2):523–538
Venable J, Baskerville R (2012) Eating our own cooking: toward a more rigorous design science of
research methods. Electron J Bus Res Methods 10(2):141–153
Willaby HW, Costa DS, Burns BD, MacCann C, Roberts RD (2015) Testing complex models with
small sample sizes: a historical overview and empirical demonstration of what partial least
squares (PLS) can offer differential psychology. Personal Individ Differ 84:73–78
Discovering Issues in Cross-Cultural
Adaptation of Questionnaire Through
PLS-SEM Analysis

Fariha Reza and Huma Amir

1 Background

Customers who belong to the bottom of the pyramid (BoP) are marginalized from
the mainstream consumer market because of their financial constraints. Literature
suggests they desire to bridge the gap between them and the relatively more affluent
people by purchasing products that reflect a better social status (Srivastava et al.
2020; Yurdakul et al. 2017). One contemporary example of such aspirational
products is smartphones penetrating the BoP segment (Prahalad 2019).
While literature suggests that BoP customers purchase aspirational goods to avoid
poverty-related shame, it is not fully understood whether such purchases increase
healthy engagement with one’s environment and subjective well-being (Dahana
et al. 2018). Furthermore, earlier research suggested that BoP customers sometimes
sacrificed their immediate needs to save enough to purchase their aspirational
products (Atkin et al. 2021). Patience in their everyday lives facilitated such
sacrifices and did not let them despair in challenging circumstances. Not much
Western marketing literature is available on patience; therefore, studying patience
in conjunction with motivation for purchasing aspirational goods, healthy engage-
ment with one’s circumstances, and the subjective well-being of BoP customers
gave novel insights (Haybron 2016). Social identity theory provided the theoretical
justification for studying motivation for buying aspirational products because of
social comparisons that BoP customers made with those who were better off in status

F. Reza (✉)
Institute of Business Management, Karachi, Sindh, Pakistan
e-mail: [email protected]
H. Amir
Institute of Business Administration, Karachi, Sindh, Pakistan
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 45

(Jolliffe and Prydz 2021). On the other hand, self-determination theory (Ryan and
Deci 2000) provided the theoretical background for studying the relationship
between aspirational purchases and healthy engagement with one’s life and the
subjective well-being of BoP customers (Jaikumar et al. 2018). Therefore, this
research empirically tested the relationships between motivation to purchase a
smartphone, intention to purchase a smartphone, self-determination, patience, sub-
jective well-being, and attitude toward life of BoP customers.

2 Methodology

2.1 Population and Sampling

People who had, at times, made an occasional aspirational purchase, despite earning
less than USD 8 per day, were the population of this research (United Nations
Development Program [UNDP] 2008). Furthermore, such people were living in
rented accommodations and were often unable to engage in socially relevant con-
sumption practices, and consequently, they were facing negative social conse-
quences of poverty (Yurdakul et al. 2017). A sample of 641 respondents was
chosen through chain referral (Reza et al. 2021). Data was collected through a
personally administered questionnaire.

2.2 Data Collection Tool and Method

Well-established scales were adopted to measure motivation to purchase a

smartphone, intention to purchase a smartphone, self-determination, subjective
well-being, patience, and attitude toward life of BoP customers (Reza et al. 2021).
The scale items were translated from English to Urdu (Pakistan’s national language)
to make them understandable to the targeted population, keeping in view its shallow
linguistic proﬁciency (Alyami et al. 2021). During pilot testing, it was realized that
the 7-point Likert scale response format contained indistinguishable choices for the
respondents; therefore, the response format was reduced to 5-point scale.

2.3 Choice of SmartPLS for Data Analysis

Various software and applications are available to analyze quantitative data; among
them, SPSS and SmartPLS are more commonly used in market research (Memon
et al. 2021). Compared with SPSS PROCESS macro, PLS-SEM software becomes a
better choice because it can test multiple dependency relationships altogether,
whereas PROCESS macro tests each dependency relationship individually (Hair
Discovering Issues in Cross-Cultural Adaptation of Questionnaire. . . 47

et al. 2019; Hayes et al. 2017; Sarstedt et al. 2022). Hence, multiple dependency
relationships were tested in this research through PLS-SEM analysis through
SmartPLS.
In PLS-SEM analysis, measurement model fit is assessed through outer loadings,
composite reliability, bootstrap confidence interval, convergent reliability, and dis-
criminant validity (Sarstedt et al. 2019). After establishing the goodness of fit for the
measurement model, the analysis proceeds to determine structural model fit through
the coefficient of determination, effect size, Q-Square, and robustness. However, it
should be noted that without determining the goodness of fit for a measurement
model, a researcher should not test the dependency relationships between constructs
that are shown in the structural model (Henseler et al. 2016).

3 Results and Discussion

A reﬂective measurement model was constructed in SmartPLS that complied with

theories of motivation, self-determination, and subjective well-being (Reza et al.
2021). After dropping weak indicators from the model, the composite reliability of
the scales that measured motivation, intention to make aspirational purchases,
patience, and subjective well-being improved (Henseler et al. 2012) as shown in
Table 1.
However, the low values of average variance extracted (AVE) depicted the low
convergent validity of the scales (Hair et al. 2019), as shown in Table 2. Earlier
research suggested that if the composite reliability of all the constructs is greater than
0.6, a researcher may proceed with structural model ﬁt despite AVE being less than
0.5 (Ringle et al. 2015); therefore, researchers proceeded with the structural model ﬁt
analysis despite this limitation.
The discriminant validity of the constructs was determined with the heterotrait-
monotrait ratio (Hair et al. 2021). This ratio was less than 0.9 for the constructs, as
shown in Table 3, and hence acceptable (Hair et al. 2021; Sarstedt et al. 2019).

Table 1 Composite reliability before and after dropping weak indicators

Composite reliability
Scale Using all indicators After dropping weak indicators
ATL 0.867 0.881
ITP 0.837 0.838
Motivation 0.786 0.821
Patience 0.680 0.784
SD 0.692 0.678
SWB 0.463 0.753
Notes: ATL attitude toward life; ITP intention to purchase; SD self-determination; SWB subjective
well-being
48 F. Reza and H. Amir

Table 2 Reliability and validity

Scale Cronbach’s alpha rho_A Composite reliability AVE
ATL 0.884 0.888 0.881 0.275
ITP 0.838 0.841 0.838 0.464
Motivation 0.833 0.832 0.821 0.343
Patience 0.792 0.813 0.784 0.389
SD 0.697 0.765 0.678 0.291
SWB 0.743 0.812 0.753 0.518
Notes: ATL attitude toward life; ITP intention to purchase; SD self-determination; SWB subjective
well-being; AVE average variance extracted

Table 3 HTMT ratio for discriminant validity

Scale ATL ITP Motivation Patience SD
ITP 0.192
Motivation 0.256 0.380
Patience 0.613 0.121 0.228
SD 0.363 0.441 0.410 0.243
SWB 0.406 0.163 0.125 0.286 0.413
Notes: ATL attitude toward life; ITP intention to purchase; SD self-determination; SWB subjective
well-being

The researchers revisited the tool translation and adaptation process to ﬁnd out the
reasons behind low divergent validity. Although the methodology conformed to the
accepted practices established through literature, the unexpected results could be
attributed to the multiple understandings of patience by the targeted population and
the shallowness of communication resulting from educational and social
deprivations.

4 Conclusion

It is reiterated here that multiple definitions of “attitude toward life” and “patience”
exist in the literature (Bülbül and Arslan 2017; Kunieda 2019). However, clarity of
thought requires the medium of language for expression (Kronrod 2022). Due to
social exclusion and living reduced lives, the linguistic proficiency of the BoP
segment is shallow (Piller 2016). As a result, the respondents had difficulty
distinguishing the more nuanced differences between life attitudes and patience.
Another plausible reason for low divergent validity between attitude toward life
and patience could be the respondents’ religiosity. In the psychology of religion,
patience is life-shaping virtue and, therefore, could have been understood as a life
attitude (Bülbül and Arslan 2017).
Discovering Issues in Cross-Cultural Adaptation of Questionnaire. . . 49

The limited linguistic repertoire of the BoP segment and the religiosity were two
plausible reasons that lowered the validity of well-established tools when used in a
different context. These reasons emphasize the need to develop speciﬁc tools in the
native language by native researchers. In this case, Pakistani researchers who are
familiar with the sociocultural characteristics of the BoP population may come
forward to design data collection tools that exhibit greater reliability and validity.
Furthermore, using emojis instead of word choices may be explored when develop-
ing a tool for the BoP population (Sheth 2021). The pervasive use of cell phones
(including smartphones) has raised the level of nonverbal communication through
emojis in the BoP population. Therefore, using emojis might be a more sensitive way
to measure the response of a BoP respondent.
While this paper highlighted the limitations in the translated tool’s compromised
convergent and divergent validity, new research may target tool reﬁnement or new
tool development to measure the BoP attitudes more accurately and with greater
sensitivity. Academic and marketing professionals will appreciate such a develop-
ment as it will enable them to study a marginalized yet multitrillion-dollar segment
more effectively.

References

Alyami M, Henning M, Krägeloh CU, Alyami H (2021) Psychometric evaluation of the Arabic
version of the fear of COVID-19 scale. Int J Ment Heal Addict 19(6):2219–2232
Atkin D, Colson-Sihra E, Shayo M (2021) How do we choose our identity? A revealed preference
approach using food consumption. J Polit Econ 129(4):1193–1251
Bülbül AE, Arslan C (2017) Investigation of patience tendency levels in terms of self-
determination, self-compassion and personality features. Univ J Educ Res 5(9):1632–1645
Dahana WD, Kobayashi T, Ebisuya A (2018) Empirical study of heterogeneous behavior at the base
of the pyramid: the inﬂuence of demographic and psychographic factors. J Int Consum Mark
30(3):173–191
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult GTM, Ringle CM, Sarstedt M, Danks NP, Ray S (2021) Partial least squares structural
equation modeling (PLS-SEM) using R. Springer, Cham
Haybron DM (2016) The philosophical basis of eudaimonic psychology. In: Joar V (ed) Handbook
of eudaimonic well-being. Springer, Cham, pp 27–53
Hayes AF, Montoya AK, Rockwood NJ (2017) The analysis of mechanisms and their contingen-
cies: PROCESS versus structural equation modeling. Australas Mark J 25(1):76–81
Henseler J, Fassott G, Dijkstra TK, Wilson B (2012) Analysing quadratic effects of formative
constructs by means of variance-based structural equation modelling. Eur J Inf Syst 21(1):
99–112
Henseler J, Hubona G, Ray PA (2016) Using PLS path modeling in new technology research:
updated guidelines. Ind Manag Data Syst 116(1):2–20
Jaikumar S, Singh R, Sarin A (2018) ‘I show off, so I am well off’: subjective economic well-being
and conspicuous consumption in an emerging economy. J Bus Res 86:386–393
Jolliffe D, Prydz EB (2021) Societal poverty: a relative and relevant measure. World Bank Econ
Rev 35(1):180–206
Kronrod A (2022) Language research in marketing. Found Trends® Market 16(3):308–421
50 F. Reza and H. Amir

Kunieda M (2019) Designing for meaningfulness in life: creating new meaning. https://aaltodoc.
aalto.ﬁ/bitstream/handle/123456789/39044/master_Kunieda_Masahiro_2019.pdf;sequence=1
Accessed 8 Nov 2021.
Memon MA, Ramayah T, Cheah J-H, Ting H, Chuah F, Cham TH (2021) PLS-SEM statistical
programs: a review. J Appl Struct Eq Model 5(1):i–xiv
Piller I (2016) Monolingual ways of seeing multilingualism. J Multicult Discour 11(1):25–33
Prahalad D (2019, Jan 2) The new fortune at the bottom of the pyramid. Strategy+Business https://
www.strategy-business.com/article/The-New-Fortune-at-the-Bottom-of-the-Pyramid. Accessed
30 May 2022.
Reza F, Amir H, Kazmi HAS (2021) Impact of smartphones, self-determination and patience on
subjective well-being of bottom of pyramid customers. Revista Brasileira De Marketing 20(2):
279–308
Ringle CM, Silva DD, Bido D (2015) Structural equation modeling with the SmartPLS. Brazil J
Market 13(2):56–73
Ryan RM, Deci EL (2000) Self-determination theory and the facilitation of intrinsic motivation,
social development, and well-being. Am Psychol 55(1):68–78
Sarstedt M, Hair JF, Cheah J-H, Becker J-M, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J 27(3):197–211
Sarstedt M, Hair JF, Pick M, Liengaard BD, Radomir L, Ringle CM (2022) Progress in partial least
squares structural equation modeling use in marketing research in the last decade. Psychol Mark
39(5):1035–1064
Sheth J (2021) New areas of research in marketing strategy, consumer behavior, and marketing
analytics: the future is bright. J Mark Theory Pract 29(1):3–12
Srivastava A, Mukherjee S, Jebarajakirthy C (2020) Aspirational consumption at the bottom of
Pyramid: a review of Literature and Future Research Directions. J Bus Res 110:246–259
United Nations Development Programme UNDP (2008) Creating value for all: strategies for doing
business with the poor. https://www.undp.org/rwanda/publications/creating-value-all-strate
gies-doing-business-poor Accessed 21 Sept 2022.
Yurdakul D, Atik D, Dholakia N (2017) Redeﬁning the bottom of the pyramid from a marketing
perspective. Mark Theory 17(3):289–303
Use of PLS-SEM Approach
in the Construction Management Research

Sachin Batra

1 Introduction

Partial least squares structural equation modeling (PLS-SEM) is a contemporary

multivariate data analysis method to estimate hypothetically established cause-effect
relationship models (Zeng et al. 2021). Many scholars have demonstrated the use of
PLS-SEM in different business disciplines, namely, marketing, strategic manage-
ment, management information systems, international business, human resource
management, operations management, supply chain management, accounting, and
tourism and hospitality (Lee et al. 2011; Hair et al. 2012; Ringle et al. 2012; Peng
and Lai 2012; Kaufmann and Gaeckler 2015; Richter et al. 2016; Sarstedt et al.
2020). But in the construction management domain, scholars are less acquainted
with the PLS-SEM approach (Zeng et al. 2021). However, since 2010 when the first
article was published in the construction management domain using the PLS-SEM
approach, Scopus search results indicate an exponential increase in the number of
articles, with 2304 published in 2021.
Therefore, in this chapter, the author attempted to provide a snapshot of the
scientific activity of the use of the PLS-SEM approach in the construction manage-
ment domain by answering the following research questions:
• What are the most influential studies using PLS-SEM in the construction sector?
• What is the co-citation network?
• What are the domains and levels of analysis in which PLS-SEM is used in
construction management research?

S. Batra (✉)
NICMAR University, Pune, Maharashtra, India

© Springer Nature Switzerland AG 2023 51

2 Methodology

The mixed method approach is used in the present work, as suggested by Harden and
Thomas (2010). However, several authors have highlighted subjective judgment
issues using only the systematic review method (He et al. 2017). The mixed-method
approach will overcome the mono-method problems as it combines and applies the
methods for synthesizing and analyzing the available literary work in the field of
inquiry (Harden and Thomas 2010). Figure 1 presents the preferred reporting items
for systematic review and meta-analysis (PRISMA) flow diagram.
As shown in Fig. 1, the first analysis stage involves bibliometric analysis
followed by full content analysis. Bibliometric analysis refers to exploring and
visualizing a large volume of the scientific database in the field of inquiry (van
Eck and Waltman 2010). Many tools exist for bibliometric analysis, of which
biblioshiny (Aria and Cuccurullo 2017) is used in the present study.

Fig. 1 PRISMA ﬂow diagram

Use of PLS-SEM Approach in the Construction Management Research 53

Different bibliometric sources such as Web of Science, Scopus, EBSCOhost, or

ProQuest could be used to obtain data for bibliometric analyses. Compared to other
databases, Scopus covers a more excellent selection of journals in the ﬁeld of
construction management than Web of Science and contains more recent publica-
tions (Chadegani et al. 2013). This validated the usage of Scopus as a data retrieval
source.

3 Results and Discussion

3.1 Bibliometric Analysis

3.1.1 Most Inﬂuential Studies

The top ten most inﬂuential studies concerning total citation score are identiﬁed
through biblioshiny, as given in Table 1. The study by Banihashemi et al. (2017),
published in the International Journal of Project Management, leads the category
with 177 total citations, followed by Aibinu and Al-Lawati (2010), published in
Automation in Construction with a 144 total citation score.

3.1.2 Co-citation Network

The co-citation network reveals two clusters (Fig. 2). In the blue-colored cluster,
scholars have used the PLS-SEM approach by referring to methodological articles;
for example, Hair et al. (2011) where the articles also mentioned the use of
SmartPLS software, while in the red-colored cluster, scholars referred to Kock
(2017) and relied on WarpPLS software for analysis.

Table 1 Top 10 most inﬂuential studies of PLS-SEM in the construction sector

Author DOI Total citations
Banihashemi et al. (2017) 10.1016/j.ijproman.2017.01.014 177
Aibinu and Al-Lawati (2010) 10.1016/j.autcon.2010.02.016 144
Cao et al. (2014) 10.1061/(ASCE)CO.1943-7862.0000903 123
Lu et al. (2015) 10.1016/j.ijproman.2014.03.004 121
Le et al. (2014) 10.1061/(ASCE)CO.1943-7862.0000886 104
Hosseini et al. (2016) 10.5130/AJCEB.v16i3.5159 102
Darko et al. (2018) 10.1016/j.jclepro.2018.07.318 94
Hussain et al. (2018) 10.3390/su10051415 91
Durdyev et al. (2018) 10.1016/j.jclepro.2018.08.304 69
Wang et al. (2016) 10.1016/j.ijproman.2016.07.004 68
54 S. Batra

Fig. 2 Co-citation network

3.2 Full Content Analysis

The 24 most influential studies based on citations are selected and thoroughly
studied to identify the level of analysis using PLS-SEM presented in Table 2. The
intention is to explore the answers to the research questions defined in the introduc-
tion section. The following seven broad fields are identified in which PLS-SEM is
used in the construction sector:
• Sustainability: 9 studies (Avotra et al. 2021; Banihashemi et al. 2017; Darko et al.
2018; Durdyev et al. 2018; Hussain et al. 2018; Jain et al. 2020; Kineber et al.
2021; Yusof et al. 2016, 2017).
• Building Information Modeling: 5 studies (Ahuja et al. 2016; Cao et al. 2014;
Chen et al. 2019; Hosseini et al. 2016; Zhao et al. 2018).
Use of PLS-SEM Approach in the Construction Management Research 55

Table 2 Analysis of the 24 most inﬂuential studies

S
No Authors B A Domain
1 Ahuja et al. (2016) • Building Information Modeling
2 Aibinu et al. (2011) • Project Claim
3 Aibinu and Al-Lawati (2010) • Procurement
4 Avotra et al. (2021) • Sustainability
5 Abu Bakar et al. (2016) • Project Performance
6 Banihashemi et al. (2017) • Sustainability
7 Cao et al. (2014) • Building Information Modeling
8 Chen et al. (2019) • Building Information Modeling
9 Darko et al. (2018) • Sustainability
10 Durdyev et al. (2018) • Sustainability
11 Hussain et al. (2018) • Sustainability
12 Hosseini et al. (2016) • Building Information Modeling
13 Jain et al. (2020) • Sustainability
14 Kineber et al. (2021) • Sustainability
15 Le et al. (2014) • Human Resource Management
16 Lu et al. (2015) • Project Performance
17 Wang et al. (2016) • Human Resource Management
18 Yusof et al. (2016) • Sustainability
19 Yusof et al. (2017) • Sustainability, Project Performance
20 Zaman (2020) • Project Success, Human Resource
Management
21 Zhang and Qian (2017) • Risk Management
22 Zhao and Singhaputtangkul • Risk Management
(2016)
23 Zhao et al. (2018) • Risk Management, Building Information
Modeling
24 Zuo et al. (2018) • Human Resource Management, Project
Success
Notes: “B” is the basic level of analysis; “A” is the advanced level of analysis

• Project Success/Project Performance: 5 studies (Abu Bakar et al. 2016; Lu et al.

2015; Yusof et al. 2017; Zaman 2020; Zuo et al. 2018).
• Human Resource Management: 4 studies (Le et al. 2014; Wang et al. 2016;
Zaman 2020; Zuo et al. 2018).
• Risk Management: 3 studies (Zhang and Qian 2017; Zhao and Singhaputtangkul
2016; Zhao et al. 2018).
• Project Claim: 1 study (Aibinu et al. 2011).
• Procurement: 1 study (Aibinu and Al-Lawati 2010).
56 S. Batra

4 Conclusions, Implications, and Future Directions

Fertile grounds for research using PLS-SEM in construction management came to

light as the outcome of this chapter. Seven domains identified where construction
management scholars commonly use PLS-SEM are Sustainability, Building Infor-
mation Modeling, Project Success/Project Performance, Human Resource Manage-
ment, Risk Management, Project claims, and Procurement. These domains will
provide the direction for the scholars to explore further and take advantage of
using PLS-SEM.
The results presented in Table 2 indicate that almost 78% of the articles applied
PLS-SEM at the basic level of analysis. Hence, to get better insights into the research
work in the construction management area, there exists broad scope for using
PLS-SEM at the advanced level. For example, the use of heterogeneity (Batra
2023), higher-order constructs, measurement invariance, cross-validated predictive
ability test (CVPAT), endogeneity (Batra and Rastogi 2023), confirmatory tetrad
analysis, finite mixture analysis, and multiple mediation analysis. Further, failure to
use advanced techniques may lead to misleading interpretations.

References

Abu Bakar AH, Yusof MN, Tufail MA, Virgiyanti W (2016) Effect of knowledge management on
growth performance in construction industry. Manag Decis 54(3):735–749
Ahuja R, Jain M, Sawhney A, Arif M (2016) Adoption of BIM by architectural ﬁrms in India:
technology–organization–environment perspective. Architect Eng Design Manag 12(4):
311–330
Aibinu AA, Al-Lawati AM (2010) Using PLS-SEM technique to model construction organizations’
willingness to participate in e-bidding. Autom Constr 19(6):714–724
Aibinu AA, Ling FYY, Ofori G (2011) Structural equation modelling of organizational justice and
cooperative behaviour in the construction project claims process: contractors’ perspectives.
Constr Manag Econ 29(5):463–481
Aria M, Cuccurullo C (2017) bibliometrix: An R-tool for comprehensive science mapping analysis.
J Informet 11(4):959–975
Avotra AARN, Chenyun Y, Yongmin W, Lijuan Z, Nawaz A (2021) Conceptualizing the state of
the art of corporate social responsibility (CSR) in green construction and its nexus to sustainable
development. Front Environ Sci 9:774822
Banihashemi S, Hosseini MR, Golizadeh H, Sankaran S (2017) Critical success factors (CSFs) for
integration of sustainability into construction project management practices in developing
countries. Int J Proj Manag 35(6):1103–1119
Batra S (2023) How to use structural equation modeling. SAGE Publications Ltd., https://doi.org/
10.4135/9781529670325
Batra S, Rastogi A (2023) Authentic leadership, future prospects and affective commitment in early
career employees—A mediation model. South Asian J Human Resources Manage. https://doi.
org/10.1177/23220937221148074
Cao D, Li H, Wang G (2014) Impacts of isomorphic pressures on BIM adoption in construction
projects. J Constr Eng Manag 140(12):04014056
Chadegani AA, Salehi H, Yunus MM, Farhadi H, Fooladi M, Farhadi M et al (2013) A comparison
between two main academic literature collections: Web of Science and Scopus databases. Asian
Soc Sci 9(5):18–26
Use of PLS-SEM Approach in the Construction Management Research 57

Chen Y, Yin Y, Browne GJ, Li D (2019) Adoption of building information modeling in Chinese
construction industry: the technology-organization-environment framework. Eng Constr Archit
Manag 26(9):1878–1898
Darko A, Chan APC, Yang Y, Shan M, He BJ, Gou Z (2018) Influences of barriers, drivers, and
promotion strategies on green building technologies adoption in developing countries: the
Ghanaian case. J Clean Prod 200(1):687–703
Durdyev S, Ismail S, Ihtiyar A, Bakar NFSA, Darko A (2018) A partial least squares structural
equation modeling (PLS-SEM) of barriers to sustainable construction in Malaysia. J Clean Prod
204:564–572
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Sarstedt M, Ringle CM, Mena JA (2012) An assessment of the use of partial least squares
structural equation modeling in marketing research. J Acad Mark Sci 40(3):414–433
Harden A, Thomas J (2010) Mixed methods and systematic reviews: examples and emerging
issues. In: Tashakkori A, Teddlie C (eds) Sage handbook of mixed methods in social &
behavioral research, 2nd edn. Sage, Thousand Oaks
He Q, Wang G, Luo L, Shi Q, Xie J, Meng X (2017) Mapping the managerial areas of Building
Information Modeling (BIM) using scientometric analysis. Int J Proj Manag 35(4):670–685
Hosseini M, Banihashemi S, Chileshe N, Namzadi MO, Udaeja C, Rameezdeen R, McCuen T
(2016) BIM adoption within Australian Small and Medium-sized Enterprises (SMEs): an
innovation diffusion model. Construct Econ Build 16(3):71–86
Hussain S, Fangwei Z, Siddiqi AF, Ali Z, Shabbir MS (2018) Structural equation model for
evaluating factors affecting quality of social infrastructure projects. Sustainability 10(5):1415
Jain S, Singhal S, Jain NK, Bhaskar K (2020) Construction and demolition waste recycling:
investigating the role of theory of planned behavior, institutional pressures and environmental
consciousness. J Clean Prod 263:121405
Kaufmann L, Gaeckler J (2015) A structured review of partial least squares in supply chain
management research. J Purch Supply Manag 21(4):259–272
Kineber AF, Othman I, Oke AE, Chileshe N, Zayed T (2021) Exploring the value management
critical success factors for sustainable residential building – a structural equation modelling
approach. J Clean Prod 293:126115
Kock N (2017) WarpPLS user manual: Version 6.0. USA: ScriptWarp Systems
Le Y, Shan M, Chan AP, Hu Y (2014) Investigating the causal relationships between causes of and
vulnerabilities to corruption in the Chinese public construction sector. J Constr Eng Manag
140(9):05014007
Lee L, Petter S, Fayard D, Robinson S (2011) On the use of partial least squares path modeling in
accounting research. Int J Account Inf Syst 12(4):305–328
Lu P, Guo S, Qian L, He P, Xu X (2015) The effectiveness of contractual and relational
governances in construction projects in China. Int J Proj Manag 33(1):212–222
Peng DX, Lai F (2012) Using partial least squares in operations management research: a practical
guideline and summary of past research. J Oper Manag 30(6):467–480
Richter NF, Sinkovics RR, Ringle CM, Schlägel C (2016) A critical look at the use of SEM in
international business research. Int Mark Rev 33(3):376–404
Ringle CM, Sarstedt M, Straub DW (2012) Editor’s comments: a critical look at the use of
PLS-SEM. MIS Q 36(1):iii–xiv
Sarstedt M, Ringle CM, Cheah JH, Ting H, Moisescu OI, Radomir L (2020) Structural model
robustness checks in PLS-SEM. Tour Econ 26(4):531–554
Van Eck N, Waltman L (2010) Software survey: VOSviewer, a computer program for bibliometric
mapping. Scientometrics 84(2):523–538
Wang CM, Xu BB, Zhang SJ, Chen YQ (2016) Influence of personality and risk propensity on risk
perception of Chinese construction project managers. Int J Proj Manag 34(7):1294–1304
Yusof NA, Abidin NZ, Zailani SHM, Govindan K, Iranmanesh M (2016) Linking the environ-
mental practice of construction firms and the environmental behaviour of practitioners in
construction projects. J Clean Prod 121:64–71
58 S. Batra

Yusof NA, Awang H, Iranmanesh M (2017) Determinants and outcomes of environmental practices
in Malaysian construction projects. J Clean Prod 156:345–354
Zaman U (2020) Examining the effect of xenophobia on “transnational” mega construction project
(MCP) success: moderating role of transformational leadership and high-performance work
(HPW) practices. Eng Constr Archit Manag 27(5):1119–1143
Zeng N, Liu Y, Gong P, Hertogh M, König M (2021) Do right PLS and do PLS right: a critical
review of the application of PLS-SEM in construction management research. Front Eng Manag
8(3):356–369
Zhang L, Qian Q (2017) How mediated power affects opportunism in owner–contractor relation-
ships: the role of risk perceptions. Int J Proj Manag 35(3):516–529
Zhao X, Singhaputtangkul N (2016) Effects of ﬁrm characteristics on enterprise risk management:
case study of Chinese construction ﬁrms operating in Singapore. J Manag Eng 32(4):05016008
Zhao X, Wu P, Wang X (2018) Risk paths in BIM adoption: empirical study of China. Eng Constr
Archit Manag 25(9):1170–1187
Zuo J, Zhao X, Nguyen QBM, Ma T, Gao S (2018) Soft skills of construction project management
professionals and project success factors: a structural equation model. Eng Constr Archit Manag
25(3):425–442
Part II
Consumer Behavior and Marketing
Understanding the Role of Consumer
Psychological Motives in Smart Connected
Objects Appropriation: A Higher-Order
PLS-SEM Approach

Zeling Zhong and Christine Balagué

1 Introduction

By 2025, there will be more than 75 billion Internet of Things (IoT) connected
devices in use in the world, according to Statista (2016). Despite the success of these
devices, extant literature shows the existence of barriers limiting their mass diffu-
sion. One of the major challenges is about how to integrate SCOs in consumers’
daily routines, which corresponds to consumer SCO appropriation. However, liter-
ature mainly focused on the acceptance context (Chuah et al. 2016; Huarng et al.
2022; Hubert et al. 2019; Lu et al. 2019); little research has studied SCO users’
appropriation stage. To ﬁll this gap, our research investigates the role of psycholog-
ical motives in SCO appropriation from a consumer perspective. To test our hierar-
chical framework, we follow the latest methodological developments about
composite-based models (Schamberger et al. 2020) and hierarchical component
models (Crocetta et al. 2021; Sarstedt et al. 2019) in partial least square structural
equation modelling (PLS-SEM) approach.

Z. Zhong (✉)
EDC Paris Business School, OCRE Research Lab, Puteaux, France
e-mail: [email protected]
C. Balagué
Paris-Saclay University, Univ Evry, IMT-BS, LITEM, Evry-Courcouronnes, France
Institut Mines-Télécom Business School, Evry, France
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 61

2 Background

Understanding consumer experience with SCO is a priority in today’s interactive

marketing research (Wang 2021). A great deal of research has analyzed consumer
SCO acceptance by applying traditional models of early adoption. In today’s
technology-enhanced world, more recent attention focused on a deeper understand-
ing of consumer experience with SCO by going beyond the early adoption stage
(Benamar et al. 2020; Hoffman and Novak 2018; Kirk et al. 2015; Zhong 2019).
Appropriation is the way in which individuals use, adapt, and integrate a technology
into their daily routines by taking possession of the technology (Carroll et al. 2003).
Most existing work on technology appropriation analyzes collective appropriation.
The individual level of appropriation has recently obtained a growing attention in
marketing research (Kirk et al. 2015). Sartre (1943) proposed three appropriation
modes: control (mastery of the technology by individuals), knowledge (information
regarding the characteristics of technology retained in individual’s memory), and
creation (changes to the technology influenced or made by individuals to better fit
their needs). Adapting this framework to marketing context, Belk (1988) suggests a
fourth appropriation mode—contamination—and confirms the relevance of the
mentioned appropriation modes in digital context (Belk 2017). Adopting these
modes, Mifsud et al. (2015) add two more appropriation modes in a service context:
psychological possession (individuals consider a technology as “it’s my own”) and
consciousness (awareness of the effects of a technology). Zhong and Balagué (2019)
confirm the relevance of the abovementioned appropriation modes in the IoT
context. To achieve theoretical parsimony (Hair et al. 2019a), we conceptualize
SCO appropriation as a higher-order construct with its concrete first-order
subdimensions. Based on the theoretical foundations of individual appropriation,
the conceptualization of the multidimensional construct “SCO appropriation” is
more suitable under the composite view (Law and Wong 1999): SCO appropriation
is the outcome of its concrete subdimensions. In line with MacKenzie et al. (2005),
after checking for five conditions of formative construct development, we formulate
the following hypothesis:
H1: Consumer SCO appropriation is a higher-order construct composed of six
reflectively measured first-order dimensions: psychological possession, self-
adaptation, knowledge, control, consciousness, and creation.
Self-identity is the personal identity developed by individuals. Three elements are
relevant to the need for self-identity (Pierce et al. 2003): coming to know oneself
(individuals want to know more about themselves), expressing self-identity (indi-
viduals want to express themselves to others), and maintaining self-identity (indi-
viduals want to maintain the continuity of self-image with their past). A large
number of prior studies (e.g., Belk 2013, 2017; Pierce and Jussila 2011; Reed
et al. 2012) describe the link between the motive for self-identity and the feelings
of possession toward objects which could be considered as a self-extension.
Understanding the Role of Consumer Psychological Motives in. . . 63

Individuals experience the feeling of possession while appropriating the target by

making it their own (Kirk et al. 2015). Thus, in the SCO context:
H2: Self-identity positively influences consumer SCO appropriation.
H2a: Coming to know self-identity positively influences consumer SCO appropriation.
H2b: Expressing self-identity positively influences consumer SCO appropriation.
H2c: Maintaining the self-identity positively influences consumer SCO
appropriation.
Individuals have a need for territoriality (Ardrey 1966); they long for owning a
space. Pierce et al. (2003) suggest that humans need a physical or virtual place that
makes them feel psychically comfortable, familiar, as at home. The motive for
owning a certain place is linked to possession feelings (Ardrey 1966). In the SCO
context, individuals would experience the feeling of possession while appropriating
the SCO by integrating it into their daily routines if they perceive it as their
territory, thus:
H3: Territoriality positively influences consumer SCO appropriation.
Individuals have a need to be effective and to feel competent for effectively
interacting with their environment (White 1959). Prior work describes a positive
relationship between the need for effectance and efficacy and the feeling of posses-
sion toward a target (Dittmar 1992; Van Dyne and Pierce 2004). In the SCO context,
individuals would appropriate the SCO to fit their need for effectively interacting
with their environment. Thus:
H4: Efficacy and effectance positively influences consumer SCO appropriation.

3 Methodology

To enhance the external validity of our study, we collected data with a quantitative
survey using a large marketing research panel in France from Kantar on 505 French
smartwatch users. To test our conceptual model, we applied PLS-SEM in software
SmartPLS 3.3.9 (Ringle et al. 2015), since PLS-SEM can efficiently handle forma-
tive constructs and complex models (Hair et al. 2019b; Sarstedt et al. 2016). We
measured all the constructs with validated scales from previous literature adapted to
the SCO context: subdimensions of appropriation (Mifsud 2016), self-identity which
is a second-order construct formed by three subdimensions (Karahanna et al. 2015),
territoriality (Zou et al. 2017), and efficacy and effectance (Bellini et al. 2016). We
measured items with five-point Likert scales ranging from “strongly disagree” to
“strongly agree” (see Appendix).
64 Z. Zhong and C. Balagué

4 Results and Discussion

4.1 First-Order Measurement Model

The reliability and validity of the first-order reflective measurement model are
satisfactory. The assessment results (Table 1) about items-loadings (Carmines and
Zeller 1979), Cronbach’s alpha (Nunnally 1978), average variance extracted (AVE)
(Fornell and Larcker 1981), composite reliability (Hair et al. 2011), and Dijkstra-
Henseler’s ρ values (Dijkstra and Henseler 2015; Rademaker et al. 2019) confirmed
the convergent validity and internal consistency reliability. Discriminant validity is
also satisfactory, confirmed by results (Table 2) regarding the heterotrait–monotrait
ratio of correlations (HTMT) after using a bootstrapping procedure with 5000 sub-
samples (Franke and Sarstedt 2019; Henseler et al. 2015).

4.2 Second-Order Measurement Model

We assessed the relationship between lower-order components and higher-order

components. Hypothesis 1 is supported by evidence (Table 3) from confirmatory
tetrad analysis in PLS (Hair et al. 2017).
To validate the formative higher-order SCO appropriation construct and self-
identity construct, we followed the procedure outlined in Hair et al. (2022). The
evidence from a bootstrapping procedure (5000 subsamples) supported the conver-
gent validity (Hair et al. 2022) of higher-order components by running a redundancy
analysis (Chin 1998). The results of the variance inflation factor (VIF) (Hair et al.
2011) showed no collinearity issues among the first-order constructs. We examined
the weights and significance of the first-order components and their higher-order
components (Fig. 1) by using a bootstrapping procedure (5000 subsamples). The
results provided clear evidence for the validity of the reflective-formative nature of
SCO appropriation construct and self-identity construct.

4.3 Structural Model

Following the recommendations of Sarstedt et al. (2021) and Hair et al. (2019a, b),
we assessed structural model based on collinearity inspection, signiﬁcance and
relevance of hypothesized relationships (Fig. 2), and explanatory power and out-
of-sample predictive power (R2, Q2, PLSpredict). To estimate higher-order con-
structs, we applied the disjoint two-stage approach which permits the application of
all structural model evaluation criteria, to minimize the parameter bias in our
structural model relationships (Sarstedt et al. 2019). Evidence from bootstrapping
procedure regarding all inner VIF (Hair et al. 2011) showed no collinearity issues.
Understanding the Role of Consumer Psychological Motives in. . . 65

Table 1 Internal consistency reliability and convergent validity of the ﬁrst-order constructs
Convergent validity Internal consistency reliability
First-order Manifest Composite Cronbach’s
construct variable Loadings AVE reliability rho_A alpha
> 0.70 > 0.50 0.70–0.95 > 0.70 > 0.70
Consciousness (C) C1 0.820 0.667 0.889 0.835 0.834
C2 0.803
C3 0.804
C4 0.840
Control (CO) CO1 0.773 0.650 0.903 0.876 0.866
CO2 0.765
CO3 0.840
CO4 0.793
CO5 0.856
Creation (CR) CR1 0.788 0.703 0.904 0.861 0.859
CR2 0.865
CR3 0.851
CR4 0.848
Knowledge (K) K1 0.834 0.672 0.911 0.887 0.878
K2 0.825
K3 0.814
K4 0.783
K5 0.840
Efﬁcacy and EE1 0.803 0.642 0.926 0.909 0.907
effectance (EE) EE2 0.789
EE3 0.826
EE4 0.802
EE5 0.829
EE6 0.772
MEE7 0.787
Come to know CKO1 0.878 0.797 0.922 0.873 0.872
oneself (CKO) CKO2 0.908
CKO3 0.891
Express self- ESI1 0.889 0.750 0.900 0.834 0.833
identity (ESI) ESI2 0.865
ESI3 0.844
Territoriality (T) T1 0.847 0.701 0.934 0.917 0.915
T2 0.832
T3 0.821
T4 0.850
T5 0.839
T6 0.834
Maintain self- MSI1 0.850 0.688 0.869 0.775 0.773
identity (MSI) MSI2 0.842
MSI3 0.794
(continued)
66 Z. Zhong and C. Balagué

Table 1 (continued)
Convergent validity Internal consistency reliability
First-order Manifest Composite Cronbach’s
construct variable Loadings AVE reliability rho_A alpha
Psychological PP2 0.849 0.749 0.900 0.833 0.833
possession (PP) PP3 0.877
PP4 0.870
Self-adaptation SA1 0.896 0.828 0.935 0.900 0.897
(SA) SA2 0.917
SA3 0.917

Evidence from blindfolding and PLSpredict procedures (Hair et al. 2019b; Shmueli
et al. 2019) supported a strong explanatory power and a medium predictive power of
the structural model.

5 Conclusions and Implications for Theory and Practice

Our study provides meaningful implications. First, our research enhances the under-
standing of the role of consumer psychological motives in SCO appropriation and
strengthens the conceptual framework for future research. Second, our study empir-
ically validates consumer SCO appropriation as a higher-order construct based on
six underlying sets of practice. The findings support three significant psychological
drivers of SCO appropriation. The study provides theoretical and empirical insights
to individual technology appropriation. Third, our study provides evidence that the
new developments of PLS-SEM are extremely suitable for analyzing complexities
and dynamics in consumer behavior, especially when modelling hierarchical
composite-based concepts. Our study also provides important insights for practice.
The findings could help SCO manufacturers and marketing managers to better
understand connected consumers’ motives to integrate SCO in their daily routines
as well as their concrete appropriation practices. This deeper understanding would
allow SCO practitioners to better adapt their offerings to SCO market. In return,
consumers whose needs are better met will engage into more loyal behaviors when
“taking possession” of a SCO and the associated services that they perceive as “it’s
my own.”
Table 2 Discriminant validity of the first-order constructs
Efficacy Come to Express Maintain
and know self- self- Psychological
Consciousness Control Creation Knowledge effectance oneself identity Territoriality identity possession
Consciousness
Control 0.829
Creation 0.713 0.676
Knowledge 0.823 0.891 0.668
Efficacy and 0.618 0.709 0.611 0.746
effectance
Come to know 0.443 0.276 0.436 0.353 0.330
oneself
Express self- 0.442 0.340 0.423 0.355 0.446 0.677
identity
Territoriality 0.448 0.361 0.658 0.372 0.441 0.494 0.568
Understanding the Role of Consumer Psychological Motives in. . .

Maintain self- 0.389 0.325 0.426 0.296 0.381 0.527 0.670 0.563
identity
Psychological 0.604 0.423 0.817 0.465 0.437 0.475 0.457 0.763 0.410
possession
Self- 0.509 0.347 0.547 0.341 0.231 0.415 0.358 0.580 0.291 0.725
adaptation
67
68

Table 3 CTA-PLS results for SCO appropriation

Standard
Original Sample deviation t-statistics (|O/ p- CI Low CI Up
SCO appropriation sample (O) mean (M) (STDEV) STDEV|) values Bias adj. adj.
1: Consciousness. Control. Creation. Knowledge -0.053 -0.052 0.032 1.674 0.095 0.001 -0.142 0.035
2: Consciousness. Control. Knowledge. Creation 0.001 0.000 0.023 0.026 0.979 -0.001 -0.062 0.064
4: Consciousness. Control. Creation. Psycho- 0.269 0.264 0.048 5.650 0.000 -0.005 0.141 0.406
logical possession
6: Consciousness. Creation. Psychological pos- -0.072 -0.071 0.015 4.622 0.000 0.001 -0.116 -0.029
session. Control
7: Consciousness. Control. Creation. Self- 0.153 0.150 0.038 3.992 0.000 -0.003 0.050 0.263
adaptation
10: Consciousness. Control. Knowledge. Psy- 0.026 0.024 0.019 1.376 0.169 -0.001 -0.025 0.078
chological possession
16: Consciousness. Control. Psychological pos- 0.288 0.283 0.048 6.037 0.000 -0.005 0.160 0.425
session. Self-adaptation
22: Consciousness. Creation. Knowledge. Self- -0.154 -0.152 0.037 4.196 0.000 0.002 -0.259 -0.054
adaptation
26: Consciousness. Creation. Self-adaptation. 0.072 0.071 0.037 1.942 0.053 -0.001 -0.030 0.177
Psychological possession
Z. Zhong and C. Balagué
Understanding the Role of Consumer Psychological Motives in. . . 69

Fig. 1 Results of assessment of the second-order formative measurement model

Fig. 2 Structural model test result

70 Z. Zhong and C. Balagué

Appendix

Questionnaire
Higher-order First-order
construct construct Items
SCO appropriation Consciousness I am aware of what I need to do to make my daily life
easier.
I can see the different tasks I have to do to take advantage
of my SCO.
I understand what is expected of me in my care (regular
use of associated services, respect for the instructions for
use).
I am aware of the uses that the SCO manufacturers expect
of me.
Control I feel in control of what I do.
I feel able to take charge of myself.
It is easy for me to realize the uses.
I have the feeling to master the uses that the SCO manu-
facturers expect from me.
I feel I can manage the uses initially conceived.
Creation I feel like I can choose the uses of my SCO.
I feel I have adapted this SCO to my everyday life.
I feel I have adapted my SCO to my everyday life.
I feel I have created a SCO that suits me.
Knowledge I feel like I know the SCO well that makes my daily life
easier.
I am familiar with the SCO I use.
I feel I know my SCO well and how to use it.
Compared to most users, I know quite a bit about my
SCO.
For those around me, I am someone who knows my
SCO well.
Psychological This SCO belongs to me.
possession I talk about my SCO, as MY “SCO.”
This SCO is part of me.
This SCO is part of my life.
Self- This SCO has led me to organize my days differently.
adaptation This SCO has led me to adapt my daily life.
This SCO has led me to change my pace of life.
Self-identity Come to know I feel a need to develop a sense of self-identity.
oneself I feel a need to discover what kind of person I am.
I feel a need to learn about myself.
Express self- I feel a need to express who I am.
identity I feel a need to express my personality.
I feel a need to express my self-identity.
Maintain self- I have a need that who I am today also incorporates
identity my past.
I have a need that my past be an important part of my self-
identity.
I feel a need that who I am today does not ignore my past.
(continued)
Understanding the Role of Consumer Psychological Motives in. . . 71

Higher-order First-order
construct construct Items
Efﬁcacy and effectance I have the skills needed to use SCO.
I am motivated to use new technologies.
I can effectively operate the functionalities of SCO.
I am conﬁdent about how to use the tools of SCO.
I can use the SCO.
I know what the features of this technology look.
I am curious about exploring new technologies.
Territoriality When using this SCO, I feel like having a place
to inhabit.
of my own territory.
of home.
of comfort.
of security.
of inner peace.

References

Ardrey R (1966) The territorial imperative. Atheneum, New York

Belk R (1988) Possessions and the extended self. J Consum Res 15(2):139–168
Belk RW (2013) Extended self in a digital world. J Consum Res 40(3):477–500
Belk R (2017) The soul and the machine: humanlike machines and machinelike humans. In:
Gneezy A, Griskevicius V, Duluth WP (eds) Advances in consumer research, vol 45. Associ-
ation for Consumer Research, MN, pp 164–169
Bellini CGP, Isoni Filho MM, de Moura Junior PJ, de Faria Pereira RDC (2016) Self-efﬁcacy and
anxiety of digital natives in face of compulsory computer-mediated tasks: a study about digital
capabilities and limitations. Comput Hum Behav 59:49–57
Benamar L, Balagué C, Zhong Z (2020) Internet of things devices appropriation process: the
Dynamic Interactions Value Appropriation (DIVA) framework. Technovation 89:102082
Carmines EG, Zeller RA (1979) Reliability and validity assessment. Sage, Thousand Oaks
Carroll J, Howard S, Peck J, Murphy J (2003) From adoption to use: the process of appropriating a
mobile phone. Australas J Inf Syst 10(2):38–48
Chin WW (1998) The partial least squares approach to structural equation modeling. In:
Marcoulides GA (ed) Modern methods for business research. Lawrence Erlbaum Associates
Publishers, NJ, pp 295–336
Chuah SHW, Rauschnabel PA, Krey N, Nguyen B, Ramayah T, Lade S (2016) Wearable technol-
ogies: the role of usefulness and visibility in smartwatch adoption. Comput Hum Behav 65:276–
284
Crocetta C, Antonucci L, Cataldo R, Galasso R, Grassia MG, Lauro CN, Marino M (2021) Higher-
order PLS-PM approach for different types of constructs. Soc Indic Res 154(2):725–754
Dijkstra TK, Henseler J (2015) Consistent partial least squares path modeling. MIS Q 39(2):
297–316
Dittmar H (1992) The social psychology of material possessions: to have is to be. St. Martin’s Press,
New York
Fornell C, Larcker DF (1981) Structural equation models with unobservable variables and mea-
surement error: algebra and statistics. J Mark Res 18(3):382–388
Franke G, Sarstedt M (2019) Heuristics versus statistics in discriminant validity testing: a compar-
ison of four procedures. Internet Res 29(3):430–447
72 Z. Zhong and C. Balagué

Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Market Theory Pract
19(2):139–152
Hair JF, Hollingsworth CL, Randolph AB, Chong AYL (2017) An updated and expanded assess-
ment of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hair JF, Sarstedt M, Ringle CM (2019a) Rethinking some of the rethinking of partial least squares.
Eur J Mark 53(4):566–584
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019b) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling, 3rd edn. Sage, Thousand Oaks
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Hoffman DL, Novak TP (2018) The path of emergent experience in the consumer IoT: from early
adoption to radical changes in consumers’ lives. GfK Market Intell Rev 10(2):11–18
Huarng KH, Yu THK, Lee CF (2022) Adoption model of healthcare wearable devices. Technol
Forecast Soc Chang 174(10):121286
Hubert M, Blut M, Brock C, Zhang RW, Koch V, Riedl R (2019) The inﬂuence of acceptance and
adoption drivers on smart home usage. Eur J Mark 53(6):1073–1098
Karahanna E, Xu SX, Zhang N (2015) Psychological ownership motivation and use of social media.
J Mark Theory Pract 23(2):185–207
Kirk CP, Swain SD, Gaskin JE (2015) I’m proud of it: consumer technology appropriation and
psychological ownership. J Mark Theory Pract 23(2):166–184
Law KS, Wong C-S (1999) Multidimensional constructs M structural equation analysis: an
illustration using the job perception and job satisfaction constructs. J Manag 25(2):143–160
Lu D, Lai IKW, Liu Y (2019) The consumer acceptance of smart product-service systems in sharing
economy: the effects of perceived interactivity and particularity. Sustainability 11(3):928
MacKenzie SB, Podsakoff PM, Jarvis CB (2005) The problem of measurement model
misspeciﬁcation in behavioral and organizational research and some recommended solutions.
J Appl Psychol 90(4):710–730
Mifsud M (2016) Appropriating service to co-create value: an empirical study in the healthcare
sector. Doctoral Dissertation, University of Montpellier
Mifsud M, Cases AS, N’Goala G (2015) Service appropriation: how do customers make the service
their own? J Serv Manag 26(5):706–725
Nunnally JC (1978) Psychometric theory. McGraw-Hill, New York
Pierce JL, Jussila I (2011) Psychological ownership and the organizational context: theory, research
evidence, and application. Edward Elgar Publishing
Pierce JL, Kostova T, Dirks KT (2003) The state of psychological ownership: integrating and
extending a century of research. Rev Gen Psychol 7(1):84–107
Rademaker ME, Schuberth F, Dijkstra TK (2019) Measurement error correlation within blocks of
indicators in consistent partial least squares: issues and remedies. Internet Res 29(3):448–463
Reed A, Forehand MR, Puntoni S, Warlop L (2012) Identity-based consumer behavior. Int J Res
Mark 29(4):310–321
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. Boenningstedt: SmartPLS.. https://www.
smartpls.com Accessed 2 Jan 2021
Sarstedt M, Hair JF, Ringle CM, Thiele KO, Gudergan SP (2016) Estimation issues with PLS and
CBSEM: where the bias lies! J Bus Res 69(10):3998–4010
Sarstedt M, Hair JF, Cheah J-H, Becker J-M, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J 27(3):197–211
Sarstedt M, Ringle CM, Hair JF (2021) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg AE (eds) Handbook of market research. Springer, Cham
Sartre JP (1943) Being and nothingness: an essay on phenomenological ontology, vol 14.
Gallimard, Paris
Understanding the Role of Consumer Psychological Motives in. . . 73

Schamberger T, Schuberth F, Henseler J, Dijkstra TK (2020) Robust partial least squares path
modeling. Behaviormetrika 47(1):307–334
Shmueli G, Sarstedt M, Hair JF, Cheah JH, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Statista (2016) https://www.statista.com/statistics/471264/iot-number-of-connected-devices-world
wide/ Accessed 13 May 2021
Van Dyne L, Pierce JL (2004) Psychological ownership and feelings of possession: three ﬁeld
studies predicting employee attitudes and organizational citizenship behavior stable. J Organ
Behav 25(4):439–459
Wang CL (2021) New frontiers and future directions in interactive marketing: inaugural editorial. J
Res Interact Mark 15(1):1–9
White RW (1959) Motivation reconsidered: The concept of competence. Psychol Rev 66(5):
297–333
Zhong Z (2019) Understanding the appropriation of consumer connected objects: a hierarchical
component modeling approach. Doctoral Dissertation. Paris Saclay University
Zhong Z, Balagué C (2019) Understanding Consumer Internet of Things appropriation: a hierar-
chical component modelling approach. In: 48th Annual Conference of European Marketing
Academy, 27–31 May, Hamburg, Germany
Zou H, Fang Y, Sun H, Lim KH (2017) Engaging in technology extra-role behavior in a human-IT
relationship: a psychological ownership perspective. In: 38th International Conference on
Information Systems ICIS, 10–13 December, Seoul, South Korea, Curran Associates, Inc.
Advertisements that Follow Users Online
and Their Effect on Consumers’
Satisfaction and Expectation Conﬁrmation:
Evidence from the Tourism Industry

Jordi López-Sintas, Giuseppe Lamberti, and Haitham Alghanayem

1 Introduction

The Internet and social media have changed consumer behavior and how tourism
companies conduct their business (Dwivedi et al. 2021). The massive potential
audience available, spending many hours a day using social media across various
platforms, pushed marketers to embrace social media as a marketing channel and
develop new marketing strategies (Appel et al. 2020).
Retargeting has emerged as an online advertising strategy (Varnali 2019).
According to Berman (2018), retargeting advertisements allow e-tourism companies
to create highly personalized travel plans and offer competitive deals based on their
previous browsing history. However, this kind of strategy is very sensitive as it may
affect how consumers view their travel-related purchases and thus their post-
purchase satisfaction. For example, better-placed deals than the one consumer
purchased can negatively affect their post-purchase satisfaction. Further, its effect
can be different on diverse consumer groups deﬁned by sociodemographic variables
such as gender, age, education, or the amount of time the consumer spends online
daily.
Research has already analyzed the retargeting effect on consumer behavior.
However, most of the literature on retargeting has been focused on pre-purchase
topics such as ethical and regulatory concerns (Aalberts et al. 2016; De Lima and

J. López-Sintas · G. Lamberti
Department of Business, School of Economics and Business, Universitat Autònoma de
Barcelona, Barcelona, Spain
e-mail: [email protected]; [email protected]
H. Alghanayem (✉)
Department of Business Administration, College of Science and Arts in Sajir, Shaqra
University, Shaqra, Saudi Arabia

© Springer Nature Switzerland AG 2023 75

Legge 2014; Leenes and Kosta 2015), consumer perceptions of retargeting (Smit
et al. 2014; Ur et al. 2012), and best practices for campaign effectiveness in terms of
timing, delivery, and repetition (Bleier and Eisenbeiss 2015; Lambrecht and Tucker
2013; Li et al. 2021; Sahni et al. 2019). Limited research has been conducted on the
impact of post-purchase retargeting on consumer behaviour, and there is a lack of
evidence regarding potential variations in this effect across diverse consumer groups.
Thus, it remains crucial that the role of retargeting is studied and viewed beyond
website visits in response to post-purchase satisfaction levels among e-tourism
industry consumers and their expectation confirmation.
Using the expectation confirmation theory (Bhattacherjee 2001) as a framework
in this chapter, we relate three main constructs: expectation confirmation, satisfac-
tion, and repurchase intention, with retargeting intended as a measurement of how
competing the post-purchase retargeted ads consumers experienced compared to
their purchase. The retargeting construct was developed considering the model of
consumer’s buying decision (Dubrovski 2001) and including four indicators (price,
features, value, and company reputation).
In this chapter we assume that the effect of more competitive retargeting adver-
tisement acts as a violation of consumers original expectations, thus affecting their
levels of expectations confirmation according to violated expectation model
(Pinquart et al. 2021). Also, it is assumed to have effects toward satisfaction as
suggested in previous retargeting literature (Baek and Morimoto 2012; Lambrecht
and Tucker 2013; Li et al. 2021).
This chapter contributes to and extends existing research on online advertising
and retargeting in the tourism sector by (1) analyzing the effect of retargeting on the
causal chain expectations-satisfaction-repurchase intention of online consumers and
(2) exploring the existence of different consumer groups identified by gender, age,
education, and daily time spent on social media where the effect of retargeting on
expectations and satisfaction is different.

2 Methodology

E-tourism online buyers in Saudi Arabia with an active account on social media
platforms, who purchased tourism services online and experienced post-purchase
retargeted ads, were sampled, surveyed, and analyzed using measurements of the
four constructs (see Table 1). A total of 402 samples was achieved, and the
demographic proﬁle of respondents for this chapter showed that consumers were
45.52% female and 54.58% male. The most frequent age was 25–34 years (37.8%).
Furthermore, most respondents hold a bachelor’s degree (51.99%) and used social
media between 1 and 5 hours daily (65.67%).
Partial least squares structural equation modeling (PLS-SEM) was used to ﬁt our
model (Hair et al. 2017). Several scholars indicate that the PLS-SEM method is the
most adequate when conceptual variables are estimated as composites (i.e., linear
combination of indicators) (Cepeda-Carrion et al. 2019). The hybrid multigroup
Advertisements that Follow Users Online and Their Effect on Consumers’. . . 77

Table 1 Deﬁnitions and sources of constructs

Construct Definition Source
Expectations Degree to which perceptions match Bhattacherjee (2001), Oliver (1980)
confirmation (confirmation) or differ from (disconfir-
mation) expectations
Satisfaction A measurement that determines how Del Bosque and San Martín (2008),
happy customers are with their purchase Bhattacherjee (2001), Oliver (1980)
Repurchase Degree of likelihood of making another Bhattacherjee (2001), Hellier et al.
intentions purchase with the e-tourism website or (2003)
application
Retargeting A measurement of how competing the New scale developed and drawn from
post-purchase retargeted adverts con- consumer buying-decision model
sumers experienced (Dubrovski 2001)

approach (Lamberti 2021) was used to explore gender, age, education, and daily
time spent on social media as possible sources of heterogeneity in consumer
behaviors. This approach combines classical multigroup analysis (MGA) with
Pathmox analysis (Lamberti et al. 2016). The most significant different groups are
first identified using Pathmox. The obtained groups are then compared using the
PLS-MGA test (Henseler et al. 2009) after assuring measurement invariance by
applying the MICOM procedure (Henseler et al. 2016). The statistical analysis was
performed using SmartPLS 3.3 (Ringle et al. 2015) and the R software genpathmox
package V.0.5 (Lamberti 2014).

3 Results

After validating the measurement model following Hair et al. (2019) (see Tables 2, 3,
and 4), our results showed (see Fig. 1 and Table 5) a significant relationship between
retargeting on expectation (β = -0.324), while surprising, retargeting on satisfaction
was low and not significant (β = -0.032). As was expected, expectation positively
affects satisfaction (β = 0.655), and satisfaction increases consumers’ intention to
repurchase (β = 0.670), corroborating previous findings (Alghamdi et al. 2018;
Rajeh et al. 2021; Zhong et al. 2015). In addition, the model’s predictive power (see
Table 6) was moderate-high according to the PLSpredict procedure (Shmueli et al.
2019).
However, the effect of retargeting was not homogenous. Instead, it varied pri-
marily according to the time spent online and the age of the consumers. The hybrid
multigroup analysis revealed the presence of three different groups of consumers
labeled, respectively, as heavy social media users (LM1), younger-moderate social
users (LM2), and older-moderate social users (LM3) that were characterized by
differences in how retargeting affect expectation and satisfaction (see Table 7). After
assuring invariance of constructs between groups (see Table 8) following the
78 J. López-Sintas et al.

Table 2 Reliability and validity criteria

Factor Cronbach’s
Constructs/items loadings 2.50% 97.50% alpha CR AVE
Expectation conﬁrmation 0.845 0.906 0.763
(EC)
EC1 0.870 0.827 0.902
EC2 0.885 0.846 0.915
EC3 0.865 0.817 0.902
Satisfaction (ST) 0.720 0.876 0.78
ST1 0.872 0.839 0.901
ST2 0.887 0.847 0.919
ST3 0.889 0.846 0.921
ST4 0.848 0.794 0.894
Repurchase intentions (RI) 0.866 0.909 0.713
RI1 0.904 0.874 0.926
RI2 0.862 0.813 0.898
Retargeting (RE) 0.897 0.929 0.765
RE1 0.814 0.747 0.864
RE2 0.879 0.844 0.908
RE3 0.867 0.824 0.899
RE4 0.816 0.764 0.860

Table 3 Heterotrait-monotrait (HTMT)

Value 2.50% 97.50%
Repurchase intentions → expectation confirmation 0.719 0.602 0.830
Retargeting → expectation confirmation 0.375 0.223 0.518
Retargeting → repurchase intentions 0.206 0.073 0.369
Satisfaction → expectation confirmation 0.763 0.680 0.832
Satisfaction → repurchase intentions 0.828 0.741 0.904
Satisfaction → retargeting 0.271 0.137 0.403

Table 4 Assessment of common method bias

Expectation conﬁrmation Repurchase intentions Retargeting Satisfaction
VIF 1.962 1.908 1.108 2.352

MICOM procedure (Henseler et al. 2016), we compared the three consumer groups
using the MGA-PLS test (Henseler et al. 2009).
We found that the effect of retargeting on expectations confirmation and satis-
faction was significant when comparing the LM1 and LM3. Furthermore, it was
significant and higher for the LM1 compared with LM3. This finding can be
explained by the fact that heaver social media users who spend longer periods of
time are more likely to be retargeted more frequently. We also found that the effect
on expectation on satisfaction was significant and higher for LM2 than LM1.
Advertisements that Follow Users Online and Their Effect on Consumers’. . . 79

Fig. 1 Path diagram results

80 J. López-Sintas et al.

Table 5 Path coefﬁcients, bootstrap IC R2, Q2, and SRMR

H Path coefficients β 2.50% 97.50% p-value Sig.
H1 Expectation confirmation on satisfaction 0.655 0.587 0.737 0.000 Yes
H2 Satisfaction on repurchase intentions 0.670 0.594 0.738 0.000 Yes
H3 Retargeting on expectation confirmation -0.324 -0.444 -0.196 0.000 Yes
H4 Retargeting on satisfaction -0.032 -0.126 0.053 0.489 No
Repurchase intentions R2 = 0.449; Q2 = 0.345
Satisfaction R2 = 0.443; Q2 = 0.335
SRMR = 0.054

Table 6 PLSpredict procedure

Constructs PLS RMSE Q2predict LM RMSE Difference
RI1 1.280 0.017 1.287 -0.007
RI2 1.313 0.018 1.324 -0.011
ST1 1.133 0.044 1.135 -0.002
ST2 1.121 0.039 1.129 -0.007
ST3 1.179 0.045 1.183 -0.004
ST4 1.222 0.027 1.217 0.005

Table 7 Multigroup comparison results

PLS-MGA test
LM1 vs LM1 vs LM2 vs
LM2 LM3 LM3
Paths LM1 LM2 LM3 p-value
Retargeting on expecta- -0.568*** -0.392*** -0.117 NS 0.203 0.001 0.091
tion confirmation
Retargeting on -0.217*** -0.750NS 0.090 NS 0.348 0.037 0.096
satisfaction
Expectation confirmation 0.440*** 0.763*** 0.609*** 0.018 0.214 0.185
on satisfaction
Satisfaction on 0.697*** 0.662*** 0.655*** 0.767 0.588 0.888
repurchase intentions
Notes: ***p < 0.001, NS non-significant. Italicized means significant <0.05.

4 Discussion

In this chapter, we provide evidence on the effect of retargeting on the causality

chain expectation-satisfaction-repurchase intention. Our results indicate that
retargeting has a strong effect on expectation, while, regarding satisfaction, its effect
is low and not significant. However, the retargeting effect can vary according to
the characteristic of the consumer. In particular, we found that it varies according to
the time spent online and the age of the consumer, supporting some findings in the
literature (Carstensen et al. 2011; Zhang et al. 2017) and indicating that marketers
have to pay attention to these two factors when implementing this online advertising
Table 8 MICOM procedure (steps 2 and 3)
Equality of composite values
Compositional invariance Mean values Variance values
Construct Groups Score corr 5.00% Mean Diff 2.5–97.5% Variance Diff 2.5–97.5%
Expectation confirmation LM1 vs LM2 0.999* 0.997 0.132 [-0.314–0.313]* -0.342 [-0.721–0.743]*
Repurchase intentions 0.997* 0.985 0.277 [-0.302–0.310]* -0.528 [-0.538–0.518]*
Retargeting 0.988* 0.988 -0.292 [-0.311–0.315]* 0.089 [-0.611–0.633]*
Satisfaction 0.999* 0.999 -0.037 [-0.307–0.316]* -0.494 [-0.653–0.636]*
Expectation confirmation LM1 vs LM3 1.000* 0.996 0.699 [-0.252–0.242] -0.063 [-0.452–0.438]*
Repurchase intentions 1.000* 0.997 0.498 [-0.254–0.243] -0.211 [-0.380–0.363]*
Retargeting 0.957* 0.954 -0.875 [-0.254–0.251] -0.050 [-0.448–0.393]*
Satisfaction 1.000* 0.998 0.590 [-0.253–0.248] -0.188 [-0.396–0.381]*
Expectation confirmation LM2 vs LM3 1.000* 0.997 0.519 [-0.255–0.247] 0.279 [-0.546–0.475]*
Repurchase intentions 0.998* 0.992 0.248 [-0.253–0.256]* 0.328 [-0.434–0.384]*
Retargeting 0.960* 0.793 -0.642 [-0.266–0.260] -0.129 [-0.480–0.407]*
Satisfaction 0.999* 0.988 0.588 [-0.255–0.255] 0.315 [-0.416–0.383]*
Advertisements that Follow Users Online and Their Effect on Consumers’. . .
81
82 J. López-Sintas et al.

strategy. However, this chapter is not without limitations. The main limitation of this
chapter was that a non-probability sampling approach was used; therefore, the results
cannot be generalized to the larger population. Also, findings are specific to
e-tourism users in Saudi Arabia and needs further validation by comparing current
findings with findings of consumers from other countries and markets. Furthermore,
this chapter did not consider examining the impact of consumer involvement on the
effects of retargeting, which future research should include to examine its impacts.

References

Aalberts RJ, Nill A, Poon PS (2016) Online behavioral targeting: what does the law say? J Curr
Issues Res Advert 37(2):95–112
Alghamdi A, Elbeltagi I, Elsetouhi A, Haddoud MY (2018) Antecedents of continuance intention of
using Internet banking in Saudi Arabia: a new integrated model. Strateg Chang 27(3):231–243
Appel G, Grewal L, Hadi R, Stephen AT (2020) The future of social media in marketing. J Acad
Mark Sci 48(1):79–95
Baek TH, Morimoto M (2012) Stay away from me. J Advert 41(1):59–76
Berman R (2018) Beyond the last touch: attribution in online advertising. Mark Sci 37(5):771–792
Bhattacherjee A (2001) Understanding information systems continuance: an expectation-
conﬁrmation model. MIS Q 25(3):351–370
Bleier A, Eisenbeiss M (2015) Personalized online advertising effectiveness: the interplay of what,
when, and where. Mark Sci 34(5):669–688
Carstensen LL, Turan B, Scheibe S, Ram N, Ersner-Hershﬁeld H, Samanez-Larkin GR et al (2011)
Emotional experience improves with age: evidence based on over 10 years of experience
sampling. Psychol Aging 26(1):21–33
Cepeda-Carrion G, Cegarra-Navarro JG, Cillo V (2019) Tips to use partial least squares structural
equation modelling (PLS-SEM) in knowledge management. J Knowl Manag 23(1):67–89
De Lima D, Legge A (2014) The European Union’s approach to online behavioural advertising:
protecting individuals or restricting business? Comput Law Secur Rev 30(1):67–74
Del Bosque IR, San Martín H (2008) Tourist satisfaction a cognitive-affective model. Ann Tour Res
35(2):551–573
Dubrovski D (2001) The role of customer satisfaction in achieving business excellence. Total Qual
Manag 12(7):920–925
Dwivedi YK, Ismagilova E, Hughes LD, Carlson J, Filieri R, Jacobson J et al (2021) Setting the
future of digital and social media marketing research: perspectives and research propositions. Int
J Inf Manag 59:102168
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2017) A Primer on Partial Least Squares Structural
Equation Modeling (PLS-SEM), 2nd edn. Sage, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hellier PK, Geursen GM, Carr RA, Rickard JA (2003) Customer repurchase intention: a general
structural equation model. Eur J Mark 37(11/12):1762–1800
Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. In: Sinkovics RR, Ghauri PN (eds) Advances in international market-
ing, vol 20. Emerald Publishing Limited, Bingley, pp 277–319
Henseler J, Ringle CM, Sarstedt M (2016) Testing measurement invariance of composites using
partial least squares. Int Mark Rev 33(3):405–431
Lamberti G (2014) R package genpathmox. http://cran.r-project.org/web/packages/genpathmox
Accessed 28 Apr 2021
Advertisements that Follow Users Online and Their Effect on Consumers’. . . 83

Lamberti G (2021) Hybrid multigroup partial least squares structural equation modelling: an
application to bank employee satisfaction and loyalty. Qual Quant:1–23
Lamberti G, Aluja TB, Sanchez G (2016) The Pathmox approach for PLS path modeling segmen-
tation. Appl Stoch Model Bus Ind 32(4):453–468
Lambrecht A, Tucker C (2013) When does retargeting work? Information speciﬁcity in online
advertising. J Mark Res 50(5):561–576
Leenes R, Kosta E (2015) Taming the Cookie Monster with Dutch Law–a tale of regulatory failure.
Comput Law Secur Rev 31(3):317–335
Li J, Luo X, Lu X, Moriguchi T (2021) The double-edged effects of e-commerce cart retargeting:
does retargeting too early backﬁre? J Mark 85(4):123–140
Oliver RL (1980) A cognitive model of the antecedents and consequences of Satisfaction Decisions.
J Mark Res 17(4):460
Pinquart M, Endres D, Teige-Mocigemba S, Panitz C, Schütz AC (2021) Why expectations do or do
not change after expectation violation: a comparison of seven models. Conscious Cogn 89:
103086
Rajeh MT, Abduljabbar FH, Alqahtani SM, Waly FJ, Alnaami I, Aljurayyan A et al (2021)
Students’ satisfaction and continued intention toward e-learning: a theory-based study. Med
Educ Online 26(1)
Ringle CM, Da Silva D, Bido D (2015) Structural equation modeling with the SmartPLS. Braz J
Market 13(2):1149–1160
Sahni SN, Narayanan S, Kalyanam K (2019) An experimental investigation of the effects of
retargeted advertising: the role of frequency and timing. J Mark Res 56(3):401–418
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S et al (2019) Predictive model
assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Smit EG, Van Noort G, Voorveld HA (2014) Understanding online behavioural advertising: user
knowledge, privacy concerns and online coping behaviour in Europe. Comput Hum Behav 32:
15–22
Ur B, Leon PG, Cranor LF, Shay R, Wang Y (2012) Smart, useful, scary, creepy: perceptions of
online behavioral advertising. In: Proceedings of the eighth symposium on usable privacy and
security, Washington, USA, July 11–13. Association for Computing Machinery, New York
Varnali K (2019) Online behavioral advertising: an integrative review. J Mark Commun 27(1):
93–114
Zhang Y, Trusov M, Stephen AT, Jamal Z (2017) Online shopping and social media: friends
or foes? J Mark 81(6):24–41
Zhong Z, Luo J, Zhang M (2015) Understanding antecedents of continuance intention in mobile
travel booking service. Int J Bus Manag 10(9):156–162
Similarities in Factors Affecting Online
Shopping Intention in Ecuador and Peru:
A Multigroup Analysis

Vasilica-Maria Margalina and Alberto Magno Cutipa Limache

1 Introduction

E-commerce penetration during the pandemic has reached very high levels due to the
impossibility to shop in many physical stores. This is also the case of Latin America,
which is one of the world’s regions with the lowest e-commerce penetration rates,
but that during the pandemic has reached levels not expected to achieve for another
ten years (Boyle 2021). The growth of e-commerce in the region was 37% in 2020
(Oxford Business Group 2021); however, the evolution has not been the same in all
countries. Brazil and Mexico are the leaders of the region, but economies such as
Argentina, Peru, and Colombia have registered a swift growth (Chevalier 2021).
Nevertheless, Latin America still lags behind e-commerce penetration when com-
pared to other regions.
Among multiple factors, shopping intention is considered to be the main barrier
for e-commerce development (Lim et al. 2016). Shopping intention is deﬁned as the
likelihood that a consumer will purchase a product or a service in the near future
(Imtiaz et al. 2019). Regarding the online shopping intention, it refers to the
likelihood that the future purchase of products will be done through an online
channel. According to Salisbury et al. (2001), the online shopping intention repre-
sents a speciﬁed shopping behavior via the Internet.

V.-M. Margalina (✉)

CESINE Centro Universitario, Santander, Cantabria, Spain
e-mail: [email protected]
A. M. Cutipa Limache
Escuela Profesional en Administración, Universidad Nacional del Altiplano de Puno, Puno,
Peru
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 85

Despite the problems of the region in reaching other regions in e-commerce

penetration, there is few research on online consumer behavior in Latin America,
and most of them are focused on countries such as Brazil and Chile (Margalina
2022). Hence, it is important to analyze the factors that affect online shopping
intention in the region in order to help businesses to increase their presence in the
online shopping channels. Previous studies have frequently analyzed factors, such
as usefulness, website characteristics and trust, as antecedents of consumer’s online
shopping intention in both developed and developing countries (Albashrawi 2021).
However, research in developing economies has been giving little attention to
demographic factors, product characteristics, risk factors, appearance, and trust in
institutional mechanisms. The last factor is important for Latin-American countries,
as the majority of their citizens distrust their political, social, and economic institu-
tions (Bergman and Rosenkratz 2019).
Thus, the objective of this research is to analyze the effect that some of the most
important factors, such as shopping channel attributes and perceived risks, have on
online shopping intention in two middle e-commerce markets of the region: Ecuador
and Peru. These two countries have been chosen for their similarities, as both have
Spanish as official language, similar formal institutions and low level of interper-
sonal and institutional trust among their citizens (Latinobarometer 2021). For this
reason, the study is focused on the identification of similarities between Latin-
American countries regarding the antecedents of consumer’s online shopping inten-
tion. Additionally, the purpose of the research is to fill the gap on the study of trust
factors in developing economies by adding institutional trust as an antecedent of
perceived risks in the proposed model.

2 Factors Affecting Online Shopping Intention

Consumer’s intention is considered to be an important predictor of consumer

behavior in the theory of the reasoned action (Azjen 1991). Nevertheless, consumer
behavior depends on different inﬂuencing factors, making measurement and analysis
of intention complex (Iqbal 2019).
The attributes of the online shopping channel, such as cost, time, convenience,
and customer service, are among the most important determinants of consumers
decision to shop online (Cunningham and De Meyer-Heydenrych 2018). The anal-
ysis of the relationship between the consumers’ perception on costs and the election
of the shopping channel is very important (Chintagunta et al. 2012). Costs are not
referring only to the price paid for the product but also to travel costs incurred when
physically purchasing the product, as well as the delivery costs. Time represents the
number of hours that consumers spend searching for a product or purchasing
it. Convenience is the most important factor when choosing the shopping channel
(Lodorfos et al. 2006) and represents the ease of use of the shopping channel
(Cunningham and De Meyer-Heydenrych 2018). Consumers search for shopping
channels that help them to save as much time as possible when purchasing products
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . . 87

(Lodorfos et al. 2006). Therefore, the online shopping channel must offered them the
easiest and quickest way to find the product and the information related to it, which
represent the main dimensions of convenience. Customer service represents the
service offered by the shopping channels to the customers and includes the interac-
tion with the staff, the returning items, and after-sales service. Therefore, the first
hypothesis is as follows:
H1: Online shopping channel attributes influence positively online shopping
intention.
Nevertheless, one of the most analyzed factors in relation with online shopping
intention is perceived risks (Cunningham and De Meyer-Heydenrych 2018; Yu et al.
2018; Iqbal 2019; Qalati et al. 2021). The perceived risks that consumers can
experience when shopping online are delivery risk, product risk, and security risk
(Cunningham and De Meyer-Heydenrych 2018). Iqbal (2019) and Qalati et al.
(2021) found a significantly positive relationship between perceived risk and online
shopping intention, while Cunningham and De Meyer-Heydenrych (2018) showed
that only delivery risks affect the consumers intention to shop online. In this case, the
second hypothesis is as follows:
H2: Perceived risks have a positive effect on online shopping intention.
Trust is one of the most important factors of the online shopping (Thomas et al.
2018). Previous studies have shown that there is a relationship between perceived
risks and trust (Zhu et al. 2011; Hong and Cha 2013). On the one hand, Zhu et al.
(2011) found that trust reduces online consumer perceived risk when shopping
online; on the other hand, Hong and Cha (2013) showed that perceived risks can
affect trust. Usually, when analyzing trust, studies focus on the trust on the online
vendor. However, in this research, it is tested the relationship between institutional
trust, defined as the trust in public institutions (Latinobarometer 2021), and per-
ceived risks when shopping online:
H3: Institutional trust has a negative effect on perceived risks of online shopping.
Figure 1 presents the variables and hypotheses included in the proposed concep-
tual model.

3 Study Context and Sample

To explore the similarities on the inﬂuence of online shopping attributes and risks on
online shopping intention, data from two Latin-American countries was collected,
Ecuador and Peru, between April and July 2022. The similar characteristics in terms
of language, formal institutions (rules, law, and regulations) and interpersonal and
institutional trust, represented the selection criteria of the two countries. Student
samples for both countries were chosen because they are more homogenous than
nonstudent sample, reducing the effect of factors other than national context, such as
88 V.-M. Margalina and A. M. Cutipa Limache

Online shipping channel attributes

Costs

Time
H1

Online shopping
H1
Convenience intention

H2
Product risks
Perceived risks
Delivery risks

Security risks H3

Institutional trust

Fig. 1 The proposed conceptual model

Table 1 Sample characteristics

Characteristics Ecuador Peru
Age (mean) 22.10 20.95
Gender (percentage of females) 74% 44%
Experience with online shopping (percentage of students that have shopped 86.1% 83.2%
online)
Interpersonal trust (percentage of student that trust the others) 5.8% 8.2%

socio-economic status, education, work experience, and age (Schlägel and Sarstedt
2016). A total of 223 students from Ecuador and 268 students from Peru were
surveyed.
In both countries, the survey was administrated in an online form to Bachelor
students by a professor. All questionnaires were completed anonymously to ensure
conﬁdentiality during the same period in both Ecuador and Peru. The characteristics
of the sample are presented in Table 1. As it can be observed, there is a higher
percentage of females (74%) in the Ecuador’s sample compared to Peru (44%),
while in the case of the other characteristics, there are only small differences.
Shopping channel attributes (cost, convenience, customer service, and time) were
measured using the scale proposed by Cunningham and De Meyer-Heydenrych
(2018). The items used for measuring online shopping intention were adapted
from Ling et al. (2010), while for institutional trust, the items and scale used by
the Latinobarometer (2020) were adapted. For all items a ﬁve-point Likert scale was
used. The questions referred to the online shopping of fashion products.
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . . 89

4 Analysis

To test the relationships, a PLS-SEM model was estimated by using the SmartPLS
3 software (Ringle et al. 2015). The reliability and validity measures were estimated
by applying the PLS Algorithm procedure, following the recommendations of Hair
et al. (2016) for the algorithm settings: path weighting, a maximum of 300 iterations,
and a stop criterion of 10-7.
The measurement invariance assessment was performed with the help of MICOM
procedure, as recommended by Henseler et al. (2016) and Schlägel and Sarstedt
(2016). The procedure includes three steps: the assessment of configural invariance,
the assessment of compositional invariance, and the assessment of the equality of
means and variances.
Finally, the multigroup analysis was performed by running Chin and Dibbern’s
(2010) permutation test, which allows to test the significance of the differences
between path coefficients.
The explicative relevance of the model was assessed by running the PLS algo-
rithm, while the predictive relevance was estimated with the help of the blindfolding
procedure as recommended by Hair et al. (2019). The predictive relevance was
evaluated by applying the PLSpredict procedure, following Sarstedt et al.’s (2021)
recommendations.

4.1 Measurement Model Assessment

Following the guidelines of Hair et al. (2019) and Sarstedt et al. (2021), the measures
for internal consistency reliability, convergent validity, and discriminant validity
were assessed (Tables 2, 3, and 4).
As some of the reﬂective indicator loadings of the constructs costs, time, and
institutional trust reached values below the 0.708 for both samples, they were
deleted. The analysis proceeding the deletion showed that the reliability all of
items was satisfactory for both samples. In the case of the variable convenience,
the items Conv2 and Conv3 were deleted because composite’s reliability was above
0.95 threshold value. Sarstedt et al. (2021) state that values higher than 0.95 are
problematic, suggesting that items are almost redundant. After the elimination,
composite reliability (rho_C) decreased from 0.966 to 0.945 in the case of the
Ecuadorian sample, and from 0.957 to 0.935 in the case of the Peruvian sample.
The values of Cronbach’s alpha, composite reliability, and AVE showed internal
consistence reliability and convergent validity for all constructs (Tables 2 and 3).
It must be mentioned that, in the case of the second order construct Risks, some
items showed values below the threshold 0.708 (Table 3). As the deletion of this
items lowers the internal consistency reliability, especially for the Ecuadorian
sample, and the objective is to compare the two sample, it was decided to retain
the items.
Table 2 Reliability and validity of the measurement model
90

Loadings Cronbach’s alpha rho_A rho_C AVE

Constructs Items Ecuador Peru Ecuador Peru Ecuador Peru Ecuador Peru Ecuador Peru
Costs Cost1 0.871 0.826 0.904 0.889 0.906 0.896 0.923 0.918 0.722 0.692
Cost2 0.846 0.842
Cost4 0.855 0.875
Cost5 0.827 0.807
Cost6 0.850 0.806
Time Time1 0.932 0.955 0.863 0.899 0.868 0.890 0.936 0.952 0.879 0.908
Time2 0.943 0.951
Convenience Conv1 0.904 0.909 0.916 0.896 0.912 0.897 0.945 0.935 0.856 0.828
Conv4 0.939 0.918
Conv5 0.933 0.904
Customer service Serv1 0.785 0.737 0.925 0.903 0.936 0.921 0.941 0.925 0.726 0.674
Serv2 0.882 0.830
Serv3 0.897 0.874
Serv4 0.826 0.807
Serv5 0.834 0.786
Serv6 0.883 0.886
Risks Delivery 0.882 0.798 0.862 0.819 0.864 0.826 0.916 0.892 0.784 0.735
Product 0.905 0.883
Security 0.865 0.884
Institutional trust TrustI5 0.891 0.877 0.903 0.879 0.945 0.921 0.926 0.909 0.716 0.669
TrustI6 0.799 0.831
TrustI7 0.803 0.738
TrustI8 0.864 0.843
TurstI9 0.870 0.792
Intention Int1 0.786 0.854 0.865 0.899 0.893 0.904 0.918 0.937 0.790 0.833
V.-M. Margalina and A. M. Cutipa Limache

Int2 0.940 0.939

Int3 0.933 0.942
Table 3 Reliability and validity of the ﬁrst-order constructs that measure Risks
Loadings Cronbach’s alpha rho_A rho_C AVE
Constructs Items Ecuador Peru Ecuador Peru Ecuador Peru Ecuador Peru Ecuador Peru
Delivery risk Risk1 0.827 0.734 0.844 0.732 0.849 0.757 0.896 0.830 0.682 0.551
Risk2 0.824 0.797
Risk3 0.863 0.791
Risk4 0.787 0.637
Product risk Risk5 0.895 0.825 0.883 0.783 0.907 0.797 0.919 0.860 0.740 0.608
Risk6 0.854 0.695
Risk7 0.907 0.861
Risk8 0.780 0.726
Security risk Risk9 0.763 0.650 0.837 0.726 0.848 0.775 0.890 0.818 0.670 0.534
Risk10 0.799 0.583
Risk11 0.847 0.815
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . .

Risk12 0.862 0.845

91
92 V.-M. Margalina and A. M. Cutipa Limache

Table 4 Discriminant validity assessment results

Institutional
Convenience Costs trust Intention Risks Service
Costs EC 0.808
PE 0.747
Institutional EC 0.187 0.125
trust
PE 0.208 0.186
Intention EC 0.659 0.592 0.205
PE 0.659 0.594 0.158
Risks EC 0.729 0.727 0.224 0.717
PE 0.706 0.665 0.183 0.738
Service EC 0.752 0.686 0.109 0.596 0.788
PE 0.698 0.561 0.159 0.567 0.722
Time EC 0.888 0.795 0.174 0.612 0.686 0.622
PE 0.830 0.702 0.251 0.537 0.675 0.589

The HTMT criterion used for the assessment of discriminant validity (Henseler
et al. 2015) reached values below the threshold 0.85. There was one exception in the
correlation of the construct time with convenience for the Ecuadorian sample, where
the HTMT value was of 0.888 (Table 4). However, as both constructs represent
attributes of the online shopping channel and the HTMT value is below the threshold
0.90, the discriminant validity criterion is met.

4.2 Invariance Measurement Assessment

The measurement invariance assessment was performed by applying the MICOM

procedure, following the guidelines of Schlägel and Sarstedt (2016). In Step 1, the
configural invariance was assessed. As the PLS path model setup is equal across both
countries and algorithm settings are identical for both group-specific model estima-
tions, configural invariance is established.
In Step 2, the similarity of the composite scores across the group was tested by
running the permutation procedure with 1000 permutations and a 5% significance
level. Table 5 shows that c exceeds the 5% of cu in all cases, and, therefore,
compositional variance is established.
Finally, the full measurement invariance between the samples of Ecuador and
Peru was tested by assessing the equality of the composite mean values and
variances, applying the permutation test recommended by Henseler et al. (2016).
In general, the results show that full measurement invariance is met, as the composite
means and variances are equal across the samples of the two countries, with the
exception of the construct customer service, in which case the mean values are not
equal. Thus, it cannot be analyzed a path model by using a pooled sample of
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . . 93

Table 5 Compositional invariance assessment

Original Correlation permutation Permutation p-
correlation mean 5.0% values
Convenience 1.000 1.000 1.000 0.803
Costs 1.000 0.999 0.998 0.502
Institutional 0.991 0.990 0.972 0.368
trust
Intention 0.999 1.000 0.999 0.133
Risks 1.000 0.999 0.998 0.850
Service 1.000 0.999 0.997 0.891
Time 1.000 1.000 0.999 0.342

Table 6 Results of the structural model

Difference (EC vs. p-value of
Construct Ecuador Peru PE) difference
Costs → Intention 0.029 0.141 -0.112 0.280
Time → Intention 0.086 -0.070 0.156 0.185
Convenience → Intention 0.238* 0.268* -0.031 0.843
Customer service 0.048 0.076 -0.028 0.843
→ Intention
Risks → Intention 0.353* 0.384* -0.028 0.791
Institutional trust → Risks 0.210* 0.162* 0.003 0.657
R2 0.460 0.487
Q2 0.346 0.395
SRMR 0.059 0.055
Notes: *p ≤ 0.05

Ecuadorian and Peruvian respondents. However, since the compositional invariance

is established and the objective of the research is to compare the two countries, the
country-speciﬁc effects can be analyzed.

4.3 Structural Model Assessment

Next, the structural model validity was assessed. As all the VIF values of the
structural model are lower than the threshold 5, it can be concluded that the
collinearity is not an issue. The significance of the path coefficients was established
by running the bootstrapping procedure with 5000 subsamples (Table 6).
The analysis of the explicative relevance of the model was assessed by running
the PLS Algorithm. Both R2 values are superior of 0.45, showing explicative
relevance. These results are supported by Q2 values obtained from the blindfolding
procedure, which are above 0.30 for both samples showing a medium predictive
relevance. In the case of both countries, convenience and risks show a significantly
94 V.-M. Margalina and A. M. Cutipa Limache

Table 7 RMSE values from PLS LM

the PLS-SEM and LM analy-
RMSE RMSE
sis obtained with PLSpredict
Items Ecuador Peru Ecuador Peru
Intention 1 0.980 0.960 1.024 0.960
Intention 2 0.795 0.907 0.791 0.904
Intention 3 0.871 0.901 0.887 0.905

positive effect on intention, while institutional trust affects positively the perception
on risks. Nevertheless, it must be mentioned that the relevance of institutional trust
as an explicative and predictive factor of risks is low, with R2 values below 0.05
(0.044 for Ecuador and 0.028 for Peru) and Q2 values also below 0.05 (0.032 for
Ecuador and 0.016 for Peru). Only convenience ( f2 = 0.027), and risks ( f2 = 0.095)
show effect sizes higher than 0.02 on intention, while institutional trust has an effect
size of 0.046 on risks.
To assess the out-of-sample predictive power, the PLSpredict procedure was
executed, according to the specifications recommended by Shmueli et al. (2019).
In the case of Ecuador, two of the items, Intention 1 and Intention 3, show lower
RMSE values of the PLS-SEM analysis than the LM’s values, indicating a medium
predictive power (Table 7). However, the out-of-sample predictive power of the
model is low for the Peruvian sample, as only Indicator 3 reaches smaller RMSE
values of the PLS-SEM analysis compared to those of LM analysis.
The results of Chin and Dibbern’s (2010) permutation test with 1000 permutation
show that path coefficient estimates are invariant across the two samples, since no
significant difference emerge between Ecuador and Peru.
According to the results, Hypothesis 1 is partially validated, as the only channel
attribute with a significative effect on online shopping intention is convenience.
Perceived risks show not only a significant effect but also contribute the most to the
explanatory and predictive power of the model; therefore, Hypothesis 2 is validated.
Surprisingly, higher levels of institutional trusts have a significant and positive effect
on perceived risks. However, even though Hypothesis 3 was not validated, results
must be interpreted with caution because even though Institutional trust has a
significant effect on perceived risks, the explanatory and predictive relevance is
very low. Though, it must be taken into consideration that the predictive power of the
model is lower for the Peruvian sample. This result together with the MICOM
assessment indicate that the path analysis should be analyzed for each country
separately.

5 Discussion

The assessment of reliability and validity of the measurement model conﬁrms the
validity of scales proposed by Cunningham and De Meyer-Heydenrych (2018) for
measuring shopping channel attributes and perceived risks and the scales proposed
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . . 95

by Ling et al. (2010) for measuring online shopping intention. However, item
validity is higher for the Ecuadorian sample compared to the Peruvian sample. On
the contrary, discriminant validity analysis performs better for the Peruvian sample.
It must be mentioned that, in the case of both samples, discriminant validity issues
arise regarding two constructs of the online channel attributes: convenience and
time. Nevertheless, even though HTMT values for these two constructs are generally
higher for both samples, they reach the maximum threshold in the case of the
Ecuadorian sample.
Convenience was revealed to be the most important channel attribute for online
shopping intention, confirming the results found in the study of Cunningham and De
Meyer-Heydenrych (2018). The results of the analysis performed in this chapter also
confirm that perceived risks represent a significant factor in the study of online
shopping intention, as found in previous studies (Cunningham and De Meyer-
Heydenrych 2018; Iqbal 2019; Qalati et al. 2021). However, in the case of Ecuador
and Peru, higher perceived risks explain and predict higher levels of online shopping
intention. Therefore, it seems that Ecuadorian and Peruvian customer decide to shop
online even when they perceive high risks. Further research should be performed in
order to identify the factor o factors that affect the relationships between these two
constructs.
The same issue was found when analyzing the effect of institutional trust on
perceived risks. Specifically, higher levels of institutional trust explain higher levels
of perceived risks, but the levels of explanatory and predictive power are low. This
could be explained by the different expectations on the competences and duties that
citizens can attribute to the institutions take into account in measuring this type of
trust (government, judicial power, political parties, the electoral institution) that may
or may not include attributions related to e-commerce risks (product, delivery, and
security). Even though the results were not as expected, they highlight the impor-
tance of analyzing the effect of institutional trust on e-commerce transactions.
Nevertheless, the data were collected in a post-pandemic context, characterized by
the economic crisis and the erosion of norms and institutions in Latin American
countries (Marczak and Guevara 2021). This context could explain the results and
the low levels of explanatory and predictive relevance of institutional trust for
perceived risks.
The multigroup analysis shows that there are no significant differences in struc-
tural relationships between the two countries, confirming previous studies that
compared the effect of these factors on online shopping intention among countries,
such as the one performed by Choi and Lee (2003). But other studies have found
differences in the effect of perceived risk (Park and Jun 2003). This result could be
explained by the degree of similarity in language, culture, and institutions between
the analyzed countries. Despite the similarities between the two countries, the
analysis should not be done by using a pooled sample of both countries, as the
items’ relevance and predictive relevance of the path model are different. Addition-
ally, there are also differences in item and discriminant validity. These issues should
be assessed in future research.
96 V.-M. Margalina and A. M. Cutipa Limache

6 Conclusion

In this chapter the similarities between two countries of Latin America on the effect
of online shopping attributes and perceived risks on online shopping intention were
analyzed. The results show that when the countries have cultural, economic, and
social similarities; there are no differences in the factors that affect online shopping
intention. They also reveal that e-commerce businesses from Ecuador and Peru can
increase online shopping intention by improving the convenience of the shopping
channel. It is interesting that online shopping intention increases even in the presence
of high perceived risks. Therefore, further research must identify what factors
moderate the relationship between perceived risks and online shopping intention.
Additionally, even though institutional trust has a significant effect on perceived
risks, it contributes very little to explain and predict this construct. Further research is
needed in order to determine how institutional mechanisms can affect perceived
risks. First, the research should be applied in other Latin American countries, where
norms and institutions have not been eroded as much as in Ecuador and Peru.
Additionally, the measurement of this variable must be assessed and take into
considerations the use of items related to institutional mechanisms that can have
direct effect on the efficiency and security of online transactions, such as the quality
of the law and regulations and the efficiency of these institutions to ensure
compliance.
On the other hand, it must be evaluated the relationship of institutional trust with
other trust factors, such as trust in online seller, online transactions, or brands.
Even though results are similar for Ecuador and Peru, the external model assess-
ment, the MICOM, and PLSpredict analysis indicate that the PLS-SEM analysis
should be performed separately for each country. Nevertheless, the analysis should
be extended to higher samples and evaluate if the differences in external model
assessment and discriminant validity persists.
The results contribute to the literature that investigate the factors that affect
e-commerce adoption in emergent markets such as countries of Latin America.
Additionally, the results can be used by companies to improve the efficiency of
their strategies for penetrating the e-commerce Latin-American market.

References

Albashrawi M (2021) Intention to adopt e-commerce: a comparative review across developed and
developing economies. Afr J Inform Syst 13(1):113–140
Azjen I (1991) The theory of planned behavior. Organ Behav Hum Decis Process 50(2):179–211
Bergman M, Rosenkratz C (2019) Conﬁanza y derecho en América Latina. Fondo de Cultura
Económica, Ciudad de México
Boyle TC (2021) The e-commerce boom coming in America Latina. Lazar Fund Managers. https://
wwwlazardassetmanagementcom/it/en_uk/references/fundamental-focus/e-commerce-boom-
latin-america Accessed 13 May 2022
Similarities in Factors Affecting Online Shopping Intention in Ecuador. . . 97

Chevalier S (2021) E-commerce in Latin America – statistics & facts. Statista https://
wwwstatistacom/topics/2453/e-commerce-in-latin-america/#dossierKeyfigures Accessed
14 May 2022
Chin WW, Dibbern J (2010) An Introduction to a permutation based procedure for multigroup PLS
analysis: results of tests of differences on simulated data and a cross-cultural analysis of the
sourcing of information systems services between Germany and the U.S. In: Esposito Vinzi V,
Chin WW, Henseler J, Wang H (eds) Handbook of partial least squares: concepts, methods and
applications. Springer, Berlin, pp 171–193
Chintagunta PK, Chu J, Cebollada J (2012) Quantifying transaction cost in online/offline grocery
channel choice. Mark Sci 31(1):96–114
Choi J, Lee K-H (2003) Risk perception and e-shopping: a cross cultural study. J Fash Mark Manag
7(1):49–64
Cunningham N, De Meyer-Heydenrych C-F (2018) Consumer perceptions of shopping channel
attributes and risk factors of apparel purchase. Retail Market Rev 14(2):28–42
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2016) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Los Angeles
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43:115–135
Henseler J, Ringle CM, Sarstedt M (2016) Testing measurement invariance of composite using
partial least squares. Int Mark Rev 33(3):405–431
Hong IB, Cha HS (2013) The mediating role of consumer trust in an online merchant in predicting
purchase intention. Int J Inf Manag 33(6):927–939
Imtiaz R, Ain Kamzi QS, Amjad M, Aziz A (2019) The impact of social network marketing on
consumer purchase intention in Pakistan: a study on female apparel. Manag Sci Lett 9(7):
1093–1104
Iqbal W (2019) Impact of perceived risk on customer’s online purchase intention towards branded
apparel. J Market Strat 1(1):38–62
Latinobarometer (2020) Questionnaires. https://www.latinobarometro.org/latContents.jsp
Accessed 12 May 2022
Latinobarometer (2021) Informe 2021. https://www.latinobarometro.org/latContents.jsp Accessed
12 May 2022
Lim YJ, Osman A, Salahuddin SN, Romle AR, Abdullah S (2016) Factors influencing online
shopping behavior: the mediating role of purchase intention. Proc Econ Fin 35:401–410
Ling KC, Chai LT, Piew TH (2010) The effects of shopping orientations, online trust and prior
online purchase experience toward customers’ online purchase intention. Int Bus Res 3(3):63
Lodorfos GN, Trosterud TA, Whitworh C (2006) E-consumers’ attitude and behaviour in the online
commodities market. Innov Mark 2(3):77–96
Marczak J, Guevara C (2021) COVID-19 recovery in Latin America and the Caribbean: a
partnership strategy for the Biden administration. Atlantic Council. https://www.
atlanticcouncil.org/in-depth-research-reports/covid-19-recovery-in-latin-america-and-the-carib
bean-a-partnership-strategy-for-the-biden-administration/ Accessed 30 Jan 2023
Margalina VM (2022) Factores que afectan la intención de compra de los consumidores de moda en
el comercio electrónico: un modelo para América Latina. Revista de Investigación Sigma 9(1):
84–96
Oxford Business Group (March 4, 2021) How Covid-19 triggered a Latin American e-commerce
boom. https://oxfordbusinessgroup.com/news/how-covid-19-triggered-latin-american-e-com
merce-boom Accessed 10 May 2022
Park C, Jun J-K (2003) A cross-cultural comparison of internet buying behavior: effects of Internet
usage, perceived risks and innovativeness. Int Mark Rev 20(5):534–553
98 V.-M. Margalina and A. M. Cutipa Limache

Qalati SA, Galvan Vela E, Li W, Dakhan SA, Hong Thuy TT, Merani SH (2021) Effects of
perceived service quality, website quality, and reputation on purchase intention: the mediating
and moderating roles of trust and perceived risk in online shopping. Cogent Bus Manag 8(1)
Ringle CM, Wende S, Becker J (2015) SmartPLS 3 [computer software]. SmartPLS GmbH.. www.
smartpls.com. Accessed 15 May 2022
Salisbury WD, Pearson RA, Pearson AW, Miller DW (2001) Perceived security and worldwide
web purchase intention. Ind Manag Data Syst 101(4):165–177
Sarstedt M, Ringle CM, Hair JF (2021) Partial least square equation modelling. In: Homburg C,
Klarmann M, Vomberg A (eds) Handbook of marketing research. Springer, Cham, pp 1–47
Schlägel C, Sarstedt M (2016) Assessing the measurement invariance of the four-dimensional
cultural intelligence scale across countries. A composite model approach. Eur Manag J 34:633–
649
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Thomas MR, Kavya V, Monica M (2018) Online website cues inﬂuencing the purchase intention of
generation Z mediated by trust. Ind J Commer Manag Stud IX(1):13–23
Yu S, Hudders L, Cauberghe V (2018) Selling luxury products online: the effect of quality label on
risk perception, purchase intention and attitude toward the brand. J Electron Commer Res 19(1):
16–35
Zhu DS, Lee ZC, O’Neal GS, Chen YH (2011) Mr. Risk! Please Trust Me: trust antecedent that
increase online consumer purchase intention. J Internet Bank Commer 16(3):1–23
The Impact of Subjective Norms, Perceived
Behavioral Control, and Purchase Intention
on Purchase Behavior of Eco-Friendly Food
Packaging Products

Julen Castillo-Apraiz, Jesús Manuel Palma-Ruiz,

and Mauricio Iván García-Montes

1 Introduction

Current patterns of consumption seem to be non-suitable considering the growing

needs of natural limited resources (Koenig-Lewis et al. 2014). Consumers are aware
of that fact and are encouraged to contribute to long-term growth, including envi-
ronmental preservation (Prakash and Pathak 2017). Hence, consumer knowledge of
environmental concerns is raising demand for environmentally friendly items
(Ramesh and Samudhra Rajakumar 2019), which implies taking into account prod-
ucts’ packaging.
Consumers are becoming increasingly worried about the environmental impact of
products’ packaging (Nguyen et al. 2020). Given the role of packaging in solid and
liquid waste, customers now demand ecologically friendly or green packaging
(Singh and Pandey 2018). Food packaging makes it easier to store, handle, transport,
and preserve food and helps to avoid food waste. Besides these advantages, food
packaging is causing increasing environmental concern due to its large manufactur-
ing volume, typically short usage period, and waste management and littering issues
(Geueke et al. 2018). As a result, any decrease in packaging waste helps to achieve
sustainability-related goals (Herbes et al. 2018).

J. Castillo-Apraiz (✉)
Department of Financial Economics II, Faculty of Economics and Business Administration,
University of the Basque Country UPV/EHU, Bilbao, Spain
e-mail: [email protected]
J. M. Palma-Ruiz · M. I. García-Montes
Facultad de Economía Internacional, Universidad Autónoma de Chihuahua, Chihuahua,
Mexico
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 99

Not surprisingly, multinational corporations are concentrating on packaging

advancements in order to decrease waste, increase recycled content, and develop
more renewable packaging materials (Prakash and Pathak 2017). Globally active
corporations are aligning their business strategies with the sustainable development
agenda (Kulshreshtha et al. 2019) by focusing on green marketing, which is ﬁrms’
commitment to developing safe, environmentally friendly products by using recy-
clable and readily decomposable packaging, improved pollution control strategies,
and more efﬁcient energy usage (Mukonza and Swarts 2019). In fact, not only
consumers but also governments are pushing businesses to use environmentally
friendly packaging for their products (Nguyen et al. 2020). Lawmakers, environ-
mental lobbyists, consumers, and the packaging sector itself have all expressed
concern over excessive packaging waste (Steenis et al. 2017).

2 Background: Environmentally Friendly Packaging

In-store purchase decisions for food goods are characterized by low participation and
impulsive processes, and packaging plays a crucial role in influencing these deci-
sions (Liao et al. 2015). Packaging is a coordinated system of preparing goods for
safe, secure, efficient, and effective handling, transport, storage, retailing, consump-
tion and recovery, reuse or disposal combined with maximizing consumer value,
sales, and hence profit (Saghir 2004, p. 34). Ecologically friendly or environmentally
aware packaging (also known as green packaging, eco-friendly packaging, or sus-
tainable packaging) has become increasingly popular in recent years (Herbes et al.
2018). Hence, packing’s impact on purchasing intention is increasing. Nonetheless,
the extent to which food and beverage packaging impacts customer purchase
decisions is still fairly unknown (Fraser 2018).

3 Literature Review and Hypotheses Development

The theory of planned behavior (TPB) (Ajzen 1991) and, to a lesser degree, its
precursor, the theory of reasoned action (Madden et al. 1992), have been used in
most investigations involving the prediction of behavior from attitudinal variables
(Ajzen and Fishbein 2000). Correspondingly, people act in line with their intentions
and perceptions of control over the behavior, whereas intentions are impacted by
subjective norms and perceived behavioral control.
The Impact of Subjective Norms, Perceived Behavioral Control, and. . . 101

3.1 Subjective Norms

Perceived social pressures or subjective norms (SN) to carry out an action are based
on people’s view about others’ expectations and motivation to meet those expecta-
tions (or not) (Ajzen 2001). Individuals differ in the relative importance they place
on SN, thus contributing differently across actions and subject populations (Ajzen
1991; Ajzen and Fishbein 2000). SN play an important role inﬂuencing purchasing
intention (Ajzen 2012; Bech-Larsen 1996; Prakash and Pathak 2017; Rokka and
Uusitalo 2008; Zakersalehi and Zakersalehi 2012). Hence, we hypothesize:
H1: Subjective norms (SN) is positively related to purchase intention (PI).

3.2 Perceived Behavioral Control

Most human social behavior is under voluntary control and so can be predicted only
on the basis of intentions. The ideas regarding perceived power, also known as
perceived behavioral control (PBC), are concerned with the ease or difﬁculty of
carrying out an activity (Ajzen 2002). A high sense of control should enhance a
person’s desire to do the action, as well as their effort and persistence. In this way,
the impact of PBC on intention might inﬂuence behavior indirectly (Fishbein and
Ajzen 2010). Thus, we hypothesize:
H2: Perceived behavioral control (PBC) is positively related to purchase intention
(PI).

3.3 Purchase Intention and Purchase Behavior

According to the TPB hypothesis, behavior may be predicted since intention inﬂu-
ences behavior (Ajzen 1991). Hence, intentions are thought to capture the motivat-
ing variables that drive conduct; they are indicators of how hard individuals are
willing to try, and how much effort they plan to put in to complete the activity (Ajzen
2012). Previous studies have shown that intentions and perceptions of behavioral
control that were relatively constant were stronger predictors of subsequent behav-
ior. For example, purchase intention and purchase behavior are linked in the context
of sustainable food choices (Campbell and Fairhurst 2016) and pro-environmental
purchase behavior (Nguyen et al. 2017). As a result, there is a need of promoting
individuals to engage in pro-environmental actions (Steg et al. 2014), as to demon-
strate the causal link between variables to impact the purchase behavior of
eco-friendly food packaging as shown in Fig. 1. Therefore, we hypothesize:
H3: Purchase intention (PI) is positively related to purchase behavior (PB).
102 J. Castillo-Apraiz et al.

Fig. 1 Research model

4 Methodology, Sample, and Data Collection

Data were obtained from 388 completed surveys in late 2020 through Qualtrics
platform. It was mainly distributed in-person to supermarket customers in Mexico.
The variables were measured based on previously validated ﬁve-point Likert scales.

5 Results and Discussion

We aim to use partial least squares structural equation modeling (PLS-SEM) for data
analysis, which allows estimating path models with latent variables (Hair et al.
2019). Furthermore, PLS-SEM is extensively used in areas such as marketing
(Sarstedt et al. 2022a) and is suitable when the research is primarily causal-predictive
in nature (Sarstedt et al. 2022b). We aim to apply necessary condition analysis
(NCA) too, whose use in conjunction with PLS-SEM constitutes a methodological
contribution (Bergh et al. 2022). Data will mainly be analyzed through SmartPLS
software (Ringle et al. 2015).

6 Conclusions

We aim to demonstrate that subjective norms, perceived behavioral control, and

purchase intention are effective predecessors of purchase behavior of eco-friendly
food packaging. Using rigorous and robust study methods, a more extensive inves-
tigation of individual variables and their inﬂuence on customer purchase intention is
required (Fraser 2018). As a result, packaging designers will be more informed on
how to create packages that customers want and are willing to buy if there is a solid
evidence foundation.
The Impact of Subjective Norms, Perceived Behavioral Control, and. . . 103

Acknowledgments The authors appreciate the support received from the Fundación Emilio
Soldevilla para la Investigación y el Desarrollo en Economía de la Empresa (FESIDE) foundation.

References

Ajzen I (1991) The theory of planned behavior. Organ Behav Hum Decis Process 50(2):179–211
Ajzen I (2001) Nature and operation of attitudes. Annu Rev Psychol 52(1):27–58
Ajzen I (2002) Perceived behavioral control, self-efficacy, locus of control, and the theory of
planned behavior. J Appl Soc Psychol 32(4):665–683
Ajzen I (2012) The theory of planned behavior. In: van Lange PAM, Kruglanski AW, Higgins ET
(eds) Handbook of theories of social psychology, vol 1. SAGE Publications Ltd, US, pp
438–459
Ajzen I, Fishbein M (2000) Attitudes and the attitude-behavior relation: reasoned and automatic
processes. Eur Rev Soc Psychol 11(1):1–33
Bech-Larsen T (1996) Danish consumers’ attitudes to the functional and environmental character-
istics of food packaging. J Consum Policy 19(3):339–363
Bergh DD, Boyd BK, Byron K, Gove S, Ketchen DJ Jr (2022) What constitutes a methodological
contribution? J Manag 48(7):1835–1848
Campbell JM, Fairhurst AE (2016) Reducing the intention-to-behaviour gap for locally produced
foods purchasing. Int J Retail Distrib Manag 44(5):508–523
Fishbein M, Ajzen I (2010) Predicting and changing behavior: the reasoned action approach.
Psychology Press, New York
Fraser A (2018) The influence of package design on consumer purchase intent. In: Pascall MA, Han
JH (eds) Packaging for nonthermal processing of food. John Wiley & Sons, Ltd., Hoboken, NJ,
pp 225–249
Geueke B, Groh K, Muncke J (2018) Food packaging in the circular economy: overview of
chemical safety aspects for commonly used materials. J Clean Prod 193:491–505
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Herbes C, Beuthner C, Ramme I (2018) Consumer attitudes towards biobased packaging – a cross-
cultural comparative study. J Clean Prod 194:203–218
Koenig-Lewis N, Palmer A, Dermody J, Urbye A (2014) Consumers’ evaluations of ecological
packaging – rational and emotional approaches. J Environ Psychol 37:94–105
Kulshreshtha K, Bajpai N, Tripathi V, Sharma G (2019) Consumer preference for eco-friendly
appliances in trade-off: a conjoint analysis approach. Int J Prod Dev 23(2–3):212–243
Liao LX, Corsi AM, Chrysochou P, Lockshin L (2015) Emotional responses towards food
packaging: a joint application of self-report and physiological measures of emotion. Food
Qual Prefer 42:48–55
Madden TJ, Ellen PS, Ajzen I (1992) A comparison of the theory of planned behavior and the
theory of reasoned action. Personal Soc Psychol Bull 18(1):3–9
Mukonza C, Swarts I (2019) The influence of green marketing strategies on business performance
and corporate image in the retail sector. Bus Strategy Environ 29:1–8
Nguyen TN, Lobo A, Greenland S (2017) The influence of cultural values on green purchase
behaviour. Mark Intell Plan 35(3):377–396
Nguyen AT, Parker L, Brennan L, Lockrey S (2020) A consumer definition of eco-friendly
packaging. J Clean Prod 252. (April):119792
Prakash G, Pathak P (2017) Intention to buy eco-friendly packaged products among young
consumers of India: a study on developing nations. J Clean Prod 141:385–393
Ramesh M, Samudhra Rajakumar C (2019) Determinants of online purchase decision of green
products. Int J Eng Adv Technol 9(1):1477–1481
104 J. Castillo-Apraiz et al.

Ringle CM, Wende S, Becker J-M (2015) Smartpls 3. Hamburg, SmartPLS computer software
Rokka J, Uusitalo L (2008) Preference for green packaging in consumer product choices – do
consumers care? Int J Consum Stud 32(5):516–525
Saghir M (2004) The concept of packaging logistics. In: Proceedings of the ﬁfteenth annual POMS
conference. April: Cancún
Sarstedt M, Hair JF, Pick M, Liengaard BD, Radomir L, Ringle CM (2022a) Progress in partial least
squares structural equation modeling use in marketing research in the last decade. Psychol Mark
39(5):1035–1064
Sarstedt M, Hair JF, Ringle CM (2022b) “PLS-SEM: indeed a silver bullet”–retrospective obser-
vations and recent advances. J Mark Theory Pract 1:1–15
Singh G, Pandey N (2018) The determinants of green packaging that inﬂuence buyers’ willingness
to pay a price premium. Australas Mark J 26(3):221–230
Steenis ND, van Herpen E, van der Lans IA, Ligthart TN, van Trijp HCM (2017) Consumer
response to packaging design: the role of packaging materials and graphics in sustainability
perceptions and product evaluations. J Clean Prod 162:286–298
Steg L, Bolderdijk JW, Keizer K, Perlaviciute G (2014) An integrated framework for encouraging
pro-environmental behaviour: the role of values, situational factors and goals. J Environ Psychol
38:104–115
Zakersalehi M, Zakersalehi A (2012) Attitude and purchasing intention of Malaysian consumers
toward green packaged foods. Int J Trade Econ Financ 3(1):46–51
Sustainability of Food Waste
in the Agro-industry of Michoacan, Mexico

Jesús Sigifredo Gastélum-Valdez, Irma Cristina Espitia-Moreno,

and Betzabé Ruiz-Morales

1 Introduction

This research was developed with the objective of identifying the variables that
determine the sustainability of food waste in the agro-industrial sector of Michoacan.
The hypothesis proposed is that agrifood management, supply chain and green
marketing are the variables that positively explain the sustainability of food waste
management in the agro-industrial sector of Michoacan. As justiﬁcation for the topic
and study variables, the main objectives of the United Nations Sustainable Devel-
opment Program were studied: to ensure access for all people to a sufﬁcient amount
of healthy (Sachs 2012), nutritious food throughout the year, ensuring the sustain-
ability of food production systems and implementing agricultural practices that
increase productivity and production, contribute to the maintenance of ecosystems
and strengthen the capacity to adapt to climate change (Gruda 2019). This target is
strongly linked to halving global per capita food waste at the retail and consumer
levels and reducing food losses in production and supply chains, including post-
harvest losses (UN 2020).

J. S. Gastélum-Valdez (✉)
Universidad Latina de América, Morelia, Mexico
e-mail: [email protected]
I. C. Espitia-Moreno · B. Ruiz-Morales
Universidad Michoacana de San Nicolás de Hidalgo, Morelia, Mexico
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 105

2 Background

The article ‘Agri-food sustainable marketing research: a bibliometric analysis of the

evolution of the field since 1990’ was made to provide a bibliometric analysis of the
evolution of the study of agrifoods with sustainability and marketing (Ruiz-Morales
et al. 2023). The data were obtained from an exhaustive analysis. Using Web of
Science (WoS), 1,134 papers from 1900 to 2019 were retrieved and analysed. Recent
years have witnessed growth in the number of publications referring to sustainabil-
ity, marketing and agrifoods, and the scientific community is becoming increasingly
interested in the relationship between agrifoods and sustainability. Important results
and information are presented about the most cited, more productive and influential
authors, organizations, journals, countries and papers in this field.
On the other hand, the agrifood supply chain is a topic of great interest for various
public and private sectors. In terms of sustainability, it has great importance due to
the continuous deterioration of the environment. One of the biggest challenges is the
design of the supply chain. Due to the complexity it represents, many researchers
prefer to study only one or some parts of the supply chain (Allaoui et al. 2018). The
agrifood supply chain, unlike other chains, involves indicators of food quality,
safety, seasonality and a very limited shelf life of products (Alvarenga et al. 2015).
Consumers increasingly demand organic and sustainable food; due to this trend, the
agrifood supply chain considers not only the cultivation of food but also its distri-
bution from the field to the customer, processing, packaging, storage, and handling
of by-products and waste (Gutiérrez and Suarez 2014). Consumers have become
more inquisitive about food attributes, such as quality, diversity, health and sustain-
ability. To meet these demands, companies seek to incorporate sustainability into
their operations, and the design of agrifood supply chain management has become
extremely important to determine a competitive advantage in the sector. The
agrifood supply chain unlike other chains involves indicators of food quality, safety,
seasonality and a very limited shelf life of products (Alvarenga et al. 2015). In
addition, the agrifood supply chain variable is used by the Food and Agriculture
Organization (FAO 2020) for its annual reports on food waste at each stage of the
supply chain.
In agrifood management, organic wastes of any type are involved, including food
waste; unlike other wastes, they must be treated in a special way because they
degrade easily, allowing the use of nutrients and energy from them. If they do not
have the proper treatment, they can produce undesirable degradation results, such as
methane emissions to the atmosphere and the proliferation of pests (Platt et al. 2014).
Approaches to organic waste management are paramount to promote waste utiliza-
tion and sustainability and the creation of new business ventures (Gómez-Brandón
and Podmirseg 2013).
Recent waste-sorting studies have indicated that large amounts of food waste that
are disposed of can be used to nourish the soil and the components of the waste by
their organic nature can be converted into energy (Platt et al. 2014). At the same
time, the interest in decreasing or taking advantage of food waste stems from
Sustainability of Food Waste in the Agro-industry of Michoacan, Mexico 107

growing concerns about world hunger, resources, food security and by a trend in the
waste generating industry toward a transition to the use of more sustainable prac-
tices. In addition, society has developed social and environmental awareness,
including food waste (Griffin et al. 2009). This has supported the prevention and
recovery of food waste through alternative management methods to reduce the
environmental impacts caused by waste generation (Schott et al. 2013).
To achieve the above, there is a pressing need to move toward a more advanced
technological approach in the revaluation of food by waste into higher value
applications such as fuels and chemicals through biorefinery technologies (Banasik
et al. 2017). Much research has been conducted on the revalorization of food waste
into chemicals, in which a multidimensional assessment approach of the resource
recovery system from waste is proposed to capture the complex values of environ-
mental, economic and social impacts (Govindan 2018). To develop food waste
management and revaluation strategies, it is desirable to have a framework that
focuses more on integrated systems design and sustainability analysis to preserve the
value of materials and products over a longer period of time by closing the life cycle
of products, creating coproducts (they maintain a value for human use or consump-
tion) or by-products (their value is for animal, plant or energy consumption or use)
(Padfield et al. 2016).
For the marketing of coproducts or by-products, there is green marketing, another
variable proposed in the hypothesis, which shows consumers of a new generation,
who are concerned about the environmental damage caused by the production of
goods and services they buy (Hahladakis and Iacovidou 2019). This type of con-
sumer aims to purchase products, in which the companies that manufacture them are
mindful of environmental costs and minimize or avoid the damage caused to the
environment (De las Fuentes et al. 2004).
Finally, the supply chain management variable is a set of means used to effi-
ciently integrate suppliers and manufacturers, warehouses and stores, so that mer-
chandise is produced and distributed at the right time, potentially leading to cost
minimization (Allaoui et al. 2018). In a traditional supply chain, raw materials are
procured, and product transformation is performed in one place, subsequently
shipped to wholesale or retail intermediaries, and finally purchased by customers
(Ng et al. 2019). To be efficient and decrease costs, effective strategies must be
implemented within the entire supply chain, to identify interactions at each level of
the supply chain. The levels include logistics, suppliers, manufacturing sites, ware-
houses, distribution, retail outlets, raw materials, in-process and finished goods
inventories. The causes of food waste in the supply chain, according to Raak et al.
(2017), are related to logistical handling, product processing and consumption.
108 J. S. Gastélum-Valdez et al.

3 Method

To obtain data, a survey was conducted in 2021 to 93 agro-industrial producers and

marketers of avocado, mango, blackberry, lemon, guava and strawberries belonging
to the state of Michoacan to determine their perceptions regarding sustainability and
agrifood management. The initial model has 40 items, 15 corresponding to the
dependent variable sustainability of food waste, 10 to the independent variable
agrifood management, 11 to the supply chain variable and 4 to the green marketing
variable. This model proposes that relationship between agrifood management,
marketing and supply chain management positively influences the sustainability of
food waste in the agrifood industry in the state of Michoacan and is a reflective
model.
For data management, a structural equation model using the partial least squares
approach was used to determine the influence of agrifood management, supply chain
and green marketing on the sustainability of food waste. To test the model,
SmartPLS 3 was used. The first step to evaluate the PLS-SEM results according to
Hair et al. (2019) is to examine the measurement models; common protocols suggest
that the reliability for exploratory research should be a minimum of 0.60, to be
considered an acceptable value in exploratory research; in the results, a 0.599
coefficient of determination was obtained; also, loadings higher than 0.708 are
recommended, since they indicate that the construct explains more than 50% of
the variance of the indicator, thus providing an acceptable reliability of the item, in
the model performed.
The final model consists of five items for the dependent variable sustainability of
food waste, six items for agrifood management, eight items for supply chain and four
for green marketing. The R2 coefficient is 0.599; the following image shows the
result of the model (Fig. 1).

4 Results and Discussion

The first step for the report of a PLS-SEM model is the measurement model results.
The results depend on the type of model; in this case, the measurement model is
reflective, so the first step is to evaluate the indicator loadings (Hair et al. 2019).
From the original 40 items, 17 items with loadings below 0.708 were eliminated.
The second evaluation in the results is assessing internal reliability, measured by
Cronbach’s alpha and the composite reliability coefficient. Table 1 shows these
results. All latent variables used in the model have a good level of internal reliability,
having values above 0.70 in both coefficients (Joreskög 1971). Additionally, it is
necessary to bootstrap to test the confidence intervals of the internal reliability; in
this case, in a 95% confidence interval, the reliability of the construct should be
greater than 0.70, so that both measures are acceptable.
Sustainability of Food Waste in the Agro-industry of Michoacan, Mexico 109

Fig. 1 Structural equation model results

Table 1 Internal reliability

Cronbach’s alpha Composite reliability coefﬁcient
Supply chain 0.867 0.894
Agrifood management 0.825 0.874
Marketing 0.783 0.860
Sustainability 0.818 0.873

The third step of reporting an PLS-SEM model is to assess the convergent validity
of each construct. The average variance extracted (AVE) is the metric used to
evaluate convergent validity. The minimum acceptable AVE is 0.50, and 0.50 or
higher indicates that the constructs explain 50% or more of the variance of the items
that form the constructs. All constructs used in the model show that the AVE are
above the 0.50 threshold value. Table 2 shows the results of rho_A and the AVE
coefﬁcients.
The next step is to assess discriminant validity, which is the extent to which a
construct is empirically distinct from other constructs (Hair et al. 2019). Henseler
et al. (2015) proposed the heterotrait-monotrait (HTMT) ratio to assess discriminant
validity. The HTMT threshold value must be below 0.90 to suggest that discriminant
validity is not present. In this model the constructs have coefﬁcients below 0.90, and
these results are reported in Table 3.
110 J. S. Gastélum-Valdez et al.

Table 2 Rho_A and average variance extracted

rho_A Average variance extracted (AVE)
Supply chain 0.891 0.894
Agrifood management 0.837 0.874
Marketing 0.875 0.860
Sustainability 0.831 0.873

Table 3 Heterotrait-monotrait ratio

Supply chain Agrifood management Green marketing
Agrifood management 0.769
Marketing 0.557 0.376
Sustainability 0.863 0.690 0.392

Table 4 Collinearity assessment

VIF
Desp_1 3.382
Desp_2 3.441
Desp_3 2.918
Desp_4 2.386
Desp_7 2.946
Dist_1 1.742
Dist_2 1.900
Dist_5 1.345
Dist_6 1.268
Ecol_2 2.185
Ecol_3 3.829
Ecol_6 2.004
Econ_1 3.465
Prod_1 2.714
Prod_2 2.096
Prod_3 2.639
Prod_4 1.594
Rev_3 2.988

If the model assessment is satisfactory, the next step is evaluating the structural
model. In this part standard assessment criteria are the coefﬁcient of determination
R2, blindfolding-based cross-validated redundancy measure Q2 and the statistical
signiﬁcance and relevance of the path (Hair et al. 2019). Additionally, it is
recommended that the model predictive power be assessed using the PLSpredict
procedure (Shmueli et al. 2016).
Before assessing the structural model, it is necessary to examine collinearity
measured with VIF values, and VIF values above 5 indicate probable collinearity.
Table 4 shows the VIF values for the model. These values are below the threshold
value of 5.
Sustainability of Food Waste in the Agro-industry of Michoacan, Mexico 111

Fig. 2 Structural equation model assessment

If the model is not a problem, the next step is the R2 coefficient evaluation. The R2
coefficient measures the variance that is explained in each of the endogenous
constructs and is also a measure of the model explanatory power (Shmueli and
Koppius 2011); the closest to 1 more explanatory power has the model. As a
guideline, Henseler et al. (2009) and Hair et al. (2011) consider values of 0.75,
0.50 and 0.25 as substantial, moderate and weak, respectively. In this research, the
coefficient R2 is 0.852, so the explanatory power is considered substantial.
Another significant structural model assessment is the path coefficient and its
statistical significance. The results of the model are shown in Fig. 2. In this figure, the
most significant path goes from the supply chain to sustainability with a path
coefficient of 0.755, the second path coefficient is from agrifood management to
sustainability with a value of 0.189, and finally, the path coefficient of -0.004 is
from marketing to sustainability.
With the statistical significance of the path coefficients, the results of the
bootstrapping algorithm show that only the path between supply chain and sustain-
ability is significant and green marketing and agrifood management are nonsignif-
icant, having p values above the 0.05 threshold value.
112 J. S. Gastélum-Valdez et al.

4.1 Implications

The research carried out allows establishing a contextualization of the subjects of

study to define the particular characteristics of the agro-industrialists of the state of
Michoacán with respect to the sustainable practices they carry out for the manage-
ment of the wastes that originate in the whole productive process and the supply
chain.
It was identified that sustainable practices for waste management are carried out
by few agro-industrialists in Michoacán, most of whom are avocado producers or
traders, which allows favourable results to be obtained with respect to the research
topic, allowing a future comparison with individuals and companies with agro-
industrial activities that do not carry out sustainable practices.
This research corroborated that the majority of agro-industrialists are extremely
interested in taking advantage of the food waste that is generated, allowing them to
enter new target markets, to obtain greater profits and improve their competitive
position and their environmental performance.
On the other hand, the development of this research showed that agro-
industrialists carry out at least one sustainable practice within their productive,
operational and distribution processes, highlighting that the greatest food waste
originates in the post-harvest period and that it is possible to use this waste for its
revaluation, turning it into a new product for human, animal or vegetable use.

4.2 Limitations

The research focused on the directory of agro-industrialists attached to CIDAM,

leaving aside farmers and individuals and companies with agro-industrial activities
that are not included in the directory.
Most of those surveyed commercialize or produce avocado, which is one of the
products from which greater use can be made of waste, in addition to the fact that all
the agro-industrialists surveyed have formal economic activities in the SAT and are
registered in DENUE.

5 Conclusions

To test the hypothesis, the path coefficients from the evaluation of the structural
model and the statistical significance of these coefficients were considered. In this
case, the supply chain was considered to be the independent variable with the
greatest influence on sustainability, since it has both a positive relationship and an
influence of 0.755 of explanation.
Sustainability of Food Waste in the Agro-industry of Michoacan, Mexico 113

The second most influential variable was agrifood management, which also had a
positive relationship with sustainability, with a value of 0.189. Finally, the marketing
variable had a negative sign and a low value of -0.004 for sustainability. Based on
the first part of the hypothesis test, the condition that the three independent variables
chosen have a positive influence on the sustainability of waste is fulfilled.
With the results obtained from the structural equation model, it is considered that
the hypothesis is accepted, because the three independent variables positively
influence the sustainability of losses, despite the small contribution that marketing
has to sustainability and because only the other two variables are significant.

5.1 Future Lines of Research

This research could be replicated for each product produced in the state of
Michoacán, allowing a comparative analysis between the different fruits and vege-
tables and the sustainable management of waste applied to each product and the
forms of revaluation that they can have with respect to the nature of the product
itself.
Apply the same measuring instrument to agricultural producers in different
regional zones of the state, to enrich the research and obtain precise data on the
sustainability of waste management directly in the ﬁeld.
Apply the measurement instrument to individuals or companies engaged in
marketing agricultural products, to obtain the food waste that originates in the
following post-harvest stages and the sustainable practices that are carried out.
Include the dimensions of green marketing, to have more detailed knowledge
about the market, price and promotion of the products revalued from food waste.
Conduct research through the study of marketing, where national and interna-
tional markets can be identiﬁed, which can be accessed through the sale of food
waste or new products made from food waste.

References

Allaoui H, Guo Y, Choudhary A, Bloemhof J (2018) Sustainable agro-food supply chain design
using two-stage hybrid multi-objective decision-making approach. Comput Oper Res 89:369–
384
Alvarenga P, Mourinha C, Farto M, Santos T, Palma P, Sengo J, Morais M-C, Cunha-Queda C
(2015) Sewage sludge, compost and other representative organic wastes as agricultural soil
amendments: beneﬁts versus limiting factors. Waste Manag 40:44–52
Banasik A, Kanellopoulos A, Claassen GDH, Bloemhof-Ruwaard JM, Van der Vost JGAJ (2017)
Closing loops in agricultural supply chains using optimization: a case study of an industrial
mushroom supply chain. Int J Prod Econ 183, Part B:409–420
De Las Fuentes L, Sanders B, Lorenzo A, Alber S (2004) Agro-food wastes minimisation and
reduction network. In: Waldron K, Faulds C, Smith A (eds) Total food: exploiting
co-products—minimizing wastes. Norwich, Institute of Food Research
114 J. S. Gastélum-Valdez et al.

FAO (2020) Desperdicio de alimentos. Organización de las Naciones Unidas para la Alimentación
y la Agruciltura. https://www.fao.org/news/story/es/item/1310444/icode/ Accessed
29 Mar 2020
Gómez-Brandón M, Podmirseg SM (2013) Biological waste treatment. Waste Manag Res 31(8):
773–774
Govindan K (2018) Sustainable consumption and production in the food supply chain: a conceptual
framework. Int J Prod Econ 195:419–431
Griffin M, Sobal J, Lyson T (2009) An analysis of a community food waste stream. Agric Hum
Values 26(1–2):67–81
Gruda NS (2019) Increasing sustainability of growing media constituents and stand-alone sub-
strates in soilless culture systems. Agronomy 9(6):298
Gutiérrez J, Suarez M (2014) Agrocombustibles y agroalimentos. Considerando las externalidades
de la mayor encrucijada del siglo XXI. Agroecología 4:123–134
Hahladakis JN, Iacovidou E (2019) An overview of the challenges and trade-offs in closing the loop
of postconsumer plastic waste (PCPW): focus on recycling. J Hazard Mater 380:120887
Hair JF, Ringle CM, Sarsted M (2011) PLS-SEM: indeed, a silver bullet. J Mark Theory Pract
19(2):139–151
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. In: Sinkovics RR, Ghauri PN (eds) Advances in international market-
ing. Bingley, Emerald, pp 277–320
Henseler J, Ringle CM, Sarsted M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Joreskög KG (1971) Simultaneous factor analysis in several populations. Psychometrika 36(4):
409–426
Ng SK, Yang A, Yakovleva N (2019) Sustainable waste management through synergistic utilisation
of commercial and domestic organic waste for efficient resource recovery and valorisation in the
UK. J Clean Prod 227:248–262
Padfield R, Drew S, Syayuti K, Page S et al (2016) Landscape in transition: an analysis of
sustainable policy initiatives and emerging corporate commitments in the palm oil industry.
Landsc Res 41(7):744–756
Platt B, Goldstein N, Coker C, Brown S (2014) State of composting in the U.S. Institute for Local
Self Reliance, Emmaus, PA
Raak N, Symmank C, Zahn S, Aschemann-Witzel J, Rohm H (2017) Processing- and product-
related causes for food waste and implications for the food supply chain. Waste Manag 61:461–
472
Ruiz-Morales B, Espitia-Moreno IC, Galeana-Figueroa E, Alfaro-Garcia VG (2023) Agri-food
sustainable marketing research: a bibliometric analysis of the evolution of the field since 1990.
Int J Innov Sustain Dev 17(1/2):85–103
Sachs JD (2012) From millennium development goals to sustainable development goals. Lancet
379(9832):2206–2211
Schott ABS, Vukicevic S, Bohn I, Andersson T (2013) Potentials for food waste minimization and
effects on potential biogas production through anaerobic digestion. Waste Manag Res 31(8):
811–819
Shmueli G, Koppius OR (2011) Predictive analytics in information system research. MIS Quart
35(3):553–572
Shmueli G, Ray S, Estrada JMV, Chatla SB (2016) The elephant in the room: predictive perfor-
mance of PLS models. J Bus Res 69(10):4552–4564
UN (2020) Global indicator framework for the sustainable development goals and targets of the
2030 agenda for sustainable development. https://unstats.un.org/sdgs/indicators/Global%20
Indicator%20Framework%20after%202020%20review_Eng.pdf Accessed 6 Mar 2022
Predicting Sustainable Consumption
Behavior of Europeans Using the CVPAT

Lena Frömbling, Svenja Damberg, Ulla A. Saari, and Christian M. Ringle

1 Introduction

Predictive power assessment has become an integral part of evaluating statistical

models (e.g., Shmueli et al. 2019). Hence, researchers are more and more interested
in the congruence of explanatory and predictive modeling. While explanatory
modeling focuses mainly on model fit as well as the significance and direction of
path coefficients included in a model, the focus in predictive modeling is on making
statements regarding observations that are not included in the sample at hand.
Predictive modeling further provides the possibility to infer about generalizability
and practical performance (Sarstedt and Danks 2022).
In our research, we build on the study by Saari et al. (2021). In their article, the
authors constructed a structural equation model with the aim of explaining and
predicting sustainable consumption behavior among Europeans. Therein, they
included environmental knowledge and risk perception as independent constructs

L. Frömbling (✉) · C. M. Ringle

Department of Management Sciences and Technology, Hamburg University of Technology,
Hamburg, Germany
e-mail: [email protected]; [email protected]
S. Damberg
Hamburg University of Technology, Department of Management Sciences and Technology,
Institute for Technology and Innovation Management, Am Schwarzenberg-Campus 4(D),
Hamburg, Germany
e-mail: [email protected]
U. A. Saari
Tampere University, Faculty of Management and Business, Industrial Engineering and
Management, Tampere, Finland
e-mail: ulla.saari@tuni.ﬁ

© Springer Nature Switzerland AG 2023 115

and environmental concern and behavioral intention as mediators. However, the

authors further suggested that the causal-predictive relationship from risk perception
to environmental concern could also be reversed (Saari et al. 2021). The goal of this
chapter is to test the proposed alternative model in terms of prediction. For this
purpose, we apply the cross-validated predictive ability test (CVPAT; Liengaard
et al. 2021; Sharma et al. 2023), which assists researchers in their predictive model
assessment (Chin et al. 2020; Hair 2021).

2 Theoretical Background

Environmental protection (Stern 2000) and sustainable consumption (Pagiaslis and

Krontalis 2014) are key issues in modern society. Highly researched constructs that
are assumed to be connected to sustainable consumption behavior are environmental
concern and risk perception of individuals. Environmental concern can be seen as an
attitude or value orientation related to the environment (Fransson and Gärling 1999),
whereas risk perception refers to the subjective, nonexpert interpretation of certain
hazards in their bandwidth of riskiness (Bickerstaff 2004). In other words, environ-
mental concern is a more global view on the environment, while risk perception is
related to speciﬁc dangers (Saari et al. 2021).
Even though there is evidence that risk perception can lead to environmental
concern, there is also a rationale for assuming the reverse direction. In the former
case, experienced risks add up to an overall concern regarding the environment
(Saari et al. 2021). However, we could also argue that an overall concern might open
the mind and lead to perceiving certain external circumstances that are dangerous to
the environment. Underlying this assumption is, for example, evidence by Lima
et al. (2005), who found that environmental awareness is related to risk perception.
Therefore, we propose an alternative model that we test in comparison to the model
of Saari et al. (2021). Figure 1 shows the established as well as the alternative model.

3 Methodology and Data

In our analysis, we use the same sample (N = 11,675) as Saari et al. (2021), which is
included in the Environment III dataset from the ISSP Research Group (2012).
Consequently, we include the following countries: Austria, the Czech Republic,
Denmark, Finland, France, Germany, the Netherlands, Norway, and Sweden, which,
according to the Sustainable Development Goals index score, ranked among the ten
most sustainable countries in 2020 (Sachs et al. 2020).
Partial least squares structural equation modeling (PLS-SEM) is the method of
choice when the aim of a research study is to model causal-predictive relationships in
a path model (Sarstedt et al. 2023). Furthermore, the PLSpredict analysis provides
researchers with the ability to assess a model’s out-of-sample predictive power
Predicting Sustainable Consumption Behavior of Europeans Using the CVPAT 117

Fig. 1 The established and alternative model (based on Saari et al. 2021). Note: Established model
at the top, alternative model at the bottom

(Shmueli et al. 2016; Shmueli et al. 2019). The CVPAT goes one step beyond the
PLSpredict analysis, in that this methodology enables researchers to pairwise com-
pare and test theoretically derived models in terms of their out-of-sample predictive
power (Chin et al. 2020; Liengaard et al. 2021). Hence, the CVPAT is particularly
suitable when researchers strive to make statements regarding prediction and gen-
eralization that go beyond their own sample (Liengaard et al. 2021). In this sense,
the CVPAT provides an overall predictive view on the models considered instead of
solely comparing path differences (Liengaard et al. 2021; Sarstedt et al. 2023).
118 L. Frömbling et al.

Table 1 CVPAT results

Average losses
EM AM EM-AM p-value
CVPAT results 0.90989 0.90943 0.00046 0.00000
Notes: CVPAT cross-validated predictive ability test, EM established model, AM alternative model.
p-value < 0.001

For the CVPAT analysis, we follow the instructions by Liengaard et al. (2021).
Step 1 of the analysis deals with the selection of two theoretically derived models. As
shown in Fig. 1, we base our reasoning on Saari et al. (2021) and provide a theory-
driven rationale to back up the alternative model. Step 2 involves testing the
established and alternative model by means of a PLS-SEM analysis to ensure data
and measurement quality. Step 3 is applying the CVPAT to assess both models’ out-
of-sample predictive power. To do so, we use the R statistical software (R Core
Team 2022) and the CVPAT code (Liengaard et al. 2020). In Step 4, we select a
model based on the results of the analysis.
During Step 2 of the procedure, we assess the alternative model1 in terms of their
measurement and structural models’ quality criteria by using SmartPLS 4 (Ringle
et al. 2022) and follow the instructions by Hair et al. (2019, 2022). Hence, we
consider the outer loadings of the reflective measurement models as well as the
construct reliability (ρA and ρC) and construct validity (AVE) measures. Further-
more, according to the variance inflation factor (VIF) values, we do not have critical
collinearity issues. Also, the heterotrait-monotrait (HTMT) ratio values, considering
the upper bound of the one-sided confidence intervals, are below the conservative
threshold and confirm our reflective constructs’ discriminant validity. In terms of
in-sample explanatory power, an example R2 value for sustainable consumption
behavior is 0.289. We therefore confirm our model’s suitable explanatory power.
Furthermore, we use PLSpredict and the Bayesian information criterion (BIC) in
a multi-method approach to assess both models’ out-of-sample predictive power. We
find that the value differences between the PLSpredict results for PLS-SEM and the
linear model benchmark are small in both cases. While they point in the direction of
the PLS-SEM results, we cannot make clear statements regarding their superior
predictive power. When comparing the BIC values for the established as well as the
alternative model, we see similar outcomes. The difference between both models is
small but speaks in favor of the alternative model. Hence, we find another rationale
for applying the CVPAT procedure.

1
For the established model values, see Saari et al. (2021).
Predicting Sustainable Consumption Behavior of Europeans Using the CVPAT 119

4 Results and Discussion

In our analysis, we consider the results from the CVPAT analysis by focusing on the
target construct, which is sustainable consumption behavior (see Table 1), and
choose our preferred model in accordance with the predictive power. A lower
average loss indicates better predictive capabilities of a model. This analysis entails
testing the difference in the average losses of the two alternative models.
Before comparing the established and the alternative model, we assess the
established model in relation to the naïve indicator-averages prediction benchmark
and the linear model prediction benchmark (Sharma et al. 2023). We find that the
model reaches the indicator averages benchmark, but not the linear model bench-
mark. Hence, we have a further indication that the model inherits predictive validity.
When comparing the established and the alternative model, we find a significant
positive value. Consequently, the alternative model shows higher out-of-sample
predictive power than the established model proposed by Saari et al. (2021). We
therefore conclude that an alternative model, which includes environmental concern
as a predecessor to risk perception, holds more predictive power considering the
target construct of sustainable consumption behavior.
We deem these results to provide a valuable direction for future research,
particularly for research related to future-oriented sustainable consumption behavior.
In this sense, the United Nations (Brundtland 1987) determined sustainable devel-
opment as a state, in which the needs of today’s generation are met, while this does
not go at the expense of future generations’ needs. However, making predictions
about the future is always an uncertain endeavor. Hence, it is of utmost necessity that
we provide models that are as accurate in their prediction as possible. The Covid-19
pandemic has changed what we previously had known about sustainable consump-
tion. Ali et al. (2021), for example, found that the perception of Covid-19 had a
significant and positive impact on people’s sustainable behavior, including sustain-
able consumption and environmental awareness.

5 Conclusions and Implications for Theory

We conclude that both presented models (i.e., Saari et al. 2021’s original model and
our proposed alternative model) have valid grounds in theory and show out-of-
sample predictive power. However, according to the CVPAT analysis, the alterna-
tive model is the preferred choice when considering predictive accuracy. This does
not mean that the model, which was established by Saari et al. (2021), should be
neglected, but it further needs to be tested, as our proposed alternative model needs
revalidation. Such an opportunity might be provided by the most recent ISSP
Environment IV dataset (ISSP Research Group 2022). The ISSP datasets in each
release are partial replications of versions from the same module in earlier years
(GESIS 2019), which provides a good opportunity for researchers to repeatedly test
120 L. Frömbling et al.

models. Another possibility would be to assess a model, which would represent both
directions in recursive relationships and would thereby help to understand sustain-
able consumption behavior even better. For this purpose, longitudinal data should be
collected. Moreover, further research could apply other novel approaches, such as
the necessary condition analysis (NCA; Dul 2016). The NCA makes it possible to
complement PLS-SEM with a focus on constructs’ sufﬁciency, by testing constructs
in their necessity for another outcome to exist (Richter et al. 2020). Thereby, the
NCA could provide opportunities for testing the abovementioned model versions
from a different angle in the future.

Acknowledgments This research uses the statistical software SmartPLS 4 (https://www.smartpls.

com). Christian M. Ringle acknowledges a ﬁnancial interest in SmartPLS.

References

Ali Q, Parveen S, Yaacob H, Zaini Z, Sarbini NA (2021) Covid-19 and dynamics of environmental
awareness, sustainable consumption and social responsibility in Malaysia. Environ Sci Pollut
Res 28(40):56199–56218
Bickerstaff K (2004) Risk perception research: socio-cultural perspectives on the public experience
of air pollution. Environ Int 30(6):827–840
Brundtland G (1987) Report of the World Commission on Environment and Development: our
common future. United Nations General Assembly Document A/42/427
Chin W, Cheah J-H, Liu Y, Ting H, Lim X-J, Cham TH (2020) Demystifying the role of causal-
predictive modeling using partial least squares structural equation modeling in information
systems research. Ind Manag Data Syst 120(12):2161–2209
Dul J (2016) Necessary condition analysis (NCA): logic and methodology of "necessary but not
sufficient" causality. Organ Res Methods 19(1):10–52
Fransson N, Gärling T (1999) Environmental concern: conceptual definitions, measurement
methods, and research findings. J Environ Psychol 19(4):369–382
GESIS (2019) ISSP 2010—Environment III, Variable Report: Documentation Release 2019/06/13,
Related to the International Dataset Archive-Study-No. ZA5500 Version 3.0.0, https://doi.org/
10.4232/1.13271. https://www.gesis.org Accessed 27 Sep 2022
Hair JF (2021) Next-generation prediction metrics for composite-based PLS-SEM. Ind Manag Data
Syst 121(1):5–11
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks, CA
ISSP Research Group (2012) ZA5500: International Social Survey Programme: Environment III—
ISSP 2010. https://www.gesis.org/issp/modules/issp-modules-by-topic/environment/2010/
Accessed 27 Sep 2022
ISSP Research Group (2022) ZA7650: International Social Survey Programme: Environment IV—
ISSP 2020. https://www.gesis.org/en/issp/modules/issp-modules-by-topic/environment/2020
Accessed 27 Sep 2022
Liengaard BD, Sharma PN, Hult GTM, Jensen MB, Sarstedt M, Hair JF, et al. (2020) CVPAT R
Code. https://github.com/ECONshare/CVPAT/ Accessed 27 Sep 2022
Predicting Sustainable Consumption Behavior of Europeans Using the CVPAT 121

Liengaard BD, Sharma PN, Hult GTM, Jensen MB, Sarstedt M, Hair JF et al (2021) Prediction:
coveted, yet forsaken? Introducing a cross-validated predictive ability test in partial least squares
path modeling. Decis Sci 52(2):362–392
Lima ML, Barnett J, Vala J (2005) Risk perception and technological development at a societal
level. Risk Anal 25(5):1229–1239
Pagiaslis A, Krontalis AK (2014) Green consumption behavior antecedents: environmental con-
cern, knowledge, and beliefs. Psychol Mark 31(5):335–348
R Core Team (2022) The R Project for Statistical Computing. https://www.r-project.org Accessed
27 Sep 2022
Richter NF, Schubring S, Hauff S, Ringle CM, Sarstedt M (2020) When predictors of outcomes are
necessary: guidelines for the combined use of PLS-SEM and NCA. Ind Manag Data Syst
120(12):2243–2267
Ringle CM, Wende S, Becker J-M (2022) SmartPLS 4. http://www.smartpls.com. Accessed
27 Sep 2022
Saari UA, Damberg S, Frömbling L, Ringle CM (2021) Sustainable consumption behavior of
Europeans: the inﬂuence of environmental knowledge and risk perception on environmental
concern and behavioral intention. Ecol Econ 189:107155
Sachs J, Schmidt-Traub G, Kroll C, Lafortune G, Fuller G, Woelm F (2020) Sustainable develop-
ment report 2020: The sustainable development goals and Covid-19. https://www.sdgindex.org/
reports/sustainable-development-report-2020/ Accessed 27 Sep 2022
Sarstedt M, Danks NP (2022) Prediction in HRM research—a gap between rhetoric and reality.
Hum Resour Manag J 32(2):485–513
Sarstedt M, Hair JF, Ringle CM (2023) “PLS-SEM: indeed a silver bullet”—retrospective obser-
vations and recent advances. J Mark Theory Pract 31(3):261–275
Sharma PN, Liengaard BD, Hair JF, Sarstedt M, Ringle CM (2023) Predictive model assessment
and selection in composite-based modeling using PLS-SEM: extensions and guidelines for
using CVPAT. Eur J Mark 57(6):1662–1677
Shmueli G, Ray S, Velasquez Estrada JM, Chatla SB (2016) The elephant in the room: evaluating
the predictive performance of PLS models. J Bus Res 69(10):4552–4564
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S et al (2019) Predictive model
assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Stern PC (2000) New environmental theories: toward a coherent theory of environmentally
signiﬁcant behavior. J Soc Issues 56(3):407–424
Part III
Hospitality and Tourism
The Relationship Among Perceived Value,
Tourist Satisfaction, and Citizenship
Behaviors: The Difference Between
Overnight Tourists and Non-overnight
Tourists

Yangpeng Lin, Di Zhang, and Hongfeng Zhang

1 Introduction

Before the current coronavirus disease 2019 (COVID-19), Macao’s per capita
overnight visitor rate had been lower than its non-overnight visitor rate. However,
due to the impact of COVID-19 and the preferential accommodation program
launched by Macao’s tourism and hospitality industry, the overnight visitor rate
per capita has increased signiﬁcantly, from 47.9% in 2020 to 49.6% now (Macao
Government Tourist Ofﬁce—MGTO 2021). And Macao government has been
trying to extend the stay time of tourists and has introduced a number of measures
to do so. However, there are certain differences in the depth of travel experience
between overnight tourists and non-overnight tourists, which may lead to differences
in their attitudes and behaviors toward destinations. And in tourism and hospitality
industry, value-attitude-behavior (VBA) model is found to be a good model to
predict consumer behavior (Kiatkawsin and Han 2017). Therefore, the VAB
model was used to provide valuable explanations for tourist behavior by referring
to relevant literature and to explore the relationship and difference between over-
night tourists and non-overnight tourists in value perception, satisfaction, and
citizenship behavior (Fig. 1).

Y. Lin
Faculty of Hospitality and Tourism Management, Macau University of Science and
Technology, Taipa, Macau, China
e-mail: [email protected]
D. Zhang · H. Zhang (✉)
Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 125

Fig. 1 Value-attitude-behavior model

The perceived value of destination experience is considered to be the key to

understanding tourist behavior (Prebensen et al. 2013), but few studies have con-
sidered the difference between the perceived value of overnight tourists and
non-overnight tourists. Studying the relationship between tourist perceived value
and tourist satisfaction with destination and their citizenship behaviors can provide
inspiration for destination marketing organizations and local authorities to create and
deliver value to tourists in the process of revitalizing and commercializing destina-
tion space. Theoretically, the research results will enrich relevant literature, and
compare the strength of the inﬂuence of perceived value of overnight tourists and
non-overnight tourists on satisfaction and citizenship behavior, ﬁlling the gap of
such research.
Overnight tourists and non-overnight tourists in Macao were selected as the
research objects to explore the dimension of customer perceived value as the
antecedent of tourist satisfaction with destination and citizenship behavior. So far,
no comparative study has been conducted on satisfaction and citizenship behavior by
using two different tourists’ value perception. Tourism scholars believe that different
types of tourists have different tourism behaviors (Gardiner et al. 2014) and point out
the need to conduct research on different types of tourists (Robinson and Schänzel
2019). In particular, the existing literature highlights the need to examine traveler
perceived values and subsequent attitudes and behaviors (Haddouche and Salomone
2018).
This is important to test the differences between overnight tourists and
non-overnight tourists in the value-attitude-behavior model (Fig. 2). And the ques-
tions included the following:
1. Is there a difference in multidimensional perceived value between overnight
tourist and non-overnight tourist?
2. Do overnight tourist perceived value and non-overnight tourist perceived value
have an impact on tourist satisfaction? Which of the two has a stronger impact on
satisfaction?
3. Do overnight tourist perceived value and non-overnight tourist perceived value
have an impact on tourist citizenship behavior (TBC)? Which of the two has a
stronger impact on citizenship behavior?
4. Does tourist satisfaction have an impact on tourist citizenship behavior?
The Relationship Among Perceived Value, Tourist Satisfaction,. . . 127

Fig. 2 Hypothesis model

2 Research Methods

To remain consistent with previous research, the measures will be taken or adapted
from previous tourism research. The questionnaire design will be based on a
multiple-item measurement scale, which has been validated in previous studies.
The sample is tourists who have had one or more trip experiences in Macao. After
an extensive literature review, we will develop a questionnaire to test our proposed
relationship outcome model. As of October 2021, tourists from the Chinese main-
land accounted for 91.1% of the total inbound tourists in Macao, ranking first in the
source market and maintaining it for many years (MGTO 2021). Therefore, the
questionnaire was mainly conducted in English and presented in both Chinese and
English to help some respondents better understand the contents.
Due to the outbreak of COVID-19 and the movement restrictions, online survey
will be deemed the most appropriate method to collect data. The online form contain
the cover page outlining the purpose of the research, assurance on data anonymity,
and confidentiality. Other than preventing physical contact with the respondents,
online survey facilitated data collection from a large number of individuals in a short
period of time (Braun et al. 2021). And the online survey had the advantages of fast
response, few missing items, and convenient statistical analysis (Zang et al. 2015).
After identifying the items for the questionnaire, a “personal management question-
naire” (data collected by the author) will be used to distribute and collect the
questionnaire. All questions must be answered to ensure that there was no bias.
Incomplete questionnaires and questionnaires providing the same answers for all
measurement items were excluded. And then, the pre-test will be carried out to
ensure the questionnaire would not have any comprehensibility issue.
Partial least squares structural equation modeling (PLS-SEM) analysis was be
used as the main method to evaluate the proposed hypothesis. PLS-SEM was useful
in marketing research due to its ability to perform all the needed analyses in a
structural manner and provide the predictive relevance of study (Fam et al. 2020).
PLS has several benefits for researchers when compared with other multivariate
models. For example, due to the partial nature of the PLS algorithm and principal
128 Y. Lin et al.

method, PLS requires minimal restrictions on the measurement scale, sample size,
and residual distribution (Tenenhaus et al. 2005). Thus, PLS is an appropriate way to
evaluate models containing complex relationships and many explicit variables
(Esposito Vinzi et al. 2010).
The tourist perceived value scale will be adapted from a scale developed by Jiang
and Hong (2021). The tourist satisfaction scale will be adapted from Suhartanto et al.
(2020). In addition, this study will examine the TCBs exhibited by the participating
tourists using the TCB scale proposed by Liu and Tsaur (2014).

Acknowledgments This work was supported by Macao Polytechnic University (RP/FCHS-02/

2022).

References

Braun V, Clarke V, Boulton E, Davey L, McEvoy C (2021) The online survey as a qualitative
research tool. Int J Soc Res Methodol 24(6):641–654
Esposito Vinzi V, Trinchera L, Amato S (2010) PLS path modeling: from foundations to recent
developments and open issues for model assessment and improvement. In: Esposito Vinzi V,
Chin WW, Henseler J, Wang H (eds) Handbook of partial least squares (Springer handbooks of
computational statistics). Springer, Berlin, pp 47–82
Fam KS, Ting H, Tan KL, Hussain K, Cheah JH (2020) Does it matter where to run? Intention to
participate in destination marathon. Asia Pac J Mark Logist 32(7):1475–1494
Gardiner S, Grace D, King C (2014) The generation effect: the future of domestic tourism in
Australia. J Travel Res 53(6):705–720
Haddouche H, Salomone C (2018) Generation Z and the tourist experience: tourist stories and use
of social networks. J Tour Futures 4(1):69–79
Jiang Y, Hong F (2021) Examining the relationship between customer-perceived value of night-
time tourism and destination attachment among generation Z tourists in China. Tour Recreat Res
48:1–14
Kiatkawsin K, Han H (2017) An alternative interpretation of attitude and extension of the value–
attitude–behavior hierarchy: the destination attributes of Chiang Mai, Thailand. Asia Pac J Tour
Res 22(5):481–500
Liu JS, Tsaur SH (2014) We are in the same boat: tourist citizenship behaviors. Tour Manag 42:88–
100
Macao Government Tourism Office—MGTO (2021) Global indicators. https://dataplus.
macaotourism.gov.mo/document/CHT/Report/GlobalIndicators/2021/1.%20Global%20Indica
tors%20%202021%20Oct.pdf Accessed 10 Jan 2022
Prebensen NK, Woo E, Chen JS, Uysal M (2013) Motivation and involvement as antecedents of the
perceived value of the destination experience. J Travel Res 52(2):253–264
Robinson VM, Schänzel HA (2019) A tourism inflex: generation Z travel experiences. J Tour
Futures 5(2):127–141
Suhartanto D, Brien A, Primiana I, Wibisono N, Triyuni NN (2020) Tourist loyalty in creative
tourism: the role of experience quality, value, satisfaction, and motivation. Curr Issue Tour
23(7):867–879
Tenenhaus M, Esposito Vinzi V, Chatelin YM, Lauro C (2005) PLS path modeling. Comput Stat
Data Anal 48(1):159–205
Zang M, Li J, Zang G (2015) Which is better? A comparative analysis of tourism online survey and
field survey. Tourism Tribune 30(4):95–104
Nonlinear Relationship Between Intellectual
Capital and Hotel Performance: Quadratic
Effect in PLS-SEM

Lorena Ruiz-Fernández, Bartolomé Marco-Lajara, Pedro Seva-Larrosa,

and Javier Martínez-Falcó

1 Introduction

Under knowledge-based view (KBV), the firm’s or manager’s abilities linked to the
creation and management of knowledge are key elements in explaining the sources
of competitive advantage (Kogut and Zander 1992). Nowadays, business and man-
agement scholars focus a great deal of attention on intangibles and knowledge,
which makes intellectual capital management a key task for companies (Baima
et al. 2020; Costa et al. 2020; Martín-de Castro et al. 2019).
Even though the popularity and awareness of these issues has increased dramat-
ically in recent decades (Serenko and Bontis 2013), intellectual capital research
remains fragmented and far from conclusive (Martín-de Castro et al. 2019).
These phenomena may be related to the underdevelopment of studies analyzing
U-shaped relationships in the field of strategic management (Haans et al. 2016).
Researchers fail to explain the relationships between different latent mechanisms
because they are not using adequate procedures to detect the presence of nonlinear or
U-shaped relationships in tourism (Rasoolimanesh et al. 2018). In addition, Martí-
nez-Martínez et al. (2020) have highlighted the necessity to focus on intellectual
capital research and practice related to tourism and hotels problems.
In addition, an ongoing trend of interest in the academic literature which is
particularly interesting involves manager’s gender as a key aspect in the manage-
ment of knowledge and hotel intangibles, and its resulting performance. Even
though gender equality has been defined as a main goal in the 2030 Agenda for
the sustainable development, parity is still far from having been achieved in

L. Ruiz-Fernández (✉) · B. Marco-Lajara · P. Seva-Larrosa · J. Martínez-Falcó

Business Management Department, Faculty of Economic and Business, University of Alicante,
Alicante, Spain
e-mail: [email protected]; [email protected]; [email protected];
[email protected]

© Springer Nature Switzerland AG 2023 129

leadership positions. Recently, authors such as Isola et al. (2020) have claimed that
women managers are better drivers of intellectual capital efficiency in companies.
Thus, it is also important to consider the essential moderator role that hotel gender
managers play in managing intellectual capital and influencing the hotel’s perfor-
mance (Li and Liu 2018). Therefore, the present research aims to analyze the
nonlinear influence of intellectual capital on hotel performance, also considering
the moderating effect of gender.

2 Methodology

To obtain the necessary information to provide empirical evidence, a survey has

been designed focused on managers of Spanish hotels. A questionnaire was distrib-
uted via Qualtrics online survey tool during the first half of 2020. Previously
validated Likert scales were used to measure the latent variables of our model.
Intellectual capital was operationalized on the basis of the 14-item scale proposed
by de Subramaniam and Youndt (2005). Perceived measures have been used to
operationalize firm performance with 8 items (traditional performance indicators and
specific hotel performance variables) following other hotel studies (Wilke et al.
2019). The contents of the questionnaire were validated using a pre-test with
4 experts in strategic hotel management and 3 hotel managers. After sending out
some 7000 emails and phone calls, the final sample was composed of 212 responses
checked with secondary information published.
Structural equation Model based on the variance (PLS-SEM) was used to link
data and theory in our research that includes moderator variables and nonlinear
effects relationships. The use of PLS is recommended since it can effectively manage
these aspects (Ali et al. 2018). This statistical technique has undergone massive
growth over the last decade (Hair et al. 2019), and it is now commonly applied in
several social science fields, in particular strategic management (Hair et al. 2012)
and hospitality management (Ali et al. 2018), areas of research under which our
work is framed.

3 Results and Discussion

Our results show that the quadratic effect of intellectual capital on hotel performance
is positive and statistically signiﬁcant (β = 0.156). This means that the relationship
between intellectual capital and hotel performance is U-shaped, so that, although
until reaching certain levels of accumulated intellectual capital in the hotel, it may
exert a negative inﬂuence on performance, in general terms and in the long term, its
effect is positive and becomes more strongly positive.
In addition, to know the size of the quadratic effect and to offer greater statistical
rigor to the presence of a nonlinear relationship, we observed the value of f 2. The
Nonlinear Relationship Between Intellectual Capital and Hotel Performance:. . . 131

value obtained in the case of the quadratic effect is 0.049 (> 0.025) which tradi-
tionally would indicate a moderate effect but in the case of nonlinear relationships
implies a large effect (Hair et al. 2021). Although our results are in line with the ones
presented in the systematic review of the literature by Pedro et al. (2018), who finds
the relationship between the three components of intellectual capital and perfor-
mance to be nonlinear, they are not in line with the findings of Haris et al. (2019), as
they found an inverse U-shaped relationship.
Taking into account the trend toward the introduction of moderation in quadratic
relations (Haans et al. 2016), we also cheek the moderating effect of manger gender
on the abovementioned relationship which is statistically significant (β = 0.132), and
the effect size is also large ( f2 = 0.026). That means that intellectual capital in hotels
runs by women has a higher effect on hotel performance than intellectual capital
managed by men. However, our findings shed light on the role of manager gender in
intellectual capital performance considering the gap raised previously in other
research (Giuliani and Poli 2019) where it is stated that the effects of manager
gender on intellectual capital performance were still unclear and needed further
research as the results achieved up to date did not converge. Nowadays, these
findings could also be found in emerging economics where female directors drive
intellectual capital performance (Smriti and Das 2021).

4 Conclusions

This research was aimed to explore the intellectual capital within a currently topical
sector, hospitality. Hotels were facing important challenges derived from the knowl-
edge society, but now they are immersed in a turbulent environment due to the
COVID-19 crisis. Under this situation, scholars need to develop empirical models to
test hypotheses that will help business managers to improve their resilience.
The turbulent context in which hotels compete today requires a manager with the
ability to combine and integrate intangible resources such as highly skilled
employees, partner relationships, and accumulated organizational knowledge. In
addition, we have justiﬁed the necessity of hotels to enhance the knowledge and
intangible resources embedded in their employees, routines, social relationships, and
so forth as one of their main targets in the search of strength to cope with the
environment.

References

Ali F, Rasoolimanesh SM, Sarstedt M, Ringle CM, Ryu K (2018) An assessment of the use of
partial least squares structural equation modeling (PLS-SEM) in hospitality research. Int J
Contemp Hosp Manag 30(1):514–538
Baima G, Forliano C, Santoro G, Vrontis D (2020) Intellectual capital and business model: a
systematic literature review to explore their linkages. J Intellect Cap 22(3):653–679
132 L. Ruiz-Fernández et al.

Costa V, Silva L, Paula L (2020) Intellectual capital and its impact on business performance: an
empirical study of Portuguese hospitality and tourism sector. Intang Cap 16(2):78–89
Giuliani M, Poli S (2019) Which relationship between gender diversity, intellectual capital and
financial performance? Int J Bus Manag 14(10):101–115
Haans RF, Pieters C, He ZL (2016) Thinking about U: theorizing and testing U-and inverted
U-shaped relationships in strategy research. Strateg Manag J 37(7):1177–1195
Hair JF, Sarstedt M, Pieper TM, Ringle CM (2012) The use of partial least squares structural
equation modeling in strategic management research: a review of past practices and recommen-
dations for future applications. Long Range Plan 45(5–6):320–340
Hair JF, Sarstedt M, Ringle CM (2019) Rethinking some of the rethinking of partial least squares.
Eur J Mark 53(4):566–584
Hair JF, Sarstedt M, Ringle CM, Gudergan SP (2021) Advanced issues in partial least squares
structural equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Haris M, Yao H, Tariq G, Malik A, Javaid HM (2019) Intellectual capital performance and
profitability of banks: evidence from Pakistan. J Risk Financ Manag 12(2):56
Isola WA, Adeleye BN, Olohunlana AO (2020) Boardroom female participation, intellectual capital
efficiency and firm performance in developing countries: evidence from Nigeria. J Econ Finance
Adm Sci 25(50):413–424
Kogut B, Zander U (1992) Knowledge of the firm, combinative capabilities, and the replication of
technology. Organ Sci 3(3):383–397
Li YQ, Liu CHS (2018) The role of problem identification and intellectual capital in the manage-
ment of hotels’ competitive advantage-an integrated framework. Int J Hosp Manag 75:160–170
Martín-de Castro G, Díez-Vial I, Delgado-Verde M (2019) Intellectual capital and the firm:
evolution and research trends. J Intellect Cap 20(4):555–580
Martínez-Martínez A, Cegarra-Navarro JG, Garcia-Perez A, Vicentini F (2020) Extending struc-
tural capital through pro-environmental behaviour intention capital: an outlook on Spanish hotel
industry. J Intellect Cap 22(3):633–652
Pedro E, Leitão J, Alves H (2018) Intellectual capital and performance: taxonomy of components
and multi-dimensional analysis axes. J Intellect Cap 19(2):407–452
Rasoolimanesh SM, Ali F, Jaafar M (2018) Modeling residents’ perceptions of tourism develop-
ment: linear versus non-linear models. J Destin Mark Manag 10:1–9
Serenko A, Bontis N (2013) The intellectual core and impact of the knowledge management
academic discipline. J Knowl Manag 17(1):137–155
Smriti N, Das N (2021) Do female directors drive intellectual capital performance? Evidence from
Indian listed firms. J Intellect Cap 23(5):1052–1080
Subramaniam M, Youndt MA (2005) The influence of intellectual capital on the types of innovative
capabilities. Acad Manag J 48(3):450–463
Wilke EP, Costa BK, Freire OBDL, Ferreira MP (2019) Interorganizational cooperation in tourist
destination: building performance in the hotel industry. Tour Manag 72:340–351
The U-Shape Influence of Family
Involvement in Hotel Chain: Examining
Dynamic Capabilities in PLS-SEM

Lorena Ruiz-Fernández, Laura Rienda, and Rosario Andreu

1 Introduction

Family businesses are a major part of the global business community, and their
development and evolution are fundamental to the success of the global economy.
These businesses face an increasingly dynamic and turbulent environment that
forces them to develop capabilities to adapt to changes and maintain their compet-
itive advantages. In particular, the analysis of the strategies followed by these
companies has attracted the attention of numerous researchers, both from a theoret-
ical and practical point of view, giving rise to the progressive construction of a rich
and exhaustive field of study (Sharma 2004; Kraus et al. 2011; Alayo et al. 2021).
Even though it is true that the relevance of family businesses extends to all sectors
of the economy, representing the largest group of companies in the world (Arregle
et al. 2019), the percentage of family businesses present in the Spanish hotel industry
is very high (Andreu et al. 2018), which underpins the importance of studying their
idiosyncratic characteristics in the development of dynamic capabilities that influ-
ence the innovative and economic performance of hotels.
In our case, we focus on the dynamic capabilities (DC) framework following the
idea of Park et al. (2019), who state the need to use the theory DC as a frame of
reference to research how family businesses develop their capabilities and how they
affect performance. In addition, despite the fact that DC have been considered the
basis of firms surviving in a dynamic environment, there has been little research on
understanding the DC of family firms (Wang 2016).

L. Ruiz-Fernández (✉) · L. Rienda · R. Andreu

Business Management Department, Faculty of Economic and Business, University of Alicante,
Alicante, Spain
e-mail: [email protected]; [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 133

Considering recent research that suggested linear relationships do not always

explain the business reality well (Sarstedt et al. 2020; Ahrholdt et al. 2019) and the
complexity and uniqueness of the underlying relationships between familiarity and
the tourism enterprise, as well as the contradictory results that can be found in the
empirical literature (Miller and Le Breton-Miller 2006; García-Castro and Aguilera
2014; De Massis et al. 2015), we propose to assess whether there is a nonlinear
relationship between family involvement and the development of dynamic capabil-
ities of sensing, capturing opportunities, and reconﬁguring their resources.
Given the need to take into account the perceptions, attitudes, and intentions of
those involved in order to achieve a deeper understanding in this area, our method
has been based on data obtained from a single survey of Spanish 3–5-star hotel
chains with and without family inﬂuence (Marco-Lajara et al. 2021). To model and
test the underlying relationships, we have used the PLS technique based on structural
equation modelling (Wilson et al. 2014).

2 Methodology

The final sample was composed by 107 hotel chains located in Spain with 3- to 5-star
hotels. We addressed a survey to managers to ask them about the constructs under
study using previously validated Likert measurement scales, with response ranges
from 1 = strongly disagree to 7 = strongly agree. Family management was measured
with the percentage of family member in management positions. Dynamic capabil-
ities have been measured according to Teece’s (2007) proposal for empirical data
analysis (second-order variable created by sensing, seizing, and reconfiguring).
The collected data were subsequently analyzed with SmartPLS v3.3 (Hair et al.
2022), using the technique of variance-based partial least squares structural equation
modelling (PLS-SEM). This methodology allows representing, estimating, and
testing a theoretical network of linear relationships between variables that may be
unobserved, i.e., latent variables such as family involvement or dynamic capabilities,
and has been defined as a suitable means to estimate non-linear effects in the field of
family business (Basco et al. 2022).

3 Results and Discussion

Nonlinear (quadratic) effects involve an interaction of Y1 with itself (Y1*Y1), so

they can be understood as a special case of a moderation model in which the
relationship between Y1 and Y2 is moderated by Y1 -self-moderation (Hair et al.
2021). Thus, in line with Rigdon et al. (2010), we can model self-moderation using
an interaction term, and the same approaches can be used as with moderation effects
(Hair et al. 2022), although the 2-stage approach is recommended. After introducing
the quadratic effect as an interaction term, we explain the results from bootstrapping
The U-Shape Inﬂuence of Family Involvement in Hotel Chain:. . . 135

based on 10,000 subsamples, the percentile approach and the “unsigned changes”
option (Hair et al. 2022; Streukens and Leroi-Werelds 2016).
Both the linear relationship (β estimated: -0.241) and the quadratic term (β
estimated: 0.267) are statistically significant, implying a U-shaped relationship
between family involvement and the development of dynamic capabilities. How-
ever, a significant coefficient for the quadratic term is necessary, but not sufficient
according to Haans et al. (2016). Therefore, following Hair et al. (2021), we test the
size of the f 2 effect to assess the impact of the interaction term on the R2 and test the
relevance of this curvilinear relationship. Since the f 2 of the quadratic effect was
0.043, as suggested by Ghasemy et al. (2020) or Hair et al. (2021) following Kenny
(2018), we concluded that the effect size was large, implying a high relevance of the
significant nonlinear relationship between family address and the deployment of
dynamic capabilities.
These results are in line with recent calls for papers such as Basco et al. (2022)
and respond to their call to shed more light on nonlinear relationships in the field of
family business.

4 Conclusions

In line with the most burning question in the ﬁeld of family business research, and
the subject of study of works such as Daspit et al. (2019), our work allows us to delve
into topical questions in the science of family business: How does family involve-
ment in Spanish hotel chains affect results?
The above leads us to extend a call to researchers in the ﬁeld of family business to
delve deeper into the relationships analyzed to bring to light more complex relation-
ships that allow us to explain the business reality in a more profound way. It is not
enough to detect mediations or moderations; it is also important to know the
U-shaped or inverse U-shaped relationships. This consideration may be latent and
would allow a better understanding of the particularities of the development and
strategic management of family businesses through the view of dynamic capabilities.

References

Ahrholdt DC, Gudergan SP, Ringle CM (2019) Enhancing loyalty: when improving consumer
satisfaction and delight matters. J Bus Res 94:18–27
Alayo M, Iturralde T, Maseda A, Aparicio G (2021) Mapping family ﬁrm internationalization
research: bibliometric and literature review. Rev Manag Sci 15(6):1517–1560
Andreu R, Claver-Cortés E, Quer D, Rienda L (2018) Family ownership and Spanish hotel chains:
an analysis of their expansion through internationalization. UCJC Bus Soc Rev 59:40–75
Arregle JL, Hitt MA, Mari I (2019) A missing link in family ﬁrms’ internationalization research:
family structures. J Int Bus Stud 50(5):809–825
136 L. Ruiz-Fernández et al.

Basco R, Hair JF Jr, Ringle CM, Sarstedt M (2022) Advancing family business research through
modeling nonlinear relationships: comparing PLS-SEM and multiple regression. J Fam Bus
Strat 13(3):100457
Daspit JJ, Long RG, Pearson AW (2019) How familiness affects innovation outcomes via absorp-
tive capacity: a dynamic capability perspective of the family firm. J Fam Bus Strat 10(2):
133–143
De Massis A, Kotlar J, Campopiano G, Cassia L (2015) The impact of family involvement on
SMEs’ performance: theory and evidence. J Small Bus Manag 53(4):924–948
García-Castro R, Aguilera RV (2014) Family involvement in business and financial performance: a
set-theoretic cross-national inquiry. J Fam Bus Strat 5(1):85–96
Ghasemy M, Teeroovengadum V, Becker J-M, Ringle CM (2020) This fast car can move faster: a
review of PLS-SEM application in higher education research. High Educ 80:1121–1152
Haans RF, Pieters C, He ZL (2016) Thinking about U: theorizing and testing U-and inverted
U-shaped relationships in strategy research. Strateg Manag J 37(7):1177–1195
Hair JF, Sarstedt M, Ringle CM, Gudergan SP (2021) Advanced issues in partial least squares
structural equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Hult T, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural equation
modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks, CA
Kenny DA (2018) Categorical moderator and continuous causal variable. http://davidakenny.net/
cm/moderation.htm Accessed 14 Mar 2021
Kraus S, Harms R, Fink M (2011) Family firm research: sketching a research field. Int J Entrep
Innov Manag 13(1):32–47
Marco-Lajara B, Ruiz-Fernández L, Seva-Larrosa P, Sánchez-García E (2021) Hotel strategies in
times of COVID-19: a dynamic capabilities approach. Anatolia 33(4):525–536
Miller D, Le Breton-Miller I (2006) Family governance and firm performance: agency, stewardship,
and capabilities. Fam Bus Rev 19(1):73–87
Park HY, Misra K, Reddy S, Jaber K (2019) Family firms’ innovation drivers and performance: a
dynamic capabilities approach. J Fam Bus Manag 9(1):4–23
Rigdon EE, Ringle CM, Sarstedt M (2010) Structural modeling of heterogeneous data with partial
least squares. In: Malhotra NK (ed) Review of marketing research (review of marketing
research), vol 7. Emerald Group Publishing Limited, Bingley, pp 255–296
Sarstedt M, Ringle CM, Cheah JH, Ting H, Moisescu OI, Radomir L (2020) Structural model
robustness checks in PLS-SEM. Tour Econ 26(4):531–554
Sharma P (2004) An overview of the field of family business studies: current status and directions
for the future. Fam Bus Rev 17(1):1–36
Streukens S, Leroi-Werelds S (2016) Bootstrapping and PLS-SEM: a step-by-step guide to get more
out of your bootstrap results. Eur Manag J 34(6):618–632
Teece DJ (2007) Explicating dynamic capabilities: the nature and microfoundations of (sustainable)
enterprise performance. Strateg Manag J 28(13):1319–1350
Wang Y (2016) Environmental dynamism, trust and dynamic capabilities of family businesses. Int J
Entrep Behav Res 22(5):643–670
Wilson SR, Whitmoyer JG, Pieper TM, Astrachan JH, Hair JF, Sarstedt M (2014) Method trends
and method needs: examining methods needed for accelerating the field. J Fam Bus Strat 5(1):
4–14
Part IV
Innovation Management
and Entrepreneurship
Two-Way Business Innovation in Central
Eastern Europe: Analysing Innovative
Enterprises Using the PLS-SEM Method

Márton Gosztonyi

1 Introduction

Innovation is a complex phenomenon. Innovation recognizes both the ﬁrst light bulb
invented by T.A. Edison in 1879 and new product packaging (Podręcznik 2005).
Although the innovation comes from the Latin word innovatis, meaning renewal,
creating something new (Tokarski 1980), nowadays it is mostly used as a synonym
for change (Janasz and Kozioł 2007). Enterprises are complex drivers of economic
growth (Acs and Szerb 2007; Audretsch and Keilbach 2007; Audretsch and Peña-
Legazkue 2012; Audretsch et al. 2006; Audretsch et al. 2008; Fritsch 2008; Noseleit
2013; Spencer et al. 2008); thus, enterprises’ ability to innovate is a key factor in
stimulating economic growth in Central Eastern Europe (CEE) (Acs and Szerb 2007;
González-Pernía and Peña-Legazkue 2015; Wong et al. 2005). Consequently, the
study of innovation has always played a key role in business research. Recently,
however, an even greater increase in research on innovative or productive enterprises
has been noted (Audretsch and Keilbach 2007; Baumol 2010; Davidsson 2006;
Samuelsson and Davidsson 2009). However, most studies explore entrepreneurial
innovation in developed economies; thus, we know little about the extent to which
enterprises in CEE economies implement innovation-driven entrepreneurship. To
this end, in our paper, we examine the factors that contribute to the emergence of
innovation among enterprises in the context of a CEE economy based on two main
theories: the classical Schumpeterian theory and the enterprises’ productive-side
effects theory.
According to the literature, CEE companies can be characterized by many
limitations in relation to innovative activity. Since this kind of activity involves a
big risk (Gorzeń-Mitka 2013), a high ﬁnancial commitment, and requires adequately

M. Gosztonyi (✉)
Universiti Malaya, Asia-Europe Institute, Kuala Lumpur, Malaysia

© Springer Nature Switzerland AG 2023 139

educated and creative staff, innovation often goes beyond the capacities of CEE
companies (Okręglicka 2014). However, the researchers also draw attention to the
fact that a strengthening process seems to be emerging among those enterprises
which products or services are based on the innovation-based demand of their
customers (Sachpazidu-Wójcicka 2017; Latusek-Jurczak 2014). Another barrier in
the CEE market is the creation of interorganizational linkages and different kinds of
cooperative relationships in order to create a competitive advantage, such as inno-
vation networks (Latusek-Jurczak 2014). Therefore, CEE companies’ owners, man-
agement and employee’s perspective of innovation shows an internal contradiction
that, on the one hand, affirms innovation as the primary source of income but, on the
other hand, shows significant risk due to the necessity of assigning additional
expenses for this purpose (Szopik-Depczyńska 2015; Zygmunt 2017; Lewandowska
and Stopa 2018). Semi-peripheral countries are countries with strong entrepreneurial
intensity (de Noronha Vaz and Nijkamp 2009), as incomes are relatively low, so an
entrepreneurial career is more favourable to earn a decent income. A growing trend
suggests that knowledge-based economy facilitate the flow of innovations within
and between semi-peripheral, centre and peripheral regions and are key sources of
innovation and economic growth in these countries (Huggins and Izushi 2007;
Huggins and Johnston 2009; Grossman and Helpman 1994; Harris 2001; Ibert
2007; Zucker et al. 2007; Romer 1986; Glaeser 1999; Ikeda 2008). For this reason,
it is essential to analyse semi-peripheral countries in innovation research, as their
interaction with centre and peripheral countries outlines the process of innovation in
the developed countries as well.
Linking innovation and business is, by no means, a new idea. The classical
economics theories of enterprise have all given prominence to the concept of
innovation. Perhaps one of the most influential ones, Schumpeter’s (1934) and
Weber’s (1978 [1921–1922]) entrepreneurial theory has defined innovation as a
central concept of entrepreneurs. It creates ‘creative destruction’, which can under-
pin long-term economic growth. Innovation, in this sense, is a necessary condition
for the action of the ‘cultural innovator’, that is, the entrepreneur. The causal chain
leading to business innovation in this theory clearly leads from the entrepreneur to
the innovation. In these theories, the understanding of the entrepreneur’s personality,
innovation attitude and psychological and sociological characteristics are key ele-
ments for the creation of business innovation. Innovation is, thus, assumed to be the
key function of the sociological archetype of an entrepreneur. Kirton’s (1987) high-
impact theory explores the styles, cognitive elements of creativity, decision-making
and problem-solving processes and outlines the profile of an innovative entrepre-
neur. In our research, we use these theories, as we emphasise the entrepreneur’s
personality as playing a key role in innovation. Thus, we focus on what visions can
be related to entrepreneurs, how they can be placed in a creative-cognitive context,
what their incentives are, why they started their business, how they perceive
problems and how they respond to them.
Contradicting Schumpeter’s theory, Baumol (1968) did not interpret business
innovation from the demand side of entrepreneurs but from the supply side of
enterprises and from the competition among them. According to this theory,
Two-Way Business Innovation in Central Eastern Europe:. . . 141

productive enterprises are the ones primarily encouraging entrepreneurs to innovate

and not the other way around. Moreover, in Kirzner’s (1973) theory, the exploration
of opportunity as a source of innovation is a central element of entrepreneurship.
This theoretical direction reverses the chain of causality and designates the innova-
tion environment as an entity that leads to the development of the innovative
entrepreneur. Thus, business innovation occurs mainly in firms that combine mul-
tiple abilities, knowledge, resources and skills (Fagerberg 2006). Higher levels of
economic activity create new business opportunities, which means that entrepre-
neurs may become interested in entering new markets and exploiting new business
opportunities by creating more competitive products (Hayton 2005). Therefore, the
enterprise’s economic activity positively affects innovation (Zotto and Gustafsson
2008).
Although at first glance a contradiction can be discovered between the two
theoretical directions, we will show in our paper that the two opposite theories can
drive the innovation on the market at the same time. In our paper, we work with a
complex definition of entrepreneurial innovation: business innovation in our
research is a successful realisation and materialisation of something new; this can
refer to gradual, radical or revolutionary changes in techniques, products, processes,
ways of thinking or organisations (Mckeown 2008; Taylor and Schroeder 2003).
Thus, innovation is about creating positive change, which reduces costs and maxi-
mises productivity (Taylor and Schroeder 2003). Therefore, we focus on technolog-
ical innovation, technology change and technology development. Furthermore, the
emergence of entrepreneurial innovation is assumed to be a three-sided effect in our
paper. It is assumed, in line with the classical Schumpeterian theory, as an effect
from the entrepreneur side, as an effect from the enterprise itself and as a result of
socio-economic mechanisms (environment).
Since the 2000s, certain segments of innovation research that emphasise the role
of enterprises in the innovation process have begun to build their theories on the
typology of the sustainable enterprise (Hockerts and Wüstenhagen 2010). Indeed,
the market system has seen the emergence of sustainable development companies
that have created innovative techniques, products or processes that drive environ-
mental or social goals and succeed in the consumer market. In these enterprises,
environmental and social goals are an integral part of these businesses’ economic
strategy (Dacin et al. 2002; DiMaggio 1988; Holm 1995; Ostrom 1990). Behind the
achievement of social goals and the emergence of entrepreneurial activities for
environmental development are specific market failures that firms seek to address
through innovation (Cohen and Winn 2007; Cohen et al. 2008). Agreeing with
Stevenson and Jarrillo-Mossi (1986), we believe that the emergence of innovative
enterprises is not merely an entrepreneurial skill or exclusively an attribute of a
particular enterprise, but a process. Hence, the contextual role of culture and the
socio-economic-environmental context must be included in the analysis of innova-
tion to have a systemic understanding of innovative enterprises in semi-peripheral
markets. However, the inclusion of the entire macro-context in the analysis is not an
easy task due to its complexity. In our research, we create a simulation of the macro-
context, in which we examine the effect of the macro-context on innovation only
142 M. Gosztonyi

through cognitive, contextual beliefs based on Bandura (1986) and Bruton

et al. (2010).
The triple (entrepreneur–enterprise–environment) theoretical impact mechanism
is an appropriate context for exploring business innovation in CEE countries and can
highlight the direction of causal chains that primarily determine the development of
innovation.

2 Methodology

Our research fits into the methodological framework of process tracing, which is
primarily an inductive methodological approach useful in theory development
(Bennett and Checkel 2015). Our theoretical framework includes hypothetical,
interrelated events and processes that require the (re)constitution of agents through
social and economic structures, and an analysis of cognitive processes and process
tracing was performed to examine these complex mechanisms of action (Blatter and
Haverland 2012). Our process tracing analysis was performed using partial least
squares-structural equation modelling (PLS-SEM).
PLS-SEM is a method of analysing complex systems; it provides a system-wide
interpretation (Hair et al. 2019; Hair et al. 2022). On the one hand, from a structural
analysis point of view, it approaches the data points with substantive hypothesis
testing to explore the relationships between exogenous and endogenous latent vari-
ables, and on the other hand, it builds on and analyses system theory at the system
level (Hair et al. 2016; Sarstedt et al. 2021; Basco et al. 2022). When examining
complex systems such as business innovations (Byrne and Callaghan 2013), exam-
ining the explanatory variables separately is insufficient; analysing them at the
system level would be more worthwhile. The effects in the model of business
innovations form a nonlinear (Nicolis 2012), third-order system (Deacon 2007),
which is far from the equilibrium (Reed and Harvey 1996) and has multiphase
correlations (DeLanda 2005) that can be described by autopoiesis (Maturana and
Varela 1980), structure, hierarchy (Cilliers 2001) and control parameters, respec-
tively. Therefore, being a multivariate technique that can capture latent dimensions
and examine their combined effects at a systemic level, SEM is an appropriate
method for researching business innovations in developing economies (Dijkstra
and Henseler 2015). It allows the simultaneous examination of the whole set of
equations and, where appropriate (e.g., when evaluating interactions), the correction
of errors in the equations, as it simultaneously estimates the model parameters and
the fit of the model (Johnson and Sohi 2014).
PLS-SEM seeks to maximise the full explained variance of endogenous con-
structs/indicators when estimating model parameters (Hair et al. 2017; Reinartz et al.
2009). It treats the constructs as a complex system and uses the full variance to
estimate the parameters of the model and not to explain it. PLS-SEM is a two-layer
modelling process. The first layer (measurement model) consists of the latent
generated from the measured variables and the variables to be explained defined in
Two-Way Business Innovation in Central Eastern Europe:. . . 143

the hypothesis. Thus, the first layer captures the relationship between the manifest
and the latent variables. The creation of this model can be reflective or formative
(Sarstedt et al. 2022). 1 The second layer (structural model) identifies the causal
relationships between the latent variables. The evaluation of the PLS-SEM measure-
ment models is based on the value of multicollinearity (variance inflation factor
(VIF)) developed for determining the reliability of internal consistency, the reliabil-
ity of indicators (average variance extracted (AVE)), the convergence validity and
the heterotrait-monotrait ratio (HTMT) (Rasoolimanesh 2022). In the case of reliable
and valid measurement models, it is possible to move on to the evaluation of the
structural model, for which several metrics were used as well. In the models, we
measured the direct, indirect and total effects, and their strength is expressed as the
standardised path coefficients (βs) and their significance (Hair et al. 2017). Path
coefficients range from -1 to +1, where higher absolute values indicate stronger
(predictive) relationships between constructs. The commonly used p-value and the
explanatory force f 2 were also used to evaluate the model. Perhaps the best-known
statistic to quantify the magnitude of the prediction error for the whole model is the
root mean square error of predictions (RMSE), but this measure is rarely used for
PLS-SEM. Rather, the PLS-SEM methodology uses the standardised root mean
square residual (SRMR) and the Bentler and Bonett index, also known as the normed
fit index (NFI). SRMR converts both the sample covariance matrix and the theoret-
ical covariance matrix into correlation matrices. The SRMR allows the measurement
of the differences between the average magnitude of observed and expected corre-
lations as an absolute measurement for the model’s fit criterion.

3 Sample and Variables

Our research sample is based on the 2021 Hungarian and Polish Global Entrepre-
neurship Monitor (GEM) data. Each sample was developed by random multistage
stratiﬁed sampling, as a result of which the samples are representative of the regional
location of Hungarian and Polish small- and medium-sized enterprises (SMEs) and
the age and gender of Hungarian and Polish entrepreneurs. The Hungarian sample
constituted 2016 and the Polish 8000 responses. However, for the present analysis,
we further narrowed our sample to those respondents who had businesses or planned
to start a business. Therefore, we worked with a sample of 366 responses in the
Hungarian dataset and 1037 in the Polish one, after data cleansing.
In our analysis, we sought a causal explanation of the enterprises implementing
innovation, for which we used the space stretched by 31 measured variables
(Table 1). The 31 measured variables plotted eight latent variables, which were
grouped according to the theoretical categories (entrepreneur–enterprise–

1
The reﬂective approach seeks to maximize overlap between indicators, while the formative model
seeks to minimize it.
144 M. Gosztonyi

Table 1 Measured and latent variables included in the analysis

Theory-
Variable based
name Question Latent variables Abbreviation categories
sumotiv1 To make a difference in the Reason of ROB Individual/
world business innovator
sumotiv2 To build great wealth or have context
a very high income
sumotiv3 To continue a family tradition
sumotiv4 To earn a living because jobs
are scarce
creativ Other people think you are Creative CV
highly innovative visionary
oppism You rarely see business
opportunities, even if you are
very knowledgeable in the
area
proact Even when you spot a proﬁt-
able opportunity, you rarely
act on it
vision Every decision you make is
part of your long-term career
plan
susdg_env When making decisions about Environmental BDES
the future of your business, and social impact
you always consider environ- of business
mental implications such as decisions
preservation of green areas,
reduction of the emission of
pollutants and toxic gases,
selective garbage collection
and conscious consumption
susdg_pri You prioritize the social
and/or environmental impact
of your business above prof-
itability or growth
susdg_soc When making decisions
about the future of your busi-
ness, you always consider
social implications such as
access to education, health,
safety, inclusive work, hous-
ing, transportation, quality of
life at work, etc.
omcstnat Do you have any customers in Crisis as an CO Context of
the following locations? Else- opportunity business
where in your country? innovations
sucpgovres In your country, the govern-
ment has so far effectively
responded to the economic
consequences of the corona-
virus pandemic
(continued)
Two-Way Business Innovation in Central Eastern Europe:. . . 145

Table 1 (continued)
Theory-
Variable based
name Question Latent variables Abbreviation categories
sucpnewopp The coronavirus pandemic
has provided new opportuni-
ties that you want to pursue
with this business
sucptech1 In response to the coronavirus
pandemic, has your business
made any changes in its use of
digital technologies for sell-
ing your product or service?
sucptech2 Do you expect your business
will use more digital technol-
ogies to sell your product or
service in the next 6 months?
omnewprod Are any of your products or An enterprise that EOIPS
services new to people in the offers an innova-
area where you live, people in tive product or
your country or the world? service
suacts Over the past 12 months, have
you done anything to help
start this new business?
suown Will you personally own all,
part or none of this business?
suowners How many people, including
yourself, will both own and
manage this new business?
suwage Has the new business paid
any salaries, wages or pay-
ments in kind, including your
own, for more than 3 months?
omcrgrow And compared to 1 year ago, Owned interna- OITB
are your expectations for tional turbulent
business growth much lower, business
somewhat lower, about the
same as a year ago, somewhat
higher or much higher?
omcstexp Do you have any customers in
the following locations? Out-
side your country?
omown Do you personally own all,
part or none of this business?
omrstart Did you start this business?
(continued)
146 M. Gosztonyi

Table 1 (continued)
Theory-
Variable based
name Question Latent variables Abbreviation categories
equalinc In my country, most people Observation of OB Socio-eco-
would prefer that everyone the business nomic
had a similar standard of living context context
nbgoodc In my country, most people
consider starting a new busi-
ness a desirable career choice
nbmedia In my country, you will often
see stories in the public media
and/or the Internet about suc-
cessful new businesses
nbsocent In my country, you will often
see businesses that primarily
aim to solve social problems
nbstatus In my country, those suc-
cessful at starting a new busi-
ness have a high level of
status and respect
easystart Subjective perception of the Subjective per- SC
social context ception of the
social context
omnewproc Are any of the technologies or Innovative IT Innovation
procedures used for this prod- technology
uct or service new to people in
the area where you live, the
people in your country or the
world?

environment). Based on this, we measured the effect of individual entrepreneurial

innovation with three latent variables, the effect of entrepreneurship with three latent
variables and the effect of the cognitive macroenvironment on the development of
innovation with one latent variable. We measured the effect of the entrepreneur on
innovation, as well as the extent to which the goal of the entrepreneur was to change
the world and achieve high income and wealth or continue the family tradition
(ROB), the extent to which the entrepreneur can be considered creative and visionary
(CV) and the extent to which the entrepreneur makes decisions based on sustain-
ability goals (BDES). We measured the effects of the company on the innovation, as
well as the extent to which the company sees a crisis as an opportunity (CO), if the
company has a new product or service or if it has renewed itself in some other way in
the past year (EOIPS), and whether the company is on a turbulent, growing trajectory
economically and has international connections (OITB). We measured the effect of
the macro-context from the cognitive side, and we considered in the analysis whether
the entrepreneur is convinced that the media give a prominent role to entrepreneurs
Two-Way Business Innovation in Central Eastern Europe:. . . 147

in the socio-economic context and whether the entrepreneur is convinced that the
businesses in the socio-economic environment address social problems (OB). As an
outcome variable, we used the emergence of innovative technology (IT), which
embodies business innovation, as it includes whether a particular enterprise has a
new technology, process, product or service.
The measured variables were characterised as having a small, random amount of
missing data in each sample (3–5%). Therefore, the missing data were replaced with
the artiﬁcial intelligence-based bagging methodology (Friedman and Popescu 2003).
Of the sampled enterprises, in Hungary 18.7% have implemented some form of
technological innovation (omnewproc), while this was only 7.3% in Poland, and
only 15.1% of enterprises have developed a new product or service (omnewprod) in
Hungary and only 6.3% in Poland. This means that only about one-ﬁfth of busi-
nesses are innovative businesses in Hungary and only one-eleven in Poland. These
innovative businesses are young businesses in both country (HU: 38.3%, PL:
37.4%). Surprisingly, most of them operate in the low-tech sector, in the retail or
in the professional service sector. Innovative enterprises operate mostly in the capital
and county capitals, as well as in the developed regions of the countries. There is a
particularly high proportion of men and those who have a high level of education
among their CEOs.

4 Hypotheses

In our research, we aim to determine the factor effects that distinguish innovative
companies in the emerging market of tow CEE countries (Hungary and Poland) from
the needs-based companies. Based on the literature, we assume a triple-side frame-
work for our innovation model in which the impact of the individual, that is,
the entrepreneur, the impact of business performance and the macro-context
(i.e. the cognitive socio-economic environment) are identified as factors that affect
the emergence of innovation. Thus, our hypothesis system explores the relationship
between entrepreneurship and innovation in a semi-peripheral country. Furthermore,
this system of hypotheses also allows us to explore the cause-effect relationships
between the three factors, so we can answer which theoretical trend explains why
businesses create innovations.
We formulated 17 hypotheses, which are presented in Table 2. Each hypothesis
assumes the presence of a particular mechanism of effect among the latent variables.
The system of these mechanisms of effects outlines the business innovation model
operating in the developing country, which is shown in Fig. 1.
In the PLS-SEM-based model 2 of innovative enterprises, we used first-order
latent variables both counties’ model, which are indicated by circles in Fig. 1. The
models were created from a total of eight first-order latent variables, each of which is

2
The analysis was performed with the SmartPLS 3v software (Ringle et al. 2015).
148 M. Gosztonyi

Table 2 Description of hypotheses

No. Description
H1 Crisis as an opportunity has a positive impact on the development of innovative technology
H2 Crisis as an opportunity has a positive impact on business decisions
H3 Crisis as an opportunity has a positive impact on reason for business
H4 Observation of the business context has a positive impact on the reason for business
H5 Observation of the business context has a positive impact on creative visionary
H6 Observation of the business context has a positive impact on the development of innova-
tive technology
H7 Observation of the business context has a positive impact on crisis as an opportunity
H8 Observation of the business context has a positive impact on business decisions
H9 Reason for business has a positive impact on the development of innovative technology
H10 Reason for business has a positive impact on an enterprise that offers an innovative product
or service
H11 Creative visionary has a positive impact on business decisions
H12 Creative visionary has a positive impact on an enterprise that offers an innovative product
or service
H13 Innovative technology has a positive impact on creative visionary
H14 Business decisions has a positive impact on the enterprise that offers an innovative product
or service
H15 Business decisions has a positive impact on the development of innovative technology
H16 Business decisions has a positive impact on an owned international turbulent business
H17 An enterprise that offers an innovative product or service has a positive impact on an
owned international turbulent business
H18 An enterprise that offers an innovative product or service has a positive impact on the
development of innovative technology
H19 An enterprise that offers an innovative product or service has a positive impact on the crisis
as an opportunity
H20 An owned international turbulent business has a positive impact on the development of
innovative technology
H21 Subjective perception of the social context has a positive impact on the development of
innovative technology
H22 Subjective perception of the social context has a positive impact on business decisions
H23 Subjective perception of the social context has a positive impact on reason for business
H24 Subjective perception of the social context has a positive impact on the observation of the
business context

built into the base models as a reflective measurement model. Although it is worth to
mention that the composite reliability of the measured variables of the reflective
models was above the 0.095 threshold in three cases in the Polish model, these
variables may not avoid redundancy (see Table 6 in Appendix). The AVE values for
the constructs are also higher than the cut-off value (0.50); thus, the criterion for the
convergence validity of the models is met. As conceptually similar indicators have
been included in the measurement models, the HTMT values must be less than 0.90.
All cases fulfil this requirement, which implies that the reflective models also meet
the discriminant criterion. Table 7 in the Appendix summarises the indicator
Two-Way Business Innovation in Central Eastern Europe:. . . 149

Fig. 1 The system of the research hypotheses, the measured variables and the latent variables of the
PLS-SEM model of innovation

loadings and collinear statistics (VIF) of the measured variables in each country. In
our models, the VIF values for the measured variables are adequate in all cases
(above 1.0), which indicates that collinearity is absent. Furthermore, reﬂective
indicator loadings above 0 are also suitable. 3

3
In our study, to compare Hungarian and Polish data, it would be obvious to carry out a multigroup
analysis (Matthews 2017). Although, we chose another method for the comparison, because we
think that the construction of the latent variables depends to a large extent on the given sociocultural
context, and they may show differences between the two countries. As a result of all this, we used
the same model that was described at the hypotheses part in our paper as a starting point for both
countries; however, the final latent variables were then developed taking into account the specifics
of the given country. As a result of this, the latent variables in the two models do not contain the
same measured variables in all cases, but were formed along the lines of what kind of latent variable
can be formed from the basic set of variables in the given country. Consequently, it was not possible
to carry out a multigroup analysis during the comparison, because the models of the two countries
do not match perfectly. Consequently, to facilitate easier understanding, we did not include the SC
latent variable in the Hungarian figures, because a latent variable matching the model could not be
formed based on the measured variables.
150 M. Gosztonyi

A 5000-sample bootstrap method 4 was used to test the significance of the path
coefficients. The SRMR of the models is around 0.10 (HU: 0.09, PL: 0.10), which is
a good fit (Hu and Bentler 1998), and the NFI is around 0.900 (HU: 0.912, PL:
0.900), which indicates an acceptable fit (Lohmöller 1989) in the models. Our
results, therefore, show that the models fit the statistical criteria.
The t-values for the structural model are shown in Fig. 2 and Table 3. The table
and the figure show only the significant path coefficients.
Based on our significant models (Fig. 3), business innovation is shaped by all
three dimensions (entrepreneur, enterprise and environment) in Hungary and in
Poland as well, but in a different casual system (Table 4).
In Hungary as a direct effect, technological innovation is affected by the latent
variables of environmental and social impact of business decisions (β = 0.11), an
innovative product or service (β = 0.46) and the possession of an owned interna-
tional turbulent business (β = 0.21). However, to realise these direct effects, a
number of other effects must be also present. Environmental and social impact of
business decisions should be affected by crisis as an opportunity (β = 0.13), creative
visionary (β = 0.21) and observation of the business context (β = 0.16). For the
effect of an enterprise that offers an innovative product or service to the innovative
technology, the effect of reason for business (β = 0.28) is needed, as well as the
effect of environmental and social business decisions (β = 0.34) and creative
visionary (β = 0.12). The impact of an owned international turbulent business
requires the environmental and social impact of business decisions (β = 0.16) and
the impact of an enterprise that offers an innovative product or service (β = 0.45).
Finally, it is important to highlight the impact of the observation of business context
on the model as it applies to crisis as an opportunity (β = 0.27), creative visionary
(β = 0.19) and reason for business (β = 0.21).
In the case of Hungary, the role of subjective perception of the social context in
innovation cannot be measured. On the contrary, in the case of Poland, it is one of the
most important areas of technological innovation (β = 0.15). Compared to the
Hungarian model, the Polish model is built on the basis of an opposite cause-and-
effect network. Technological innovation and the enterprise innovation determine
the innovative entrepreneur, and technological innovation depends directly on
subjective perception of environment. Technical innovation has a strong influence
on why someone intends to create a business (β = -0.19) and how creative they feel
about themselves (β = 0.12). Although no significant correlation can be observed in
the model between crisis as an opportunity and technical innovation, the factor plays
an important role in business decisions (β = 0.228) and reason of business
(β = 0.482). In the model, the social environment appears as a strong influencing
factor in the development of innovation, and the innovations coming from the side of
the business, and it shapes the innovative entrepreneur.

4
The bootstrap method is a nonparametric procedure that allows testing the statistical significance
of different PLS-SEM results, including the significance testing of the path coefficient, R2 values,
VIF and HTMT values.
Two-Way Business Innovation in Central Eastern Europe:. . . 151

Fig. 2 The PLS-SEM and bootstrap model of business innovation in Hungary (N = 366) and
Poland (N = 1037)
152 M. Gosztonyi

Table 3 Results of the bootstrap procedure (Hungary)

Path coefﬁcient (β) Bootstrap conﬁdence intervals
Original sample Bootstrap STDEV (2.5%; 97.5%)
BDES → EOIPS 0.34 0.34 0.04 (0.117; 0.536)
BDES → IT 0.11 0.11 0.04 (-0.053; 0.260)
BDES → OITB 0.16 0.16 0.05 (0.057; 0.246)
CO → BDES 0.13 0.13 0.05 (-0.065; 0.324)
CV → BDES 0.21 0.21 0.05 (0.144; 0.272)
CV → EOIPS 0.12 0.12 0.05 (0.101; 0.135)
EOIPS → IT 0.46 0.46 0.06 (0.398; 0.516)
EOIPS → OITB 0.45 0.45 0.05 (0.371; 0.525)
OB → BDES 0.16 0.16 0.05 (0.085; 0.244)
OB → CO 0.27 0.28 0.05 (0.195; 0.346)
OB → CV 0.19 0.19 0.05 (0.128; 0.246)
OB → ROB 0.21 0.21 0.05 (0.107; 0.296)
OITB → IT 0.21 0.21 0.05 (0.144; 0.272)
ROB → EOIPS 0.28 0.28 0.05 (0.117; 0.430)

In summary, the business innovation in Hungary and in Poland takes place in a

complex system; however, in the two countries, innovation develops based on a
completely different causal chain. While in Hungary the innovative entrepreneur is
the one who creates innovation in a non-innovative environment, in Poland we can
observe the exact opposite; innovative entrepreneurs are formed through the inno-
vative environment and business. Based on these, the innovation development of
Hungary can be described with the classic Schumpeterian theory, while the innova-
tion in Poland can be characterized with enterprises’ productive-side effects theory.
As a summary of our research results, Table 5 presents the veriﬁcation of our
hypotheses.
We could not verify H1, H3, H6, H9, H13, H19 and H21–H24 among our hypotheses
on the Hungarian dataset, but all the other hypotheses were veriﬁed. In the Polish
model, we cannot verify H1, H4, H6–H8, H10, H12, H14–H18, H20 and H22. Based on
which the two countries operate on the same market, the business innovation is
created with two completely different logics; therefore, business innovation can be
explained with two different theories.

5 Discussion

In our paper, we examined, in the case of semi-peripheral countries (Hungary and

Poland), based on representative data, the casual effects for business innovation,
based on the two most important theoretical trends.
Our data show that the proportion of innovative enterprises is low in both
countries, accounting for about one-ﬁfth in Hungary or one-eleven in Poland of all
Two-Way Business Innovation in Central Eastern Europe:. . . 153

Fig. 3 The path coefﬁcients and factor weights of the Hungarian (N = 366) and Polish (N = 1037)
models
154 M. Gosztonyi

Table 4 Results of the bootstrap procedure (Poland)

Path coefﬁcient (β) Bootstrap conﬁdence intervals
Original sample Bootstrap STDEV (2.5%; 97.5%)
CO → BDES 0.228 0.237 0.107 (-0.223; 0.196)
CO → ROB 0.482 0.481 0.079 (-0.163; 0.150)
CV → BDES 0.130 0.126 0.048 (-0.103; 0.086)
EOIPS → CO -0.232 -0.235 0.104 (-0.195; 0.194)
IT → CV 0.120 0.122 0.033 (-0.066; 0.062)
IT → ROB -0.019 -0.019 0.009 (-0.019; 0.015)
OB → CV -0.279 -0.282 0.031 (-0.062; 0.056)
SC → IT -0.155 -0.159 0.040 (-0.079; 0.075)
SC → OITB -0.102 -0.103 0.040 (-0.075; 0.084)
SC → ROB -0.114 -0.113 0.038 (-0.075; 0.076)

enterprises. These businesses are typically SMEs, located in the more economically
developed parts of the country, in large cities. To explore the impact mechanisms
that lead to innovations, we analysed our data using the PLS-SEM methodology.
Our results show that business innovation in semi-peripheral countries results from
the combined effect of a complex tripartite system (entrepreneur-enterprise-environ-
ment). However, the causality of these factors differs from country to country.
In countries where the environment is less innovative, such as Hungary, the
entrepreneur has a direct impact on the creation of innovation, and the inﬂuence of
the environment almost completely disappears. In these systems a creative and
visionary entrepreneur must create an enterprise with which they ‘[want] to change
the world’ while gaining a satisfactory amount of wealth. Contrary to this, in the
economy where the environment is more supportive of innovation, like Poland, the
enterprise itself have a direct impact on the creation of innovation, and this has an
impact to the creation of innovative entrepreneur. In these systems the enterprise
itself must achieve a level of economic development embedded in an international
network. In addition, the enterprise needs to handle crisis issues and obstacles as
opportunities. All this can create the appearance of a new product and service in the
company, which has a direct positive impact on the appearance of entrepreneur
innovation. Thus, while in one case the development of innovation can be explained
mainly by the Schumpeterian theory, in the other case, the enterprises’ productive-
side effects theory can be fruitful in the explanation.
An additional but important result of our study is that innovation in semi-
peripheral countries can be closely linked to sustainable goals, as both the entrepre-
neur and the business must prioritise goals and decisions in keeping with social and
environmental sustainability to crate innovation.
Our study is not without limitations. For our model, we used two sample of two
CEE countries; therefore, the study can be made more accurate if other CEE
Two-Way Business Innovation in Central Eastern Europe:. . . 155

Table 5 Veriﬁcation of the research hypotheses

Hungary Poland
Hypothesis Relationship Result Relationship Result
CO → IT H1 Not significant Not verified Not significant Not verified
CO → BDES H2 Significant, Verified Significant, Verified
positive positive
CO → ROB H3 Not significant Not verified Significant, Verified
positive
OB → ROB H4 Significant, Verified Not significant Not verified
positive
OB → CV H5 Significant, Verified Significant, Verified
positive negative
OB → IT H6 Not significant Not verified Not significant Not verified
OB → CO H7 Significant, Verified Not significant Not verified
positive
OB → BDES H8 Significant, Verified Not significant Not verified
positive
ROB → IT H9 Not significant Not verified Significant, Verified
negative
ROB → H10 Significant, Verified Not significant Not verified
EOIPS positive
CV → BDES H11 Significant, Verified Significant, Verified
positive positive
CV → EOIPS H12 Significant, Verified Not significant Not verified
positive
IT → CV H13 Not significant Not verified Significant, Verified
positive
BDES → H14 Significant, Verified Not significant Not verified
EOIPS positive
BDES → IT H15 Significant, Verified Not significant Not verified
positive
BDES → H16 Significant, Verified Not significant Not verified
OITB positive
EOIPS → H17 Significant, Verified Not significant Not verified
OITB positive
EOIPS → IT H18 Significant, Verified Not significant Not verified
positive
EOIPS → CO H19 Not significant Not verified Significant, Verified
negative
OITB → IT H20 Significant, Verified Not significant Not verified
positive
SC → IT H21 Not significant Not verified Significant, Verified
negative
SC → BDES H22 Not significant Not verified Not significant Not verified
SC → ROB H23 Not significant Not verified Significant, Verified
negative
SC → OITB H24 Not significant Not verified Significant, Verified
negative
156 M. Gosztonyi

European Union countries are analysed. Our research can be further reﬁned by
including additional macro-contextual variables as well. In our research, we included
only cognitive variables to measure the macro-context, but this could be extended to
include either performance requirements (McClelland 1961) or risk-taking variables
(Gartner and Carter 2003), as well as measures of context regulators, values and
institutions. In addition, we think that it would be important to supplement the
analysis with measures of state corruption and freedom of economic competition.
By highlighting these metrics, it would be possible to further deepen the knowledge
about the countries following Schumpeterian logic. We could understand that the
innovative entrepreneurs in these countries, like Hungary, may follow an innovation
structure which is independent of their environment because the market itself does
not operate along the lines of a free and open economy.
In summary, the materialisation of business innovation is the result of a complex
three-dimensional mechanism of effects that requires not a single factor but a
combination of factors. In all of this, there is a need for entrepreneurs who can
‘dream’ and stimulate innovation along the lines of sustainability. However, having
a macro-environment is equally important, as it would allow them to implement
these innovations and develop their entrepreneurial spirit and can create the inno-
vative entrepreneur. Furthermore, there is a need for an economically prosperous
business with international connections, which can provide an appropriate frame-
work for creating new products and services, and the tendency to see challenges as
opportunities and not as problems. The combined effect of all this is needed for the
creation of business innovation in the CEE market, which can lead later to economic
growth.

Appendix

Table 6 Reliability and validity of the latent variables (Hungary and Poland)
Hungary Poland
Composite Average variance Composite Average variance
reliability extracted (AVE) reliability extracted (AVE)
BDES 0.83 0.71 BDES 0.92 0.80
CO 0.73 0.59 CO 1.00 1.00
CV 0.79 0.65 CV 0.74 0.59
EOIPS 0.79 0.65 EOIPS 0.94 0.89
IT 1.00 1.00 IT 1.00 1.00
OB 0.76 0.62 OB 0.85 0.66
OITB 0.78 0.54 OITB 1.00 1.00
ROB 0.81 0.68 ROB 1.00 1.00
SC 0.76 0.52
Two-Way Business Innovation in Central Eastern Europe:. . . 157

Table 7 Indicator loadings of measured variables and their VIF (collinear statistics) values
(Hungary and Poland)
Hungary Poland
Indicator loadings VIF Indicator loadings VIF
creativ 0.91 1.12 creativ 0.86 1.04
nbmedia 0.89 1.07 easystartL 0.76 1.41
nbsocent 0.67 1.07 nbgoodcL 0.81 1.62
omcrgrow 0.73 1.12 nbmediaL 0.79 1.41
omcstexp 0.73 1.19 nbstatusL 0.85 1.52
omcstnat 0.52 1.06 omcrgrow 1.00 1.00
omnewproc 1.00 1.00 omnewproc 1.00 1.00
omnewprod 0.85 1.10 opportL 0.54 1.27
omown 0.74 1.24 sucpnewopp 1.00 1.00
suacts 0.76 1.10 sumotiv1 1.00 1.00
sucpnewopp 0.95 1.06 suown 0.94 1.49
sumotiv1 0.89 1.16 suowners 0.94 1.49
sumotiv2 0.75 1.16 susdg_env 0.94 1.28
susdg_env 0.87 1.22 susdg_pri 0.85 1.99
susdg_soc 0.82 1.22 susdg_soc 0.89 1.19
vision 0.69 1.12 suskillL 0.84 1.13
vision 0.67 1.04

References

Acs ZJ, Szerb L (2007) Entrepreneurship, economic growth and public policy. Small Bus Econ
28(2–3):109–122
Audretsch DB, Keilbach M (2007) The theory of knowledge spillover entrepreneurship. J Manag
Stud 44(7):1242–1254
Audretsch DB, Peña-Legazkue I (2012) Entrepreneurial activity and regional competitiveness: an
introduction to the special issue. Small Bus Econ 39(3):531–537
Audretsch DB, Keilbach MC, Lehmann EE (2006) Entrepreneurship and economic growth. Oxford
University Press, Oxford
Audretsch DB, Bönte W, Keilbach M (2008) Entrepreneurship capital and its impact on knowledge
diffusion and economic performance. J Bus Ventur 23(6):687–698
Bandura A (1986) Social foundations of thought and action: a social cognitive theory. Prentice Hall,
New Jersey
Basco R, Hair JF Jr, Ringle CM, Sarstedt M (2022) Advancing family business research through
modeling nonlinear relationships: comparing PLS-SEM and multiple regression. J Fam Bus
Strat 13(3):100457
Baumol WJ (1968) Entrepreneurship in economic theory. Am Econ Rev 58:64–71
Baumol WJ (2010) The microtheory of innovative entrepreneurship. Princeton University Press,
Princeton, NJ
Bennett A, Checkel JT (eds) (2015) Process tracing. Cambridge University Press, Cambridge
Blatter J, Haverland M (2012) Designing case studies: explanatory approaches in small-N research.
Palgrave Macmillan, Basingstoke
Bruton GD, Ahlstrom D, Li H-L (2010) Institutional theory and entrepreneurship: where are we
now and where do we need to move in the future? Entrep Theory Pract 34(3):421–440
158 M. Gosztonyi

Byrne D, Callaghan G (2013) Complexity theory and the social sciences: the state of the art. Taylor
& Francis, London
Cilliers P (2001) Boundaries, hierarchies and networks in complex systems. Int J Innov Manag 5(2):
134–147
Cohen B, Winn MI (2007) Market imperfections, opportunity and sustainable entrepreneurship. J
Bus Ventur 22(1):29–49
Cohen B, Smith B, Mitchell R (2008) Toward a sustainable conceptualization of dependent vari-
ables in entrepreneurship research. Bus Strateg Environ 17(2):107–119
Dacin MT, Goodstein J, Scott RW (2002) Institutional theory and institutional change: introduction
to the special research forum. Acad Manag J 45(1):45–56
Davidsson P (2006) Nascent entrepreneurship: empirical studies and developments. Found Trends
Entrep 2(1):1–76
de Noronha Vaz T, Nijkamp P (2009) Knowledge and innovation: the strings between global and
local dimensions of sustainable growth. Entrep Reg Dev 21(4):441–455
Deacon TW (2007) Three levels of emergent phenomena. In: Murphy A, Stoeger L (eds) Evolution
and emergence: systems, organisms, persons. Oxford University Press, Oxford, pp 88–110
DeLanda M (2005) Intensive science and virtual philosophy. Continuum impacts. Bloomsbury
Academic, New York
Dijkstra TK, Henseler J (2015) Consistent and asymptotically normal PLS estimators for linear
structural equations. Comput Stat Data Anal 81(1):10–23
DiMaggio PJ (1988) Interest and agency in institutional theory. In: Zucker LG (ed) Institutional
patterns and organizations: culture and environment. Ballinger Pub. Co, Cambridge, pp 3–23
Fagerberg J (2006) Innovation: A guide to literature. In: Fagerberg J, Mowery DC, Nelson RR (eds)
The Oxford handbook of innovation. Oxford University Press, Oxford, pp 1–27
Friedman JH, Popescu BE (2003) Importance sampled learning ensembles. J Mach Learn Res 1:1–
32
Fritsch M (2008) How does new business formation affect regional development? Introduction to
the special issue. Small Bus Econ 30(1):1–14
Gartner WB, Carter NM (2003) Entrepreneurial behaviour and ﬁrm organizing processes. In: Acs
ZJ, Audretsch DB (eds) Handbook of entrepreneurship research. Springer, New York, pp
195–222
Glaeser EL (1999) Learning in cities. J Urban Econ 46(2):254–277
González-Pernía JL, Peña-Legazkue I (2015) Export-oriented entrepreneurship and regional eco-
nomic growth. Small Bus Econ 45(3):505–522
Gorzeń-Mitka I (2013) Risk identiﬁcation tools - Polish MSMES companies practices. Probl Manag
21st Century 7:6–11
Grossman GM, Helpman E (1994) Endogenous innovation in the theory of growth. J Econ Perspect
8(1):23–44
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2016) A primer on partial least squares structural
equation modeling (PLS-SEM). SAGE Publications Inc, London
Hair JF, Matthews LM, Matthews RL, Sarstedt M (2017) PLS-SEM or CB–SEM: updated
guidelines on which method to use. Int J Multivar Data Anal 1(2):107–123
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. SAGE Publications Inc., Thousand Oaks, CA
Harris RG (2001) The knowledge-based economy: intellectual origins and new economic perspec-
tives. Int J Manag Rev 3(1):21–40
Hayton JC (2005) Promoting corporate entrepreneurship through human resource management
practices: a review of empirical research. Hum Resour Manag Rev 15(1):21–41
Hockerts K, Wüstenhagen R (2010) Greening Goliaths versus emerging Davids. Theorizing about
the role of incumbents and new entrants in sustainable entrepreneurship. J Bus Ventur 25(5):
481–492
Two-Way Business Innovation in Central Eastern Europe:. . . 159

Holm P (1995) The dynamics of institutionalization. Transformation processes in Norwegian

fisheries. Adm Sci Q 40(3):398–422
Hu L-T, Bentler PM (1998) Fit indices in covariance structure modeling: sensitivity to
underparameterized model misspecification. Psychol Methods 3(4):424–453
Huggins RA, Izushi H (2007) Competing for knowledge: creating, connecting and growing.
Routledge, London
Huggins RA, Johnston A (2009) Knowledge networks in an uncompetitive region: SME innovation
and growth. Growth Chang 40(2):227–259
Ibert O (2007) Towards a geography of knowledge creation: the ambivalences between ‘knowledge
as an object’ and ‘knowing in practice’. Reg Stud 41(1):103–114
Ikeda S (2008) The meaning of ‘social capital’ as it relates to the market process. Rev Austrian Econ
21(2/3):167–182
Janasz W, Kozioł K (2007) Determinanty działalności innowacyjnej przedsiębiorstw. PWE,
Warszawa
Johnson JS, Sohi RS (2014) The curvilinear and conditional effects of product line breadth on
salesperson performance, role stress, and job satisfaction. J Acad Mark Sci 42(1):71–89
Kirton M (1987) Kirton adaption-innovation inventory manual, 2nd edn. Occupational Research
Centre, Hatfield
Kirzner IM (1973) Competition and entrepreneurship. University of Chicago Press, Indianapolis
Latusek-Jurczak D (2014) Formy współpracy międzyorganizacyjnej. In: Koźmiński AK, Latusek-
Jurczak D (eds) Relacje międz yorganizacyjne w naukach o zarządzaniu. Oficyna Wolters
Kluwer business, Warsaw, p 59
Lewandowska A, Stopa M (2018) SMEs innovativeness and institutional support system: the local
experiences in qualitative perspective. Polish case study. Oeconomia Copernicana 9(2):333–351
Lohmöller J-B (1989) Latent variable path modelling with partial least squares. Physica Heidelberg,
Heidelberg
Matthews L (2017) Applying multigroup analysis in PLS-SEM: a step-by-step process. In: Latan H,
Noonan R (eds) Partial least squares path modeling. Springer, Cham, pp 219–243
Maturana HR, Varela FJ (1980) Autopoiesis and cognition: the realization of the living. Springer,
Dordrecht
McClelland D (1961) The achieving society. Free Press, New York
Mckeown M (2008) The truth about innovation. Pearson Prentice Hall, London
Nicolis G (2012) Introduction to nonlinear science. Cambridge University Press, Cambridge
Noseleit F (2013) Entrepreneurship, structural change, and economic growth. J Evol Econ 23(4):
735–766
Okręglicka M (2014) Adoption and use of ICT as a factor of development of small and medium-
sized enterprises in Poland. Przedsiębiorczość i Zarządzanie 15(7):393–405
Ostrom E (1990) Governing the commons: the evolution of institutions for collective action.
Cambridge University Press, Cambridge
Podręcznik O (2005) OECD/Wspólnoty Europejskie. OECD
Rasoolimanesh SM (2022) Discriminant validity assessment in PLS-SEM: a comprehensive
composite-based approach. Data Anal Perspect J 3(2):1–8
Reed M, Harvey DL (1996) Social science as the study of complex systems. In: Kiel LD, Elliott E
(eds) Chaos theory in the social sciences. University of Michigan Press, US, pp 295–324
Reinartz W, Haenlein M, Henseler J (2009) An empirical comparison of the efficacy of covariance-
based and variance-based SEM. Int J Res Mark 26(4):332–344
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. SmartPLS
Romer PM (1986) Increasing returns and long run growth. J Polit Econ 94(5):1002–1037
Sachpazidu-Wójcicka K (2017) Innovation as a determinant of the competitiveness of Polish
enterprises. Oeconomia Copernicana 8(2):287–299
Samuelsson M, Davidsson P (2009) Does venture opportunity variation matter? Investigating
systematic process differences between innovative and imitative new ventures. Small Bus
Econ 33(2):229–255
160 M. Gosztonyi

Sarstedt M, Ringle CM, Hair JF (2021) Partial least squares structural equation modeling. A
workbook. Springer International Publishing, Cham
Sarstedt M, Radomir L, Moisescu OI, Ringle CM (2022) Latent class analysis in PLS-SEM: a
review and recommendations for future applications. J Bus Res 138:398–407
Schumpeter J (1934) The theory of economic development. Harvard University Press, London
Spencer AS, Kirchhoff BA, White C (2008) Entrepreneurship, innovation, and wealth
distribution—the essence of creative destruction. Int Small Bus J 26(1):9–26
Stevenson HH, Jarrillo-Mossi JC (1986) Preserving entrepreneurship as companies grow. J Bus
Strateg 7(1):10–23
Szopik-Depczyńska K (2015) Effects of innovation activity in industrial enterprises in Eastern
Poland. Oeconomia Copernicana 6(2):53–65
Taylor SE, Schroeder K (2003) Inside intuit: how the makers of quicken beat Microsoft and
revolutionized an entire industry. Harvard Business Review Press, London
Tokarski J (ed) (1980) Słownik wyrazów obcych. PWN, Warszawa
Weber M (1978 [1921–1922]) Economy and society: an outline of interpretive sociology. Univer-
sity of California Press, Berkeley, CA (Original work published in 1921–1922)
Wong PK, Ho YP, Autio E (2005) Entrepreneurship, innovation and economic growth: evidence
from GEM data. Small Bus Econ 24(3):335–350
Zotto CD, Gustafsson V (2008) Human resource management as an entrepreneurial tool? In:
Barret R, Mayson S (eds) International handbook of entrepreneurship and HRM. Edward
Elgar Publishing, London, pp 89–110
Zucker LG, Darby MR, Furner J, Liu RC, Ma H (2007) Minerva unbound: knowledge stocks,
knowledge ﬂows and new knowledge production. Res Policy 36(6):850–863
Zygmunt A (2017) Innovation activities of Polish ﬁrms. Multivariate analysis of the moderate
innovator countries. Oeconomia Copernicana 8(4):505–521
Institutional Context and Entrepreneurship
Typologies: The Moderator Role
of Environmental Uncertainty

Serap Korkarer, Mahmut Hızıroğlu, and Joseph F. Hair Jr.

1 Introduction

In the past two decades, the field of entrepreneurship has expanded rapidly as an
independent discipline. However, the theoretical foundations of the “opportunity
recognition” phenomenon, one of the most important concepts in the discipline, are
still being questioned (Alvarez et al. 2016; Shane and Venkataraman 2000). The
debate centers around the epistemological foundations of “opportunity recognition”
and focuses on the question of whether there are opportunities waiting to be
discovered regardless of entrepreneurs’ perceptions, in contrast to opportunities
created by the actions of entrepreneurs (Alvarez and Barney 2007). “Opportunity
creation,” rooted in Schumpeter’s notion of “creative destruction,” views the oppor-
tunities as a subjective phenomenon within which entrepreneurs emerge. Moreover,
opportunity definition is linked to creativity in a fragmented and ambiguous context
in which the entrepreneur creates reality rather than choosing it (Weick 1979). On
the contrary, “opportunity discovery” from a Kirznerian perspective assumes that
opportunities exist objectively in the market, independent of the entrepreneur. The
job of the entrepreneur, therefore, is to identify these opportunities and take advan-
tage of them.
Despite their epistemological differences, and that the type of entrepreneurial
actions (EA) is context dependent (Sarasvathy et al. 2003), these two different
definitions of opportunity complement each other. Nevertheless, due to the claim

S. Korkarer (✉) · M. Hızıroğlu

Department of Management, Faculty of Economics, Istanbul University, İstanbul, Turkey
e-mail: [email protected]
J. F. Hair Jr.
Mitchell College of Business, University of South Alabama, Mobile, AL, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 161

that the phenomenon of “opportunity discovery” (Kirznerian entrepreneurship) is

more common in practice (Filser et al. 2020), only one aspect of opportunity
identification is the subject of empirical research. Criticisms of the understanding
that a single comprehensive definition of opportunity is necessary to make progress
in the field of entrepreneurship (i.e., Alvarez and Barney 2020; Foss and Klein 2020)
indicates there is a need for empirical studies that explore both aspects of this
phenomenon. Few empirical studies incorporate both aspects of opportunity defini-
tion in the research design. For instance, de Jong and Marsili (2015) investigated the
extent to which opportunities, realized in the form of new products introduced in the
market, could be considered of Schumpeterian versus Kirznerian nature. Similarly,
Smith et al. (2009) examined how relative differences in the degree of opportunity
tacitness relate to the process of opportunity identification. Another study by
Vaghely and Julien (2010) described a model which can provide a framework for
understanding the entrepreneur’s use of information to identify opportunities.
Finally, Welter (2012) described the boundary conditions between opportunity
types and used these boundary conditions to offer empirical guidance for testing
the different types of opportunities.
Although the abovementioned studies provide new and interesting insights into
the nature and antecedents of different types of EA, only Welter’s work (2012) has
linked the opportunity type to context. Studies that associate context with entrepre-
neurship conceptualize entrepreneurship as a single construct and examine the
impact of contextual features on entrepreneurship (Alvarez and Urbano 2011). To
date, empirical consideration on how institutional context (IC) influences the differ-
ent type of EA is a relatively neglected issue. Moreover, the literature on EA which is
generally combined with institutional theory has frequently been studied in devel-
oped countries (Su et al. 2017). It is implicitly assumed that countries have their own
national entrepreneurship system (Busenitz et al. 2000) and that an IC in a country is
perceived in the same way by everyone. But according to the Etzioni’s assessment of
the legitimacy of entrepreneurship, due to the different subcultures, ideologies, and
religious understandings in a society, the institutional environment can be perceived
differently within a country (Etzioni 1987). The objective of the present study,
therefore, is to empirically examine the opportunity identification debate by inves-
tigating the role of the normative, regulatory, and cognitive perceptions of IC on
creative and discovery EA.
Based on a contingency approach, the behavior of actors is further conditioned by
the environmental uncertainty (EU). Welter (2012) claims the existence of EU can
define the line between creation and discovery and therefore outline the possible
actions of entrepreneurs in the face of uncertainty (Knight 1964; Sarasvathy 2001).
In this study we further posit that EU moderates the relationship between the
perception of IC and the type of EA.
The data for this research is collected in Turkey, a rich source for institutional
theory studies (Taş and Hızıroğlu 2016). Turkey has a market with free competition,
which generally exhibits an oligopoly structure. While the role of the state is as an
actor supporting the private sector, it also shapes the market (Bugra 2007; Taş and
Hızıroğlu 2016). Moreover, the bourgeoisie is not immune to the ideological and
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 163

Environmental Uncertainty

Perceived Institutional
Context +H2a
+H1a

Regulative Context Discovery EA

+H1
Normative Context

Creative EA
Cognitive Context +H2

Fig. 1 Theoretical model

political divisions in Turkey, which in turn may affect the perceptions of entrepre-
neurs. Thus, the ideologies of interest groups and ruling elites in Turkey shape both
their own and followers’ thoughts and actions and assume the position of an
important source of power for them. Given these observations, the main contribu-
tions of this study are (1) to explore whether the ideological, political, and ethnic
differences in a developing country can lead to differentiation in the perceptions of
the institutional environment and (2) to test whether this difference has an impact on
the types of EA, and thereby provide empirical evidence of the ongoing phenomenon
of opportunity recognition in the ﬁeld of entrepreneurship.

2 Theoretical Model and Hypotheses

Figure 1 shows the model we investigate. The theoretical model suggests ﬁrms’ IC is
directly related to their strategic EA and that EU plays a moderating role in this
relationship. The proposed variables are drawn from different disciplines and have
been applied in previous theoretical explanations and empirical studies. In this
context, each variable in the model is explained below.

2.1 Entrepreneurial Actions

EA, as a set of ideas, beliefs, and actions that enable the creation of future goods and
services, are not present in the current market (Sarasvathy et al. 2003; Venkataraman
1997), thereby facilitating the potential in the environment and taking advantage of
164 S. Korkarer et al.

the opportunities (Shane and Venkataraman 2000). The EA literature focuses on

processes (Cohen and Winn 2007; Dean and McMullen 2007), opportunities, and
threats at multiple levels. However, the connections between the processes are not
well understood (Shepherd and Patzelt 2011). In this study we propose that the
relationships between multiple levels can be explained by studying causal-predictive
mechanisms, situational factors, action formation, and transformation mechanisms
(Coleman 1986; Hedström and Swedberg 1998).
Different assumptions of teleological theories (such as motivation theory, Maslow
(1943); institutional theory, DiMaggio and Powel (1983)) about the nature of the
actor, nature of the entrepreneur’s objective, and nature of the decision-making
context on entrepreneur’s same actions lead to different types of EA (Alvarez and
Barney 2007). Based on those assumptions, Sarasvathy et al. (2003) proposed three
different, context-dependent perspectives: allocative, discovery, and creative. The
allocative perspective is related to allocative efﬁciency which is valid only when a
competitive market works perfectly (Sarasvathy et al. 2003). Therefore, only the
discovery and creative perspectives are examined in this research.

2.1.1 Discovery EA

Opportunities created by external causes such as changes in technology, consumer

preferences, and the context of the firm or industry (Dew et al. 2004; Kirzner 1973)
are independent of individual opportunities, waiting to be discovered outside (Alva-
rez and Barney 2007; Knight 1964). Entrepreneurial firms must be alerted to
proactively take advantage of information asymmetry despite the errors in the
market. However, since profit opportunities will attract the attention of competitors
and imitators, companies are expected to pursue these opportunities quickly with
their competitive aggressive behavior style (Sundqvist et al. 2012). Therefore, firms
must explore and pursue existing and latent markets (Sarasvathy et al. 2003).
In this view, the focus is on asymmetric information between economic actors.
The acquisition of knowledge creates opportunities for entrepreneurs to discover
based on their knowledge advantage and other complementary skills (Venkataraman
1997). More specifically, Kirzner (1973) noted that alert entrepreneurs discover
opportunities by shifting the “invisible” hand toward pareto optimal solutions.

2.1.2 Creative EA

The “opportunity-creating entrepreneurship” approach is based on the

Schumpeterian and evolutionary economics tradition (Alvarez and Barney 2007).
The increase in the level of dynamism causes supply and demand to be unknown,
and economic innovations in areas such as ﬁnance and marketing must be identiﬁed
to create opportunities (Grove et al. 2018; Schumpeter 1934). The most important
distinguishing feature is that the entrepreneur has a causal role in the emergence of
opportunities (Alvarez and Barney 2007). Thus, there is no “end” until the creative
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 165

process takes place. That is, opportunities cannot be understood until they exist and
only after engaging in an iterative process of action and reaction (Buchanan and
Vanberg 1991; Weick 1979). Therefore, ﬁrms need autonomy to be innovative, take
more risk (Lumpkin and Dess 1996), and “research” has almost no meaning. In
creativity, the entrepreneur acts and observes how the market and customers react to
their actions (Alvarez and Barney 2007; Gartner 1985; Weick 1979).

2.2 Institutional Context

Firms need to meet the demands of the technical environment to produce/serve

effectively and efficiently (Scott 1995). However, the behavior of firms is a response
to social pressures such as conformity and legitimacy, which they are generally
exposed to due to the symbolic environment created by other organizations
(Suddaby 2013), while also a product of the IC (Greenwood and Hinings 1996;
Meyer and Rowan 1977; North 1991; Tolbert et al. 2011).
Institutions have three types of mechanisms that affect the behavior and actions of
institutions, individuals, and organizations: regulative, normative, and cognitive
(Scott 1995). These institutions send signals and support firms by providing oppor-
tunities such as stability, social support, access to resources, and legitimacy (Baum
and Oliver 1991). Thus, companies can identify strategic opportunities such as
identifying gaps in the market and understanding customer wants and needs
(Adner and Helfat 2003). Different actors in the same context choose appropriate
and legitimate options for the time being under bounded rationality (Oliver 1991),
and take new, creative, and different actions through different perspectives (Daft and
Weick 1984; Etzioni 1987; Knight 1964; Milliken 1987).
In discovering opportunities, entrepreneurial firms have to be alert to take advan-
tage of information asymmetry despite the errors in the market (Kirzner 1973).
Research shows that alert entrepreneurs regularly scan their environment for poten-
tial ideas (Busenitz 1996). Entrepreneurs rely on different types of information and
clues (Lindsay and Craig 2002) and discover more opportunities by choosing
effective options among research tactics, information processing ability, and deter-
mined opportunities (Murphy 2011). When firms assume their context to be mea-
surable and clear, they discover opportunities by collecting information and making
rational analyzes and measurements of opportunities (Daft and Weick 1984). Thus,
firms are expected to take advantage of the opportunities in the market (Khandwalla
1987). For example, Gaur and Lu state that multinational firms have built their R&D
centers in the USA due to regulatory factors such as being well advanced in
copyright protection (Gaur and Lu 2007). Therefore, we propose the following
hypothesis:
H1: There is a positive relationship between firms’ perceptions of institutional
context and their discovery entrepreneurial actions.
166 S. Korkarer et al.

Entrepreneurs create meaning about their environment based on the situations

they encounter, because “means affect ends” (Weick 1979, p. 64). Signals from the
institutions cause entrepreneurs to dream about potential opportunities in the market,
which will lead entrepreneurs to create innovations such as products, services,
processes, and ultimately markets (Khandwalla 1976). If the legal regulations
made by the state increase the cooperation and trust environment among the com-
panies, the level of using external resources to improve the innovation processes of
the companies will increase (Iturrioz et al. 2015). For example, Panasonic has
established a department called Eco Solutions North America, which focuses on
renewable energy and energy efficiency projects in the USA and Canada, based on
supportive legal regulations (Zhang et al. 2022). Moreover, national institutional
patterns, such as access to research and education institutions, finance, and trained
employees, enable managers to make informed EA decisions (Zablah et al. 2012),
thus leading individuals to be more likely to produce creative results (Fleming et al.
2007). Societies that exhibit positive value toward creativity and change create a
normative environment that leads to an increase in creative EA (Gómez-Haro et al.
2011). In addition, if the context is changing and quite demanding, managers will
find the context difficult to analyze. Thus, they will undertake creative initiatives
within the uncertain contexts of their environment that cannot be analyzed (Daft and
Weick 1984). Therefore, we propose the following hypothesis:
H2: There is a positive relationship between firms’ perceptions of institutional
context and their creative entrepreneurial actions.

2.3 The Moderating Role of Environmental Uncertainty

In the strategic management and entrepreneurship literature, the ﬁrm’s business

environment plays a regulating role as it affects the effectiveness of the firm’s
strategy (Lumpkin and Dess 2001). EU includes market dynamism, technological
dynamism, competitive intensity, and economic changes (Elbanna and Child 2007a).
Since firms’ perceptions of uncertainty (ambiguity) are different from each other,
they will make their decisions by considering the possibilities they create according
to their own perceptions (Duncan 1972). Environmental dynamism has mixed
effects on the firm. For example, it may enable firms to pursue more new opportu-
nities in a more dynamic environment (Shane and Venkataraman 2000). On the other
hand, to take advantage of these opportunities, firms must be flexible (Teece et al.
1997), take risks (Kets de Vries 1977), apply their initiative and desire to succeed
(Weick and Roberts 1993), and take timely actions (Wiklund and Shepherd 2003).
Since the demands and needs will not change much in static markets compared to
dynamic ones, it is easier to predict compared to dynamic markets. In this case,
discovery EA are expected to take advantage of the opportunities in the market
(Khandwalla 1987). However, in dynamic markets where competition is high,
customers have many alternatives and search for options that are satisfying. For
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 167

this reason, firms in such markets carry out more proactive and aggressive strategies
(Sarasvathy et al. 2003; Teece et al. 1997). Therefore, our hypothesis is the
following:
H1a: Environmental uncertainty plays a moderating role in the relationship
between the firm’s perception of institutional context and its discovery EA.
In an environment where competition does not occur, opportunities arise from the
individuals themselves to initiate change in the economy with technological change
and innovation (Buchanan and Vanberg 1991; Sarasvathy et al. 2003). In markets
where such dynamism is high, new markets such as Netscape, Airbnb, Über,
Amazon, and Tesla are being created (Grove et al. 2018); therefore, creative EA
are expected (Teece 2007). Because there is uncertainty in these markets, customer
demands are unknown, and companies need to be more innovative (García-Morales
et al. 2012). In this case, companies are expected to invest more in research and
development activities and take risks (Lumpkin and Dess 2001; Sundqvist et al.
2012). So, we propose as follows:
H2a: Environmental uncertainty plays a moderating role in the relationship
between the firm’s perception of institutional context and its creative EA.

3 Methodology

Due to inconsistent ﬁndings of previous studies and the fact that no research has
previously been conducted in the context of Turkey, the proposed research model is
considered a causal-predictive model for exploratory, explanatory, and predictive
purposes (Hair et al. 2022). Existing literature was explored for the connections of
the perspectives with each other. Then, proposed relationships are explored and
explained using quantitative analysis.

3.1 Research Procedure and Data Collection

An online survey method is used. First, participants were contacted by phone, their
permission was obtained, and information was given about the research, and then
questionnaires were sent to them in the digital environment (email, social media,
WhatsApp) (Creswell and Creswell 2018, p. 90). Convenience, purposive, and
snowball non-probabilistic sampling techniques were used (Coşkun et al. 2019,
p. 169). The scales were adapted from established and validated scales used in
previous studies in accordance with the Turkish context. Permission to use the scales
was obtained from the respective authors via email. Table 1 shows the studies from
which the scales were adapted.
168 S. Korkarer et al.

Table 1 Studies the scales adapted

# of
Scale Reference items Title
Discovery EA Jambulingam 14 The Influence of Decision, Environmental and Firm
et al. (2005) Characteristics on The Rationality of Strategic Deci-
sion-Making
Jaworski and Market orientation: antecedents and consequences
Kohli (1993)
Creative EA Jambulingam 12 The Influence of Decision, Environmental and Firm
et al. (2005) Characteristics on the Rationality of Strategic Deci-
sion-Making
Institutional Busenitz et al. 13 Country institutional profiles: Unlocking entrepre-
context (2000) neurial phenomena
Environmental Elbanna and 14 The Influence of Decision, Environmental and Firm
uncertainty Child (2007a) Characteristics on The Rationality of Strategic Deci-
sion-Making

The questionnaire statements were translated separately from English to Turkish

and from Turkish to English by two bilingual speakers (Brislin 1980). The translated
statements were evaluated by four academics in Turkey who are experts in strategic
management, institutional theory, and entrepreneurship. The expressions that were
not understood enough or had grammatical errors were re-evaluated considering the
theoretical framework and were finalized in accordance with the approach and
relevant theories. Afterward, the questionnaires were sent to the editors, academi-
cians at different universities, and senior management of the company, the research
owner works, for evaluation. After confirmation of smooth understanding of expres-
sions, the questionnaire was finalized, and the face validity of the survey was
ensured (Hair et al. 2014).
Since entrepreneurial action typologies of firms are being examined, the firm
level is the unit of analysis. Considering the most appropriate participant selection
criteria of Huber and Power (1985), key influencers (top management, company
owner, board member, etc.) are chosen as the participants (Huber and Power 1985).
Therefore, the level of measurement can be described as individual. Since the
individual data is then subject to a firm-level measurement while analyzing the
data, the analysis level is again determined as firm. The variables consist of
higher-order constructs with lower-order constructs, designed as reflective-reflective
based on theoretical considerations and other similar studies.

3.2 Analysis

The measurement and structural model analyses was executed using partial least
squares structural equation modeling (PLS-SEM) and the SmartPLS 4 software. For
the analysis, the studies of Hair and colleagues (Hair et al. 2014; Hair et al. 2022)
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 169

were considered. Before starting the analysis, multivariate outliers were checked
with the SPSS 25 program using the Mahalanobis distance test (Tabachnick et al.
2013, pp. 91–93); a total of 17 outliers were removed from the data set. In this
context, analyses were carried out with 788 survey responses. The sample size
complies with the criteria specified in the literature. The scales consist of a total of
53 statements with a sample size of 788. The final sample size was sufficient,
therefore, based on the Kock and Hadaya inverse square root method (Hair et al.
2014, p. 100).
PLS-SEM was preferred as the analysis technique due to the various advantages
mentioned in the literature (Ringle et al. 2012; Sarstedt et al. 2017). However, the
main reasons for choosing PLS are its statistical power is higher than ML-based
CBSEM (covariance based) (Chin et al. 2003) and the emphasis on prediction of
the endogenous constructs (Hair et al. 2022). A post hoc analysis was conducted of
the moderator effect. The extended repeated indicators approach was used to assess
the higher-order constructs.1

3.2.1 Common Method Variance (CMB)

To minimize the likelihood of CMB, the questionnaire was designed based on

guidelines by Podsakoff and colleagues (Podsakoff et al. 2003). The Harman’s
single factor method was applied on a post hoc basis to examine CMB, and the
common variance extracted was only 24.035% (< 50; Podsakoff et al. 2003). The
results are far lower than would indicate a problem with CMB (Babin et al. 2016).

3.2.2 Descriptive Statistics and Correlation Analysis

Table 2 presents the means, standard deviations, and correlation coefficients between
the variables (N = 788). There is a moderate relationship between most of the
variables. The patterns of the simple correlations support the findings of the multi-
variate analysis that examines the theoretical relationships between the constructs
measured with multi-item scales. The IC and EU are significantly correlated with
creative EA and discovery EA. This appears to support the EA process model.
The level of agreement with the discovery EA statements is 3.60, while 3.10 with
creative one, which could be evaluated as entrepreneurs, is more prone to discovery
EA. The average level of agreement value with IC is 2.81, while 2.89 with
EU. Based on the average level of agreement values with IC (2.81) and EU (2.89),
it may be misleading to interpret the participants as having a very positive perception

1
Details of how the variables were operationalized and original sources for the measures employed
for the ones the table of which has not been put on this document can be obtained from the
corresponding author.
170 S. Korkarer et al.

Table 2 Descriptive statistics and correlations among variables

Environmental
Mean S.D. Creative Discovery Ins_Cont. uncertainty
Creative 3.10 0.55 1.000
Discovery 3.60 0.64 0.572** 1.000
Ins_Cont 2.81 0.61 0.142** 0.079* 1.000
Environmental 2.89 0.57 0.153** 0.165** 0.286** 1.000
Uncertainty
Notes: *significant at 0.05 level; **significant at 0.01 level, p > 0.05 no statistically significant
relation. Pearson correlation coefficient power levels: 0 < r < 0.299 weak; 0.300 < r < 0.599
medium; 0.600 < r < 0.799 strong; 0.800 < r < 0.999 very strong

of their IC and external environment in terms of doing business and entrepreneurship

in Turkey.
A total of 68.7% of the respondents are male, 37.2% are between the ages of
40 and 49, and 45.9% are undergraduates. In addition, 84.5% are senior management
and company owner; 34.6% have been working in their current position for 21 years
and above, while 28.3% have been working in a similar position 11–15 years.
Therefore, it was concluded that key respondents were familiar with the survey
statements. The ﬁrms respondents work for are predominantly medium and large
enterprises (40% of them are international, 30.1% are in operation for 35 years or
more, and 57% are medium and large enterprises). Approximately 50% of the ﬁrms
are in the manufacturing sector, and 73% are at medium technological level.

3.2.3 Measurement Model Assessment

PLS-SEM theoretical models are evaluated following a two-step process, as outlined

in the CCA procedure (Hair et al. 2020). The first step is measurement model
assessment which examines and confirms reliability and validity. The second step
is structural model evaluation focusing on prediction (Hair et al. 2022). The infor-
mation presented in Table 3 confirms the validity and reliability of the measurement
model constructs.

3.2.4 Structural Model Assessment

Main Model

Structural model assessment again follows the CCA procedure (Hair et al. 2020).
The VIF values were examined to assess multicollinearity. No problems were
identified (< 3.0). Next, the significance of the structural path coefficients (betas)
was examined. The path coefficients ranged from rather small to quite high, but all
were significant <0.05 level. Explained variance in the dependent variables (coef-
ficients of determination—R2), also known as in-sample prediction, was evaluated
Table 3 The validity and reliability of the reflective constructs
Convergent validity Internal consistency reliability Discriminant validity
Structure Indicator (ind.) Loadingsi Ind. rel.i1 AVE CAi2 rho_Ai3 CRi4 F&Li5 HTMTi6
Inno Inn1 0.756 0.57 0.680 0.765 0.777 0.864 Yes Yes
Inn3 0.862 0.74 Yes Yes
Inn4 0.852 0.73 Yes Yes
Auto Aut1 0.724 0.52 0.555 0.734 0.734 0.833 Yes Yes
Aut2 0.772 0.60 Yes Yes
Aut3 0.749 0.56 Yes Yes
Aut4 0.735 0.54 Yes Yes
Creative Inn1 0.617 0.38 0.454 0.798 0.804 0.852 Yes Yes
Inn3 0.700 0.49 Yes Yes
Inn4 0.778 0.61 Yes Yes
Aut1 0.723 0.52 Yes Yes
Aut2 0.648 0.42 Yes Yes
Aut3 0.614 0.38 Yes Yes
Aut4 0.619 0.38 Yes Yes
Proac Pro3 0.868 0.75 0.755 0.676 0.676 0.861 Yes Yes
Pro4 0.871 0.76 Yes Yes
Comp_Agr CA10 0.804 0.65 0.619 0.795 0.796 0.867 Yes Yes
CA2 0.701 0.49 Yes Yes
Institutional Context and Entrepreneurship Typologies: The Moderator. . .

CA3 0.795 0.63 Yes Yes

CA9 0.807 0.65 Yes Yes
Discovery Pro3 0.664 0.44 0.506 0.804 0.804 0.860 Yes Yes
Pro4 0.670 0.45 Yes Yes
CA10 0.740 0.55 Yes Yes
CA2 0.701 0.49 Yes Yes
171

(continued)
Table 3 (continued)
172

Convergent validity Internal consistency reliability Discriminant validity

Structure Indicator (ind.) Loadingsi Ind. rel.i1 AVE CAi2 rho_Ai3 CRi4 F&Li5 HTMTi6
CA3 0.750 0.56 Yes Yes
CA9 0.737 0.54 Yes Yes
Reg. Reg1 0.667 0.45 0.588 0.764 0.770 0.850 Yes Yes
Reg3 0.774 0.60 Yes Yes
Reg4 0.827 0.69 Yes Yes
Reg5 0.790 0.63 Yes Yes
Norm Nor1 0.843 0.72 0.667 0.750 0.762 0.857 Yes Yes
Nor2 0.857 0.74 Yes Yes
Nor3 0.746 0.56 Yes Yes
Cog. Cog1 0.857 0.74 0.696 0.565 0.571 0.821 Yes Yes
Cog3 0.811 0.66 Yes Yes
Ins_Cont Reg1 0.601 0.37 0.404 0.750 0.819 0.857 Yes Yes
Reg3 0.646 0.42 Yes Yes
Reg4 0.705 0.50 Yes Yes
Reg5 0.686 0.48 Yes Yes
Nor1 0.729 0.54 Yes Yes
Nor2 0.697 0.49 Yes Yes
Nor3 0.581 0.34 Yes Yes
Cog1 0.549 0.31 Yes Yes
Cog3 0.484 0.24 Yes Yes
Notes: i: >= 0.708 (0.40 could be accepted); i1: square of external loads >= 0.5; i2: 0.60 < reliability <0.90; i3: Cronbach’s alpha; i4: 0.60 < composite
reliability <0.90; i5 Fornell and Larcker: Is the square root of each block’s AVE greater than the correlations with all other structures? i6: HTMT the heterotrait-
monotrait ratio. Is it considerably lower than 0.85 (could be 0.90) (Hair et al. 2022)?
S. Korkarer et al.
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 173

Fig. 2 Structural equation model for the relationship between institutional context and EA

next. The final step is assessment of out-of-sample prediction (Hair et al. 2022).
Figure 2 displays the structural model results.
The VIF value with the constructs predicting each internal structure is below the
specified criterion (VIF < 3; Hair et al. 2022). To calculate the significances and the
explained variance in the dependent variable, the algorithm was run by choosing 5%
significance level, percentile method, one-tailed, complete bootstrapping with 10,000
bootstrap samples as suggested. The t-value, p-value and confidence intervals meet
the specified criteria (t > 1.65; p < 0.05; CI does not contain 0 or 1). Table 4 shows
that the perceived IC has a relatively higher effect on creative EA (β = 0.142;
p < 0.05) than discovery EA (β = 0.079; p < 0.05), but it can be said that it has a low
impact on both actions ( far from the absolute value 1).
The next step is to analyze the model’s explanatory power, in-sample predictive
power, (R2 and f 2) (Table 5). The model explains discovery (0.006 < 0.25) and
creative EA (0.020 < 0.25) as weak (Hair et al. 2022, p. 219). However, the R2 value
is context dependent, and a value above 0.26 indicates the model has substantial
explanatory power (Cohen 1988; Hair et al. 2019). The f 2 effect size represents the
likelihood the relationship (β) is in fact present in the population (Hair et al. 2019).
Examination shows IC has a statistically significant positive effect on discovery and
creative EA. In addition, the effect is small on discovery EA (0.006) and medium for
creative EA (0.021) (Table 5).
Lastly, structural model out-of-sample predictive power was evaluated using
PLSpredict (Hair et al. 2022; Manley et al. 2021). Based on the PLSpredict analysis,
174 S. Korkarer et al.

Table 4 Loadings and signiﬁcance

Sig.
Std. p- level CI
Std. β error t-value value 5% 95% f2 Decision
Creative → 0.878 0.010 92.347 0.000 Yes (0.862–0.893) 3.381
Auto
Creative → 0.852 0.010 81.5592 0.000 Yes (0.834–0.869) 2.638
Inno
Discovery 0.931 0.004 219.931 0.000 Yes (0.924–0.938) 6.491
→
Comp_Agr
Discovery 0.768 0.017 44.286 0.000 Yes (0.738–0.795) 1.434
→ Proac
Ins_Cont 0.621 0.030 21.026 0.000 Yes (0.570–0.667) 0.627
→ Cog.
Ins_Cont 0.142 0.039 3.629 0.000 Yes (0.077–0.204) 0.021 Supported
→ Creative
Ins_Cont 0.079 0.038 2.076 0.022 Yes (0.016–0.142) 0.006 Supported
→
Discovery
Ins_Cont 0.824 0.014 60.061 0.000 Yes (0.800–0.845) 2.112
→ Norm
Ins_Cont 0.862 0.012 73.714 0.000 Yes (0.841–0.880) 2.879
→ Reg.

Table 5 Model’s explanatory power

R2 t-value p-value CI 5%–95% f2
Discovery 0.006 0.997 0.160 0.000–0.020 0.006
Creative 0.020 1.806 0.035 0.006–0.042 0.021
Notes: R2: 0 <= R2 <= 1; weak if 0.25 <= R2 < 0.50; medium if 0.50 <= R2 < 0.75; signiﬁcant if
R2 >= 0.75) (Hair et al. 2022, p. 219). The f 2 is small if 0.02 <= f 2 < 0.15; medium if
0.15 <= f 2 < 0.35; large if f 2 >= 0.35 (Cohen 1988; Hair et al. 2019)

the Q2 values are greater than zero (Hair et al. 2022, p. 222). A majority of the RMSE
values from the PLS-SEM analysis are lower than the naïve LM benchmark RMSE
values. If a majority of the dependent construct’s indicators produces lower
PLS-SEM prediction errors compared to the naïve LM benchmark, this indicates
the model has moderate predictive power (Shmueli et al. 2019). In this context, the
model’s prediction power is considered good.

Moderator Effect

Table 6 displays the moderation results. The t-value, p-value, and confidence
intervals do not meet the specified criteria according to the loadings and significance
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 175

Table 6 Loadings and signiﬁcance

Sig. level CI
Hypothesis Std. β t-value p-value 5% 5%–95% Decision
H1a 0.007 0.221 0.413 No -0.049 to 0.060 Rejected
H2a -0.015 0.437 0.331 No -0.074 to 0.041 Rejected

test (t > 1.645; p < 0.05; CI contains either 0 or 1). Thus, no moderator effect was
found for the relationship between IC and discovery EA (β = 0.007; t = 0.221;
p = 0.413) and creative EA (β = -0.015; t = 0.437; p = 0.331) on EU.

4 Discussion

The risk-taking dimension was excluded from the analysis since it creates a different
structure. Some studies in Turkey also have the same results (e.g., Aslan, 2015;
Demiroğlu, 2007). Although it is stated that taking risks in entrepreneurship is an
important advantage, some studies show that risk taking does not affect entrepre-
neurial behavior as much as other dimensions or even effect negatively. Therefore,
risk taking is stated as related to national culture, organizational, and regulatory
contexts (Naldi et al. 2007), as the vital importance of risk management in Turkish
culture is not seized and there is no clear perception about the positive or negative
impact on the company.
The findings are consistent with previous studies. Turkish companies prefer to
avoid uncertainty (Elbanna et al. 2014), remain on the safe side (Benzing et al.
2009), and take advantage of existing opportunities instead of emphasizing creativity
(Bozkurt and Baştürk 2009). Based on the level of agreement of values, these
findings are supported since the answers tend to be toward to discovery EA
(discovery EA:3.54; creative EA:3.10). In contrast, the level of agreement with the
IC (2.81) in terms of doing business and entrepreneurship in Turkey is negatively
related.
Moreover, Turkish firms tend to not consider IC in acting on entrepreneurial
opportunities, but IC is related to creative EA (R2 = 0.020) more so than discovery
EA (R2 = 0.006). Contrary to some studies which report that institutional practices
such as taxes, bribery, difficulty in accessing information, formal procedures, rules,
etc. are ineffective and in fact discourage companies from taking creative EA
(De Soto 2000; Johnson et al. 2000), this research reveals negative perceptions of
the institutional environment and instability motivate Turkish managers to empha-
size more creative EA than discovery. This is in line with reports that firms exhibit
more negative perceptions of their IC and motivate them to be alert and produce
innovative products (Lumpkin and Dess 2001; Sundqvist et al. 2012).
Manager’s agreement level with the external environment (2.89) is also low,
while our analysis shows no moderating effect of EU on the interaction of IC with
EAs. Studies on the EU for EA in the entrepreneurship literature have not yielded
176 S. Korkarer et al.

consistent results, likely due to the differences between the objective characteristics
of the firm’s environment and the perceived uncertainty of the environment (Elbanna
and Child 2007a; Milliken 1987). Likewise, Elbanna and Child (2007b)’s study and
our results contradict some other studies, which could be due to the other studies not
explicitly assessing EA. Other possible sources of the inconsistencies could be
variations between the proposed outcome constructs and the context-specific differ-
ences. Finally, managers are used to the instability and environmental fluctuations,
so they may discount the environment as a factor influencing EA.
It is also worth noting that the participants answered the survey in the Covid-19
pandemic (March 2021–May 2021), while the Turkish economy was in a decline
since 2015 due to various crises in Turkey (ITA 2022). These situations have led to
an increase in investor uncertainties and risk perceptions specific to Turkey (OECD
2021). Thus, the basic conditions that companies should be able to make free
decisions have diminished (Öztürkler 2020). This situation quite likely plays a
major role in the lack of significance of negative perceptions of the context and
environment and its effects on EA.

4.1 Observations and Conclusions

The results lead us to propose revisions in our initial model for future research of the
antecedents of EAs in Turkish autonomous private companies. First, the effects of
the dimensions of the IC (Hunt and Ortiz-Hunt 2018) could be studied separately in
future studies. This could bring out the most weighted context (regulative, norma-
tive, cognitive) effecting EAs. Second, the findings indicate the Turkish companies
take both EAs nearly at the same level. This would support the context dependent
EA (Greenwood et al. 2002; McMullen and Shepherd 2006; Meyer and Rowan
1977; Milliken 1987; Schumpeter 1934; Scott and Meyer 1994). To better differen-
tiate and understand the impact of the IC, longitudinal, mixed method studies would
be useful. The relationship between IC and the EAs is complex phenomenon
(Sarasvathy et al. 2003; Teece 2007). Therefore, control variables such as organiza-
tional characteristics, technological dynamism, or moderating variables should be
added.
In addition, within an unstable, uncertain market and weak IC particularly due to
the Covid-19 pandemic, some Turkish companies such as LcWaikiki, İpekyol,
Bisse, and Vakko could succeed to evolve with their product portfolio. Also, digital
workforce platforms have emerged such as Gooby, Getir, Istegelsin, and Banabi
(İlhan 2020). Getir is one of the most adorable companies as it is the world’s first
ultrafast grocery delivery. Therefore, qualitative studies could be executed to eval-
uate the impact of the internal capabilities and particularly cognitive aspects.
Yet there were other findings for which more research is necessary to verify the
moderating role of EU in different cultures. It is useful to compare the results of our
findings with those reported by Elbanna (2007b) since both studies have outcome
constructs other than firm performance and top management strategic decision-
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 177

making in different contexts (Turkey and Egypt). Both show that strategic decision-
making has various significant antecedents with EU playing a less significant role.
Although the research is designed very carefully, there are limitations. The most
important limitation is the application of established scales from another context. At
least in the light of Whitley’s (1996) national business systems typologies related to
the structures that are the subject of the research, context-specific scale development
studies would likely be revealing, and the structural model relationships can be
retested in the light of more context-specific scales. Also, cross-sectional research
was conducted through a questionnaire of the theoretical model. First, the perceptual
measures employed may not truly reflect the phenomenon of interest. It is likely that
some information, such as political tactics and unsatisfactory decision outcomes,
were not considered in the answers. Therefore, the results should be interpreted with
caution. To reduce this limitation, several procedures were followed. To reduce
response bias, (1) the independent variable was prioritized in the survey, (2) the
answer collection method was diversified by changing the scale formats, and a
5-point Likert scale was used, (3) a moderator was included to further examine the
relevant relationships, and (4) assurances of complete anonymity and confidentiality
were given. Lastly, the population is determined as commercial firms that make
autonomous decisions in Turkey. Unfortunately, there is no list of autonomous
decision-making firms. For this reason, survey data were collected by a
non-probability sampling method from those who could be reached through the
researcher’s network connections and those who agreed to participate. In future
research, it would be beneficial to include criteria in the sampling methodology to
screen out nonqualified respondents.

References

Adner R, Helfat CE (2003) Corporate effects and dynamic managerial capabilities. Strateg Manag J
24(10):1011–1025
Alvarez SA, Barney JB (2007) Discovery and creation: alternative theories of entrepreneurial
action. Strateg Entrep J 1(1–2):11–26
Alvarez SA, Barney JB (2020) Has the concept of opportunities been fruitful in the ﬁeld of
entrepreneurship? Acad Manag Perspect 34(3):1–26
Alvarez C, Urbano D (2011) Environmental factors and entrepreneurial activity in Latin America.
Academia Revista Latinoamericana de Administración 48(1):126–139
Alvarez SA, Audretsch D, Link AN (2016) Advancing our understanding of theory in entrepre-
neurship. Strateg Entrep J 10(1):3–4
Aslan G (2015) Dinamik Yetenekler, Stratejik İmprovizasyon ve Firma Performansı. Yaşar
Üniversitesi, İzmir
Babin BJ, Grifﬁn M, Hair JF (2016) Heresies and sacred cows in scholarly marketing publications. J
Bus Res 69(8):3133–3138
Baum JAC, Oliver C (1991) Institutional linkages and organizational mortality. Adm Sci Q 36:187–
218
Benzing C, Chu HM, Kara O (2009) Entrepreneurs in Turkey: a factor analysis of motivations,
success factors, and problems. J Small Bus Manag 47(1):58–91
178 S. Korkarer et al.

Bozkurt V, Baştürk Ş (2009) KOBİ Girişimcilerinde Risk Ve Belirsizlik Algilari: Bursa Örneği.
Ankara Üniversitesi SBF Dergisi 64(2):43–74
Brislin RW (1980) Translation and content analysis of oral and written materials. In: Triandis HC,
Berry JW (eds) Handbook of cross-cultural psychology: methodology, vol 2. Allyn and Bacon,
Boston, MA, pp 389–444
Buchanan JM, Vanberg VJ (1991) The market as a creative process. In: Hausman DM (ed) The
philosophy of economics: an anthology, 3rd edn. Cambridge University Press, Cambridge, pp
378–398
Bugra A (2007) Poverty and citizenship: an overview of the social-policy environment in repub-
lican Turkey. Int J Middle East Stud 39(1):33–52
Busenitz LW (1996) Research on entrepreneurial alertness. J Small Bus Manag 34(4):35–44
Busenitz LW, Gomez C, Spencer JW (2000) Country institutional profiles: unlocking entrepre-
neurial phenomena. Acad Manag J 43(5):994–1003
Chin WW, Marcolin BL, Newsted PR (2003) A partial least squares latent variable modeling
approach for measuring interaction effects: results from a Monte Carlo simulation study and an
electronic-mail emotion/adoption study. Inf Syst Res 14(2):189–217
Cohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum
Associates, Hillsdale, NJ
Cohen B, Winn MI (2007) Market imperfections, opportunity and sustainable entrepreneurship. J
Bus Ventur 22(1):29–49
Coleman JS (1986) Social theory, social research, and a theory of action. Am J Sociol 91(6):
1309–1335
Coşkun R, Altunişik R, Yildirim E (2019) Sosyal Bilimlerde Araştırma Yöntemleri SPSS
Uygulamalı, 10th edn. Sakarya Yayıncılık, Sakarya
Creswell JW, Creswell JD (2018) Research design: qualitative, quantitative, and mixed methods
approaches, 5th edn. Sage Publications, Los Angeles
Daft RL, Weick KE (1984) Toward a model of organizations as interpretation systems. Acad Manag
Rev 9(2):284–295
de Jong JPJ, Marsili O (2015) The distribution of Schumpeterian and Kirznerian opportunities.
Small Bus Econ 44(1):19–35
De Soto H (2000) The mystery of capital: why capitalism triumphs in the west and fails everywhere
else. Basic Books, New York
Dean TJ, McMullen JS (2007) Toward a theory of sustainable entrepreneurship: reducing environ-
mental degradation through entrepreneurial action. J Bus Ventur 22(1):50–76
Demiroğlu Y (2007) Firma Davranışı Olarak Girişimcilik. Master, DEÜ Sosyal Bilimleri Enstitüsü,
Turkey
Dew N, Velamuri SR, Venkataraman S (2004) Dispersed knowledge and an entrepreneurial theory
of the firm. J Bus Ventur 19(5):659–679
Duncan RB (1972) Characteristics of organizational environments and perceived environmental
uncertainty. Adm Sci Q 17(3):313–327
Elbanna S, Child J (2007a) The influence of decision, environmental and firm characteristics on the
rationality of strategic decision-making. J Manag Stud 44(4):561–591
Elbanna S, Child J (2007b) Influences on strategic decision effectiveness: development and test of
an integrative model. Strateg Manag J 28(4):431–453
Elbanna S, Thanos IC, Colak M (2014) An exploratory study of the determinants of the quality of
strategic decision implementation in Turkish industrial firms. J Gen Manag 40(2):27–46
Etzioni A (1987) Entrepreneurship, adaptation and legitimation: a macro-behavioral perspective. J
Econ Behav Organ 8(2):175–189
Filser M, Tiberius V, Kraus S, Zeitlhofer T, Kailer N, Müller A (2020) Opportunity recognition:
conversational foundations and pathways ahead. Entrep Res J 13(1):1–30
Fleming L, Mingo S, Chen D (2007) Collaborative brokerage, generative creativity, and creative
success. Adm Sci Q 52(3):443–475
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 179

Foss NJ, Klein PG (2020) Entrepreneurial opportunities: who needs them? Acad Manag Perspect
34(3):366–377
García-Morales VJ, Jiménez-Barrionuevo MM, Gutiérrez-Gutiérrez L (2012) Transformational
leadership influence on organizational performance through organizational learning and inno-
vation. J Bus Res 65(7):1040–1050
Gartner WB (1985) A conceptual framework for describing the phenomenon of new venture
creation. Acad Manag Rev 10(4):696–706
Gaur AS, Lu JW (2007) Ownership strategies and survival of foreign subsidiaries: impacts of
institutional distance and experience. J Manag 33(1):84–110
Gómez-Haro S, Aragón-Correa JA, Cordón-Pozo E (2011) Differentiating the effects of the
institutional environment on corporate entrepreneurship. Manag Decis 49(10):1677–1693
Greenwood R, Hinings CR (1996) Understanding radical organizational change: bringing together
the old and the new institutionalism. Acad Manag Rev 21(4):1022–1054
Greenwood R, Suddaby R, Hinings CR (2002) Theorizing change: the role of professional
associations in the transformation of institutionalized fields. Acad Manag J 45(1):58–80
Grove H, Clouse M, Schaffner LG (2018) Digitalization impacts on corporate governance. J Gov
Regul 7(4):51–63
Hair JF, Black WC, Babin BJ, Anderson RE, Tatham RL (2014) Multivariate Data Analysis, 7th
edn. Pearson Education Limited, Harlow, Essex
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Howard MC, Nitzl C (2020) Assessing measurement model quality in PLS-SEM using
confirmatory composite analysis. J Bus Res 109:101–110
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks, CA
Hedström P, Swedberg R (1998) Social mechanisms: an introductory essay. In: Hedström P,
Swedberg R (eds) Social mechanisms: an analytical approach to social theory. Cambridge
University Press, Cambridge, pp 1–31
Huber GP, Power DJ (1985) Retrospective reports of strategic-level managers: guidelines for
increasing their accuracy. Strateg Manag J 6(2):171–180
Hunt RA, Ortiz-Hunt LL (2018) Deinstitutionalization through business model evolution: women
entrepreneurs in the Middle East and North Africa. In: Ladislav M (ed) Entrepreneurship:
development tendencies and empirical approach. IntechOpen, London, pp 431–460
İlhan L (2020, 6 April) 300 firma maske üretimine girdi. Dünya Gazetesi. From https://www.dunya.
com/ekonomi/300-firma-maske-uretimine-girdi-haberi-466924 Accessed 26 Aug 2022
ITA (2022, 26 July) Turkey country commercial guide: market overview. International Trade
Administration. https://www.trade.gov/country-commercial-guides/turkey-market-overview
Accessed 26 Aug 2022
Iturrioz C, Aragón C, Narvaiza L (2015) How to foster shared innovation within SMEs' networks:
social capital and the role of intermediaries. Eur Manag J 33(2):104–115
Jambulingam T, Kathuria R, Doucette WR (2005) Entrepreneurial orientation as a basis for
classification within a service industry: the case of retail pharmacy industry. J Oper Manag
23(1):23–42
Jaworski BJ, Kohli AK (1993) Market orientation: antecedents and consequences. J Mark 57(3):
53–70
Johnson S, McMillan J, Woodruff C (2000) Entrepreneurs and the ordering of institutional reform:
Poland, Slovakia, Romania, Russia and Ukraine compared. Econ Transit 8(1):1–36
Kets de Vries MFR (1977) The entrepreneurial personality: a person at the crossroads. J Manag
Stud 14(1):34–57
Khandwalla PN (1976) Some top management styles, their context and performance. Organ Adm
Sci 7(4):21–51
Khandwalla PN (1987) Generators of pioneering-innovative management: some Indian evidence.
Organ Stud 8(1):39–59
180 S. Korkarer et al.

Kirzner IM (1973) Competition and entrepreneurship. University of Chicago Press, Chicago, IL

Knight FH (1964) Risk, uncertainty and profit. Sentry Press, New York
Lindsay NJ, Craig JB (2002) A framework for understanding opportunity recognition: entrepre-
neurs versus private equity financiers. J Priv Equity 6(1):13–24
Lumpkin TG, Dess GG (1996) Clarifying the entrepreneurial orientation construct and linking it to
performance. Acad Manag Rev 21(1):135–172
Lumpkin GT, Dess GG (2001) Linking two dimensions of entrepreneurial orientation to firm
performance: the moderating role of environment and industry life cycle. J Bus Ventur 16(5):
429–451
Manley SC, Hair JF, Williams RI, McDowell WC (2021) Essential new PLS-SEM analysis
methods for your entrepreneurship analytical toolbox. Int Entrep Manag J 17(4):1805–1825
McMullen JS, Shepherd DA (2006) Entrepreneurial action and the role of uncertainty in the theory
of the entrepreneur. Acad Manag Rev 31(1):132–152
Meyer JW, Rowan B (1977) Institutionalized organizations: formal structure as myth and cere-
mony. Am J Sociol 83(2):340–363
Milliken FJ (1987) Three types of perceived uncertainty about the environment: state, effect, and
response uncertainty. Acad Manag Rev 12(1):133–143
Murphy PJ (2011) A 2×2 conceptual foundation for entrepreneurial discovery theory. Entrep
Theory Pract 35(2):359–374
Naldi L, Nordqvist M, Sjöberg K, Wiklund J (2007) Entrepreneurial orientation, risk taking, and
performance in family firms. Fam Bus Rev 20(1):33–47
North DC (1991) Institutions. J Econ Perspect 5(1):97–112
OECD (2021, 1 Jan 2021) OECD economic surveys: Turkey. OECD. https://www.oecd.org/
economy/surveys/TURKEY-2021-OECD-economic-survey-overview.pdf Accessed
26 Aug 2022
Oliver C (1991) Strategic responses to institutional processes. Acad Manag Rev 16(1):145–179
Öztürkler H (2020) Covid-19’un Türkiye Ekonomisine Etkileri ce Politika Önlemlerinin Ekonomik
Özgürlükler Açisindan Değerlendirilmesi (17). Ö. A, Derneği
Podsakoff PM, MacKenzie SB, Podsakoff NP, Lee JY (2003) The mismeasure of man (argument)
and its implications for leadership research. Leadersh Q 14(6):615–656
Ringle CM, Sarstedt M, Straub DW (2012) Editor's comments: a critical look at the use of
PLS-SEM in “MIS Quarterly”. In: Burton-Jones A (ed) Management information systems
quarterly, vol 36. Management Information Systems Research Center, University of Minnesota,
Minneapolis, pp iii–xiv
Sarasvathy SD (2001) Causation and effectuation: toward a theoretical shift from economic
inevitability to entrepreneurial contingency. Acad Manag Rev 26(2):243–263
Sarasvathy SD, Dew N, Velamuri SR, Venkataraman S (2003) Three views of entrepreneurial
opportunity. In: Acs ZJ, Audretsch DB (eds) Handbook of entrepreneurship research. Springer,
London, pp 77–98
Sarstedt M, Ringle CM, Hair JF (2017) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg A (eds) Handbook of market research, vol 26. Springer,
Cham, pp 1–40
Schumpeter JA (1934) Theory of economic development: an inquiry into profits, capital, credit,
interest and the business cycle. Harvard University Press, Cambridge, MA
Scott WR (1995) Institutions and organizations, vol 2. Sage, Thousand Oaks, CA
Scott WR, Meyer JW (1994) Institutional environments and organizations: structural complexity
and individualism. Sage, Thousand Oaks, CA
Shane S, Venkataraman S (2000) The promise of entrepreneurship as a field of research. Acad
Manag Rev 25(1):217–226
Shepherd DA, Patzelt H (2011) The new field of sustainable entrepreneurship: studying entrepre-
neurial action linking “what is to be sustained” with “what is to be developed”. Entrep Theory
Pract 35(1):137–163
Institutional Context and Entrepreneurship Typologies: The Moderator. . . 181

Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Smith BR, Matthews CH, Schenkel MT (2009) Differences in entrepreneurial opportunities: the
role of tacitness and codification in opportunity identification. J Small Bus Manag 47(1):38–57
Su J, Zhai Q, Karlsson T (2017) Beyond red tape and fools: institutional theory in entrepreneurship
research, 1992–2014. Entrep Theory Pract 41(4):505–531
Suddaby R (2013) Institutional theory. In: Kessler EH (ed) Encyclopedia of management theory.
Sage, Thousand Oaks, CA, pp 379–383
Sundqvist S, Kyläheiko K, Kuivalainen O, Cadogan JW (2012) Kirznerian and Schumpeterian
entrepreneurial-oriented behavior in turbulent export markets. Int Mark Rev 29(2):203–219
Tabachnick BG, Fidell LS, Ullman JB (2013) Using multivariate statistics, vol 6. Pearson,
Boston, MA
Taş A, Hızıroğlu M (2016) İdeolojinin kurumsal değişimdeki rolü: Eğitim kurumu üzerinden
Türkiye eksenli bir değerlendirme. İşletme Bilimi Dergisi 4(2):71–104
Teece DJ (2007) Explicating dynamic capabilities: the nature and microfoundations of (sustainable)
enterprise performance. Strateg Manag J 28(13):1319–1350
Teece DJ, Pisano G, Shuen A (1997) Dynamic capabilities and strategic management. Strateg
Manag J 18(7):509–533
Tolbert PS, David RJ, Sine WD (2011) Studying choice and change: the intersection of institutional
theory and entrepreneurship research. Organ Sci 22(5):1332–1344
Vaghely IP, Julien P-A (2010) Are opportunities recognized or constructed?: an information
perspective on entrepreneurial opportunity identification. J Bus Ventur 25(1):73–86
Venkataraman S (1997) The distinctive domain of entrepreneurship research. In: Katz J
(ed) Advances in entrepreneurship, firm emergence and growth, vol 3. JAI Press, New York,
pp 119–138
Weick KE (1979) The social psychology of organizing, 2nd edn. McGraw-Hill Humanities,
New York
Weick KE, Roberts KH (1993) Collective mind in organizations: heedful interrelating on flight
decks. Adm Sci Q 3:357–381
Welter CT (2012) Distinguishing opportunity types: why it matters and how to do it. Ph.D. thesis,.
The Ohio State University, Ohio
Whitley R (1996) Business systems and global commodity chains: competing or complementary
forms of economic organisation? Compet Change 1(4):411–425
Wiklund J, Shepherd D (2003) Knowledge-based resources, entrepreneurial orientation, and the
performance of small and medium-sized businesses. Strateg Manag J 24(13):1307–1314
Zablah AR, Franke GR, Brown TJ, Bartholomew DE (2012) How and when does customer
orientation influence frontline employee job outcomes? A meta-analytic evaluation. J Mark
76(3):21–40
Zhang Y, Yang J, Liu M (2022) Enterprises’ energy-saving capability: empirical study from a
dynamic capability perspective. Renew Sust Energ Rev 162:1–13
PLS-Based Evaluation of a Digital
Transformation Adoption Model
for Biopharma Manufacturing

Frederick K. Johnson and Chinazunwa C. Uwaoma

1 Introduction

Biopharmaceutical (biopharma) companies in the United States are actively

transforming their manufacturing operations, given the devastating effects of the
COVID-19 pandemic (Ford 2020). This crisis-driven transformation comes from
Pharma 4.0 initiatives as digital transformation (DX) strategies (Yang et al. 2020).
One critical aspect of achieving transformation success is determining which DTF
can effectively meet the demands of a biopharma manufacturing environment
(Coundouris 2020). This chapter addresses the selection and adoption problems by
proposing an adoption model designed to assist IT leaders and practitioners choose
the best framework.

2 Background

We surveyed a sample population of 210 executives, managers, IT leaders, consul-

tants, and academic professionals working in the biopharma industry by asking them
to evaluate the usefulness of our proposed DTF. The primary data collection method
used a Qualtrics web-based survey from October 19 to November 1, 2021, using the
Cinetement research platform. During the survey, we introduced participants to the
“dxn Conceptual Overview” seen in Fig. 1 and the “dxn Strategic Roadmap” in

F. K. Johnson (✉) · C. C. Uwaoma

Claremont Graduate University, Center for Information Systems and Technology, Claremont,
CA, USA
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 183

Fig. 1 The dxn conceptual overview

Fig. 2 The dxn strategic road map

Fig. 2, representing two of the nine artifacts within the dxn Biopharma Digital
Transformation Framework.

2.1 Assessing Perceived Usefulness and Intention to Use

Participants evaluated the model based on their professional experience. The results
successfully established the proposed artifact’s utility, using a seven-point Likert
scale, ranging from “Strongly agree” to “Strongly disagree.” The analysis demon-
strates that the most frequent response (i.e., the mode) was “agree,” indicating the
PLS-Based Evaluation of a Digital Transformation Adoption Model. . . 185

usefulness of the proposed artifacts and the respondent’s intentions to adopt the
proposed artifacts.

3 The Digital Transformation Adoption Model

Using a structural equation (SEM) model, this chapter analyzes data collected in our
previous research project to develop a DX adoption model for biopharma
manufacturing. The primary constructs include elements from the technology accep-
tance model (TAM) by Davis et al. (1989) and hallmark functional and structural
elements of top DX models presented by Coundouris (2020).

3.1 Hypothesis

The proposed model includes intention to use (IU) as a function of perceived ease of
use (EU), job relevance (JR), and perceived usefulness (PU) as a function of
hallmark functional elements (HFE) and hallmark structural elements (HSE) of the
top DTFs. Table 1 and Fig. 3 depict each construct’s associated hypotheses and
identiﬁed relationships.

3.2 Sample Population and Demographics

We collected feedback from 210 executives and managers in the biotech and
pharmaceutical industries. The results show that over 57% of all respondents were
currently involved in research or implementations related to DX or Pharma 4.0
programs. Given the ﬁgures in Tables 2 and 3, the survey successfully targeted the
correct sample population based on relevant activities, job functions, and industries.

Table 1 Hypotheses
Hypotheses IV → DV Effect
H1 Perceived usefulness → Intention to use Positive
H2 Job relevance → Intention to use Positive
H3 Perceived ease of use → Intention to use Positive
H4 Hallmark functional elements → Perceived usefulness Positive
H5 Hallmark structural elements → Perceived usefulness Positive
186 F. K. Johnson and C. C. Uwaoma

HS1
PU1 PU3 PU5
HS2

HS3 H5, +

HS4 HSE

HFE3 H1, +
H4, + PU
HFE4

HFE JB2 H2, + IU1

JB3 IU2

JR IU

EU1 H3, +

EU2

EU5
EU

Fig. 3 Propose PLS-SEM digital transformation adoption model

Table 2 Sample population demographics

Respondents categories Frequency Percent
Executives 64 30.48
Managers and above in non-IT roles 74 35.24
Managers or IT leaders 40 19.05
Consultant 15 7.14
Academic 17 8.10
Total 210 100.0

Table 3 Sample population current digital transformation activity

Involved in digital transformation efforts Frequency Percent
Yes 121 57.62
No 89 42.38
Total 210 100.0

4 Results and Discussion

Using PLS analysis, this chapter assessed the proposed structural equation model in
Fig. 3 in two phases. Within this chapter, we evaluated the measurement model in
the first phase and the structural model in the second phase (Ringle et al. 2020). After
satisfying the measurement model requirements, we assess the structural model in
determining the significance of each construct’s path coefficient using the SmartPLS
bootstrapping method (Shehzad et al. 2021). Table 4 presents the results.
PLS-Based Evaluation of a Digital Transformation Adoption Model. . . 187

Table 4 Reliability and validity results

Constructs Items Loadings t-statistics VIF Cronbach’s α rho_A CR AVE
Intention to use 0.915 0.917 0.959 0.921
IU1 0.958 135.117 3.451
IU2 0.962 176.318 3.451
Perceived usefulness 0.871 0.878 0.920 0.794
PU1 0.896 57.697 2.231
PU3 0.881 42.608 2.555
PU5 0.896 50.906 1.929
Hallmark structural 0.848 0.856 0.897 0.686
elements
HSE1 0.866 48.117 2.102
HSE2 0.819 27.245 1.905
HSE3 0.821 26.084 1.912
HSE4 0.807 27.211 1.769
Hallmark functional 0.773 0.777 0.898 0.815
elements
HFE3 0.894 41.106 1.660
HFE4 0.911 60.395 1.660
Job relevance 0.916 0.916 0.960 0.922
JB2 0.960 114.299 3.489
JB3 0.960 96.694 3.489
Perceived ease of use 0.924 0.925 0.952 0.868
EU1 0.931 84.337 3.607
EU2 0.929 68.187 3.298
EU5 0.935 74.052 3.667
Notes: IU Intention to use, PU Perceived usefulness, JR Job relevance, EU Perceived ease of use,
HSE Hallmark structural elements, HSF Hallmark functional elements

4.1 Measurement Model Assessment

The first analysis phase validated the construct’s reliability and validity by evaluat-
ing the measurement model. Determining the reliability and convergent validity of
the measurement model means measuring the loadings of items for each construct
for items reliability, Cronbach’s alpha for the construct’s reliability (Cα), the com-
posite reliability (CR), the rho_A (rho A), and the average variance extracted (AVE;
Hair Jr et al. 2022). The thresholds for each calculated measurement must be greater
than 0.7, 0.7, 0.7, and 0.5 (Ali et al. 2018).
The assessment of the measurement model across the six reflective constructs
revealed satisfactory reliability and convergent validity for each construct. Since all
variance inflation factor (VIF) values are less than five, common method bias
variance (MBV) is not a problem (Kock 2015), as seen in Table 4.
Next, we measured the cross-loading values. In Table 5, all six items load firmly
to their respective constructs based on cross-loading factors. Since the loading
factors for all constructs are above 0.6, the model has discriminant validity (Fig. 4)
(Awang 2014).
188 F. K. Johnson and C. C. Uwaoma

Table 5 Discriminant validity: cross loadings

Items IU EU HSE JR PU HFE
IU1 0.957 0.756 0.740 0.721 0.756 0.618
IU2 0.962 0.819 0.778 0.759 0.786 0.686
EU1 0.739 0.931 0.678 0.651 0.653 0.617
EU2 0.788 0.929 0.714 0.661 0.778 0.681
EU5 0.768 0.935 0.755 0.656 0.712 0.673
HS1 0.787 0.813 0.866 0.648 0.736 0.633
HS2 0.550 0.532 0.819 0.567 0.583 0.470
HS3 0.636 0.614 0.821 0.589 0.607 0.533
HS4 0.622 0.555 0.807 0.572 0.628 0.594
JB2 0.741 0.644 0.685 0.960 0.689 0.618
JB3 0.741 0.708 0.696 0.960 0.651 0.575
PU1 0.778 0.727 0.735 0.676 0.897 0.663
PU3 0.624 0.585 0.589 0.545 0.881 0.605
PU5 0.731 0.725 0.733 0.631 0.895 0.705
HFE3 0.629 0.672 0.613 0.579 0.641 0.894
HFE4 0.601 0.605 0.611 0.544 0.695 0.911
Notes: IU Intention to use, PU Perceived usefulness, JR Job relevance, EU Perceived ease of use;
HSE Hallmark structural elements, HSF Hallmark functional elements

HS1
0.866 PU1 PU3 PU5
HS2 0.819
0.821
HS3 0.896 0.881 0.896
0.807
HSE 0.505
HS4

0.684
0.399
HFE3 0.894
0.911
PU 0.311
HFE4

HFE JB2 0.960 0.958

IU1
0.286 0.780
0.960 0.962
JB3 IU2
JR IU
0.381

EU1
0.931
EU2 0.929
0.935
EU5
EU

Fig. 4 The measurement model

Finally, according to the heterotrait-monotrait (HTMT) ratios guidelines, HTMT

ratios less than 0.90 are acceptable, as seen in Table 6 (Hair Jr et al. 2022).
After achieving discriminant validity, we estimated the coefﬁcient of determina-
tion (R2 Adj), predictive relevance (Q2) and effect size ( f2) (see Table 7). The R2 Adj
and Q2 values for intention to use are 0.780 (Q2 = 0.704), and the perceived
usefulness is 0.684 (Q2 = 0.527), which supports the model’s in-sample predictive
power (Sarstedt et al. 2014). The results of blindfolding with an omission distance of
PLS-Based Evaluation of a Digital Transformation Adoption Model. . . 189

Table 6 Discriminant validity: heterotrait-monotrait ratios (HTMT)

Constructs IU EU HFE HSE JR PU
IU
EU 0.892
HFE 0.809 0.836
HSE 0.889 0.856 0.831
JR 0.842 0.765 0.739 0.813
PU 0.893 0.848 0.898 0.889 0.776
Notes: IU Intention to use, PU Perceived usefulness, JR Job relevance, EU Perceived ease of use,
HSE Hallmark structural elements, HSF Hallmark functional elements

Table 7 Coefﬁcient of determination, predictive relevance, and effect size

f2 R2 Adj Q2
Endogenous
Constructs Intention to use Perceived usefulness constructs
Intention to use 0.780 0.704
Perceived ease of use 0.238
Hallmark functional elements 0.275
Hallmark structural elements 0.443
Job relevance 0.168
Perceived usefulness 0.162 0.684 0.527

seven provide Q2 values considerably above zero (Table 7), confirming the model’s
predictive relevance in out-of-sample prediction (Sarstedt et al. 2014). In addition,
the f 2 indicates whether an exogenous construct affects an endogenous construct
(Rehman et al. 2020). Cohen (1988) supports three categories for f 2 values, includ-
ing minor ( f2 = 0.02), intermediate ( f2 = 0.15), and stronger influence ( f2 = 0.35).
For this chapter, the effect size ranges from medium to high.
We also checked the measurement model for predictive power using PLSpredict
(r = 10, k = 10) in SmartPLS to check the measurement model, as suggested by
Shmueli et al. (2019). Figure 5 shows that prediction errors were symmetrically
distributed so that RMSE could be used. As a result, four of the five indicators of the
endogenous variables, i.e., IU, overtook the naïve benchmarks of Q2 > 0 for both
PLS-SEM and linear regression model (LM) benchmarks (see Table 8). Then, we
compared the RMSE and MAE performance measures and found that majority of the
PLS values were lower than those in LM for RMSE and MAE, indicating a medium
predictive power (Shmueli et al. 2019).
We compared the Q2 values of PLS and LM and found that Q2 was higher for
some of PLS than for LM, and after the comparison, Q2 values remained positive.
Given these comparisons, we can conclude that PLS Predict for our model fulfilled
the required criteria; therefore, we can confidently establish the measurement
model’s predictive relevance and accuracy (Shmueli et al. 2019).
190 F. K. Johnson and C. C. Uwaoma

Fig. 5 Intention to use (IU) distribution of prediction errors

Table 8 PLSpredict assessment

PLS LM PLS-LM
Indicator RMSE MAE Q2 RMSE MAE Q2 RMSE MAE Q2
IU1 0.938 0.674 0.650 0.965 0.685 0.629 -0.027 -0.011 0.021
IU2 0.706 0.501 0.751 0.739 0.537 0.727 -0.033 -0.036 0.024
PU1 0.915 0.675 0.585 0.736 0.553 0.605 0.179 0.122 -0.020
PU3 0.865 0.641 0.456 0.904 0.639 0.595 -0.039 0.002 -0.139
PU5 0.732 0.567 0.609 0.867 0.651 0.453 -0.135 -0.084 0.156
Notes: The numbers in bold represent the lowest relative value.

Table 9 Hypotheses results

H0 IV DV Beta BCI LL BCI UL t-statistics p-values Conclusion
H1 PU IU 0.311 0.156 0.443 4.028 0.000 Supported
H2 JR IU 0.286 0.141 0.434 3.773 0.000 Supported
H3 EU IU 0.381 0.232 0.537 4.831 0.000 Supported
H4 HSE PU 0.505 0.393 0.631 8.191 0.000 Supported
H5 HFE PU 0.399 0.259 0.521 5.957 0.000 Supported
Notes: IU Intention to use, PU Perceived usefulness, JR Job relevance, EU Perceived ease of use,
HSE Hallmark structural elements, HSF Hallmark functional elements

4.2 Structural Model Assessment

After confirming the model’s reliability and convergent validity, we assessed the
structural model’s statistical significance. We calculate each construct’s direct rela-
tionships and path coefficient using the Complete SmartPLS bootstrapping sampling
with 5000 resamples (Ringle et al. 2015). Figure 5 demonstrates the results.
Furthermore, Table 9 lists the hypotheses, and Fig. 6 demonstrates that all direct
observations are positively significant. Hence, H1, H2, and H3 confirm a positive
association between perceived usefulness, job relevance, and perceived ease of use
PLS-Based Evaluation of a Digital Transformation Adoption Model. . . 191

HS1
48.568 PU1 PU3 PU5
HS2 26.674
25.945
HS3 27.487 57.405 42.471 52.875
8.191
HS4 HSE

0.684
5.957
HFE3 41.504
60.191
PU 4.028
HFE4

HFE JB2 115.981 136.946

IU1
3.773 0.780
99.034 176.090
JB3 IU2
JR IU
4.831

EU1
86.013
EU2 67.808
76.305
EU5
EU

Fig. 6 The structural model

with intention to use. The impact of perceived ease of use (β = 0.381; t = 4.831;
p < 0.000) on intention to use is stronger than the effect of perceived usefulness
(β = 0.311; t = 4.028; p < 0.000) and job relevance (β = 0.286; t = 3.773;
p < 0.000), all support H1, H2, and H3.
For the remaining hypotheses, H4 and H5, we proposed that hallmark functional
elements and hallmark structural elements are positively associated with perceived
usefulness. The impact of hallmark structural elements (β = 0.505; t = 8.191;
p < 0.001) on perceived usefulness is stronger than the effect of hallmark functional
elements (β = 0.399; t = 5.957; p < 0.000), which supports H4 and H5.

5 Conclusion

5.1 Implications of Theory and Practice

In theory, perceived usefulness generally garnered the primary focus (Venkatesh and
Davis 2000); however, our research shows perceived ease of use had the largest
effect on respondents’ willingness to adopt the dxn framework. Likewise, job
relevance and perceived usefulness had nearly the same impact on intention to
use. Still, the evidence agrees that DTFs for biopharma manufacturing environments
should include structural and functional elements based on their effect on practi-
tioners’ perceived usefulness. In practice, IT leaders and practitioners are just as
interested in quickly reaching their DX goals as in making it easier for others around
them to succeed. Practitioners are also interested in using a framework that is easy to
learn, one they can become skilled at, and transforms to their needs while not
overlooking critical steps within the DX process.
192 F. K. Johnson and C. C. Uwaoma

References

Ali F, Rasoolimanesh SM, Sarstedt M, Ringle CM, Ryu K (2018) An assessment of the use of
partial least squares structural equation modeling (PLS-SEM) in hospitality research. Int J
Contemp Hosp Manag 30(1):514–538
Awang Z (2014) A handbook on SEM for academicians and practitioners: the step by step practical
guides for the beginners. Bandar Baru Bangi, MPWS Rich Resources
Cohen JE (1988) Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Asso-
ciates, Inc, Hillsdale, NJ
Coundouris A (2020) Top 5 digital transformation frameworks in 2020. Run Frictionless. https://
runfrictionless.com/blog/top-5-digital-transformation-frameworks-2020 Accessed 26 Feb 2021
Davis FD, Bagozzi RP, Warshaw PR (1989) User acceptance of computer technology: a compar-
ison of two theoretical models. Manag Sci 35(8):982–1003
Ford J (2020) Biopharma leaders prioritize R&D, technological transformation, and global market
presence. Deloitte Insights. https://www2.deloitte.com/us/en/insights/industry/life-sciences/
pharmaceutical-industry-trends.html Accessed 26 Aug 2020
Hair JF Jr, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Kock N (2015) Common method bias in PLS-SEM: a full collinearity assessment approach. Int
E-Collab 11(4):1–10
Rehman SU, Bhatti A, Kraus S, Ferreira JJ (2020) The role of environmental management control
systems for ecological sustainability and sustainable performance. Manag Decis 59(9):
2217–2237
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3 (3.3.9). SmartPLS GmbH
Ringle CM, Sarstedt M, Mitchell R, Gudergan SP (2020) Partial least squares structural equation
modeling in HRM research. Int J Hum Resour Manag 31(12):1617–1643
Sarstedt M, Ringle CM, Smith D, Reams R, Hair JF Jr (2014) Partial least squares structural
equation modeling (PLS-SEM): a useful tool for family business researchers. J Fam Bus Strat
5(1):105–115
Shehzad MU, Zhang J, Le PB (2021) Role of collaborative culture and knowledge management
process for stimulating radical and incremental innovation: a comparative moderation approach.
Bus Process Manag J 27(7):2021–2050
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Venkatesh V, Davis FD (2000) A theoretical extension of the technology acceptance model: four
longitudinal ﬁeld studies. Manag Sci 46(2):186–204
Yang T, Shah S, Chang C (2020, 23 March). The future of the pharmaceutical industry | Deloitte
Insights. Deliotte. https://www2.deloitte.com/us/en/insights/industry/health-care/future-of-phar
maceutical-industry.html Accessed 26 Feb 2021
Entrepreneurial Culture, Management,
and Innovation of Dairy Industries
in Greece, in a Bureaucracy Environment

Athanasios Falaras and Odysseas Moschidis

1 Introduction

Greece prior to the global economic crisis of 2008 was a country where the state
sector had a major role in the economy. After a decade of economic adjustment, the
state sector involvement was decreased to provide space to private entrepreneurship;
however, the Greek economy has not yet achieved to reach at the pre-crisis economic
levels. Income has been heavily decreased, and the majority of businesses are in
limbo. This fact creates the need to study entrepreneurial culture and innovation and
factors among them in the case of Greece. Entrepreneurial culture is the basis for
entrepreneurial success (Danish et al. 2019). Also entrepreneurial culture is critical
so an entrepreneur applies successful management to a business (Capelleras et al.
2018). Entrepreneurial culture and management are critical to innovation, which is
an important factor of entrepreneurial success (Kline 2009). Entrepreneurial culture
and management are two factors that belong to the internal business environment.
However the external business environment interact with the internal to innovation
process (Cao and Chen 2019). Based on the quadruple helix model, the government
is one of the external business environment dimensions (along with academia and
society), and it is responsible for various laws that might affect positively or
negatively businesses (Schütz et al. 2019). A serious issue regarding the external
business environment in Greek entrepreneurial status quo is bureaucracy. There are a
series of complex and time-consuming legal processes between businesses and
Greek public services that deteriorate the Greek external business environment
(Vlachos et al. 2019). In addition entrepreneurial culture in Greece is low, and
people have negative opinion about entrepreneurship, as actually they mostly seek

A. Falaras (✉) · O. Moschidis

Department of Business Administration, University of Macedonia, Thessaloniki, Greece
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 193

the best possible employment (Thomakis and Daskalopoulou 2021). Therefore low
entrepreneurial culture leads to low management (Jardim et al. 2021). Actually
management in Greek small and medium businesses (SMEs) is based exclusively
on practical experience (Vlados 2021).
Thus, the following question is created: Can entrepreneurship in Greece suc-
ceed? To answer the question to this problem, a model was created. So, this chapter
studies the effect of entrepreneurial culture, business management, and bureaucracy
on innovation, with business management and bureaucracy being set as intermediate
variables. Of particular interest is the indirect effect of entrepreneurial culture on
innovation, through bureaucracy and business management. To study this model a
successful sector of the Greek economy was chosen to apply the model. Dairy
industry has the greatest significance in the food industry, so that makes it worth
studying. It constitutes 19% of the total value of the food industry in Greece
(Hellenic Statistical Authority 2021). This makes the dairy industry a good enough
sector to apply the model and regard it as an example to enhance the other sectors of
the economy.
The study produces useful output for entrepreneurs, scholars, policymakers, and
society. The significance of the study for entrepreneurs lies in the fact that entrepre-
neurship can succeed in Greece, although the environment could be better. However,
for this to happen, entrepreneurs should cultivate their entrepreneurial culture and
optimize business management, so that they can be led to innovation. In addition,
this model can become the basis for further research in academia. Its application on
another field of the economy of Greece or another country would be of interest, so
that the necessary comparisons of results will be made. As far as the state is
concerned, the results of the survey about bureaucracy show that additional actions
need to be taken for the improvement in the external business environment in
Greece. Regarding the society the results indicate that entrepreneurship is the path
to economic prosperity. But to succeed it clearly must be done right.

2 Background
2.1 Innovation

“Innovation is the implementation of a new or signiﬁcantly improved product (good

or service), or process, a new marketing method, or a new organizational method in
business practices, workplace organization or external relations” (OECD 2005,
p. 46). It is analyzed in four dimensions: product innovation, process innovation,
organizational innovation, and marketing innovation. Product and process innova-
tion form technological innovation, while organizational and marketing innovation
form non-technological innovation (OECD 2005). It is regarded as a critical factor
for economic welfare (Frankovic et al. 2020).
There has been developed vast literature analyzing the stages that exist from the
conception of an idea to actual innovation. All these stages are conceptualized by the
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 195

term innovation management (Adams et al. 2006). According to Roberts (1988),

there are ﬁve stages on process innovation: The creation of new ideas, redesign of
production processes, knowledge management, development of innovation, and
redesign of market processes. During innovation management, a series on inventions
may not lead to successful innovations. However, a business that learns from failure
can be led to success (Ramirez-Portilla and Torres 2021). At next follow indicative
examples of innovations in dairy industry.
In Tunisia dairy industries proceed to organizational change by open innovation,
where the knowledge of sharing platforms takes place, achieving internal absorptive
capacity, minimizing complexity, uncertainty, and risks and reaching performance
results (Ben Arﬁ et al. 2019). Likewise, cooperation networks lead to innovation in
Brazil (Torre et al. 2019). In Italy process innovations take place in logistics to
improve the environmental imprint of dairy companies (Cannas et al. 2020). In
addition, the recent vegan trend has led to the development of new products in the
dairy industry, which contain dairy substitutes (Saari et al. 2021). Convergent
innovation also takes place in the dairy industry, improving the dairy supply chain
from plant production to retailers (Talukder et al. 2021).

2.2 Bureaucracy

A serious issue highlighted by the majority of entrepreneurs in Greece is bureau-

cracy. This term regards the amount of processes and time required that businesses in
Greece have to spend, to fulfill a series of obligations toward the state. Such
examples are time to obtain electrical connection, time to start a business, and time
required to obtain an operating license or time to register property. Currently
bureaucracy halts entrepreneurial intention, foreign direct investment, innovation,
extroversion, and therefore economic growth (Papaconstantinou et al. 2013;
Vasilopoulos and Tsitsakis 2020; Vlachos et al. 2019). To provide an overview of
bureaucracy in Greece, a relative evolution of indicative measures is presented in the
Table 1 on data provided by the World Bank (2019). In 2009, the last pre-crisis
elections took place. In 2010, the first bailout loan took place. In 2012, the first
elections during crisis were conducted. In 2015, a referendum was conducted
regarding the bailout loans. Greece officially exited from the bailout programs in
2018, and in 2019, the first post-crisis elections took place.
Based on the demonstrated literature and data, the following hypothesis is
created:
H1: Bureaucracy affects negatively and statistically significantly innovation.
Table 1 indicates that bureaucracy has been reduced during the last decade.
Especially during the quarantine period, a digital transformation took place in the
public sector (Karamalis and Vasilopoulos 2020). However, bureaucracy is still
highlighted as an important issue in Greek entrepreneurship.
196 A. Falaras and O. Moschidis

Table 1 Indicative bureaucracy issues in Greece

Series name 2010 2015 2019
Time required to start a businessa 19 13 4
Time required to register propertyb 26 26 26
Time required to get electricityc 51 51 51
Source: World Bank (2019)
a
Time required to start a business is the number of calendar days needed to complete the procedures
to legally operate a business. If a procedure can be speeded up at additional cost, the fastest
procedure, independent of cost, is chosen.
b
Time required to register property is the number of calendar days needed for businesses to secure
rights to property.
c
Time required to get electricity is the number of days to obtain a permanent electricity connection.
The measure captures the median duration that the electricity utility and experts indicate is
necessary in practice, rather than required by law, to complete a procedure.

2.3 Business Management

Fayol (1930) was the ﬁrst to deﬁne management. He referred to management as a

process of planning, organizing, and mobilizing human workforce and control it to
achieve a goal. Nowadays effective management could be analyzed in five dimen-
sions (Peters 1988):
• Immediate customer response
• Constant pursuit of innovation to all business aspects
• Teamwork mindset
• Leadership that systematically pursuits change, inspires, and shares a vision
• Manpower fit allocation
So a good manager has to create a suitable physical environment for customers, as
it is important for service experience and behavioral intentions (Chen and Shen
2020). In addition a good manager constantly tries to figure out a new innovation that
could improve business efficiency (Ghasemaghaei and Calic 2020). Furthermore
teamwork is instilled in every team type (action, production, management)
(Kerrissey et al. 2020). Moreover a manager is a leader who comprises the values
of vision, strategy, empowerment, motivation, and inspiration (Shahzad et al. 2021).
Finally there are a series of tools to optimize manpower allocation (Chan et al. 2019).
Given these conditions business management leads to innovation (Lam et al. 2021).
These dimensions are included in the eight steps of business management process:
planning, organization chart and job design, organization, division of labor, leader-
ship and motivation, human resources development, administrative assessment, and
final assessment (Vlados and Chatzinikolaou 2022).
During the planning stage, goals are set. Then the means of fulfilling the goals are
defined, as long as the deadline to achieve the initial targets (Mio et al. 2020). The
organization chart defines the business structure and fabric, as the hierarchical
relation between various divisions, berths, and manpower. Rules are also set,
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 197

business needs are defined and the way of meeting them (Lassl 2020). Also at this
stage, job positions are designed (Peiró et al. 2020). The next step is to utilize the
combination of material and nonmaterial business resources (Maraveas et al. 2022).
Then division of labor takes place (Stroeva et al. 2021). After these steps business
administration takes place. At this stage leadership traits and workforce motivation
are required to maximize employee’s productivity (Gerdenitsch et al. 2020). Then
efficiency is measured and assessed, feedback is provided, and if it is needed,
corrective actions take place (Cabinova et al. 2021). Finally, a summary assessment
takes place. This stage takes into consideration that the position of a business in the
market is relative to its competitors and evolutionary in time. So the final assessment
takes place where it is judged if the business is in the right direction (Bagdadli and
Gianecchini 2019).
Currently, management practices had to maneuver during the pandemic situation
and the subsequent lockdowns across the globe, as employees’ mental health was
challenged. It was studied that consideration; establishing work structure; allowing
flexibility, open, precise and regular communication; and anticipatory, goal-oriented
and fast problem were highlighted as the most useful measures for employees’
mental health (Bouchard and Meunier 2022; Thielsch et al. 2021). Based on the
demonstrated literature, the following hypothesis is created:
H2: Business management affects positively and statistically significantly
innovation.

2.4 Entrepreneurial Culture

The term “entrepreneurial culture” expresses the entrepreneur’s stance on a series of

issues. Such issues are willingness to take risks and initiatives, openness to change,
continuously pursuing new challenges and access to new resources, having leader-
ship traits, and considering innovation as a systematic and team pursuit (Dimitratos
and Plakoyiannaki 2003; Hayton and Cacciotti 2013; Suddle et al. 2010). It is
important because unlike various studied demographic measures (age, budget,
etc.) that are objective, entrepreneurial culture is subjective to every entrepreneur,
as none can choose age and experience, but it’s a matter of choice and personal self-
improvement to build a good entrepreneurial culture (Minguzzi and Passaro 2001).
In the case of Greece, most Greek SMEs are owned by individual entrepreneurs,
who are often supported by family members (Karanasios and Lans 2018). Therefore,
entrepreneurial culture becomes even more important since the entrepreneur is
primarily responsible to convert his/her own entrepreneurial culture into organiza-
tional culture, passing on his/her own values to the employees (Schmiedel et al.
2019). Consequently, a business owner’s entrepreneurial culture turns into organi-
zational culture, leading the business to innovation (Buccieri et al. 2020). Since the
majority of entrepreneurs in Greece have low entrepreneurial culture, this is exactly
198 A. Falaras and O. Moschidis

the reason why a strong sector of the Greek economy (dairy industry) was chosen to
study entrepreneurial culture: because it could demonstrate that innovation and,
therefore, business success require that the entrepreneur have developed the appro-
priate entrepreneurial culture. So, the following hypothesis is created:
H3: Entrepreneurial culture affects positively and statistically significantly
innovation.
Moreover, entrepreneurial culture constitutes the basis for dealing with the
relations of the company with its clients and other cooperating companies, the
right allocation of human resources according to each employee’s skills, the adop-
tion of constant search for innovation within the organization, and the rise of
company leaders who cultivate team spirit among the employees. It is already
studied that there is a positive relationship between entrepreneurial culture and
management (Basu and Bhola 2022; Jardim et al. 2021). As entrepreneurial culture
is regarded the basis for the development of management and it is already studied
that there is a positive relationship between management and innovation, it is
expected that management complementary mediates between entrepreneurial culture
and innovation. So, the following hypotheses are created:
H4: Entrepreneurial culture affects positively and statistically significantly business
management.
H5: Business management complementary mediates the relationship between entre-
preneurial culture and innovation.
Bureaucracy is a concept negatively related with entrepreneurship (Griffiths et al.
2009). It is widely regarded as a negative factor on entrepreneurial intention
(Sørensen 2007). In addition as entrepreneurial culture is a concept closely associ-
ated with entrepreneurship, it is studied that there is a negative relationship between
entrepreneurial culture and bureaucracy (Farinha et al. 2020). However it is inter-
esting to see how bureaucracy mediates the relation between entrepreneurial culture
and innovation. In specific to see if entrepreneurial culture blunts bureaucracy,
which is regarded as an obstacle of external business environment and if entrepre-
neurial culture can still positively affect innovation through bureaucracy. This is the
most interesting part of this research because if the indirect effect of entrepreneurial
culture on innovation through bureaucracy is statistically significant, it states that
external business environment obstacles can be surpassed (Sitaridis and Kitsios
2019). So, the last hypotheses of this study are formed:
H6: Entrepreneurial culture affects negatively and statistically significantly
bureaucracy.
H7: Bureaucracy competitively mediates the relationship between entrepreneurial
culture and innovation.
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 199

3 Methodology

3.1 Data Collection and Model Design

To carry out the analysis, a research was conducted with the use of an electronic
questionnaire designed on Google forms. The survey population were owners of
dairy product businesses in mainland Greece and Crete. Dairy industry has the
greatest signiﬁcance in the food industry, so that makes it worth studying. It
constitutes 19% of the total value of the food industry in Greece. This makes the
dairy industry a good enough sector of the economy to study its entrepreneurship. In
the questionnaire, the aim of the survey was stated, as well as the assurance of the
participants’ anonymity. At the moment there is no open database in Greece
regarding this population. For this reason, the method of quota sampling was chosen.
For the assurance of representativeness of the population, there were aims of
collecting responses by region, based on a census by the Hellenic Statistical Author-
ity. Details of contact with the businesses were found on commercial websites.
Phone calls took place, during which the owner of the company was asked to answer
the questionnaire. Since assent was provided, the questionnaire was sent via e-mail.
The data regarded questions about innovation, entrepreneurial culture, bureaucracy,
and management, and they constituted the latent variables of the survey. The survey
started in last January and was completed in March. 249 responses were collected
altogether, having no missing values at all. Table 2 conﬁrms population
representativeness.
To process the data, the use of latent variables and study of indirect effects, the
software SmartPLS which utilizes the partial least square (PLS) algorithm was used.
This software was chosen due to its simplicity in handling these issues. To secure
that the sample is adequate enough to apply the PLS algorithm, the study followed
Hair et al. (2011), where the sample size should be greater than 10 times the

Table 2 Population representativeness

Relative sample Population Relative population
Region Sample frequency 2019 frequency
Eastern Macedonia and 22 0.09 37 0.05
Thrace
Central Macedonia 48 0.19 127 0.16
Western Macedonia 9 0.04 39 0.05
Epirus 9 0.04 38 0.05
Thessaly 45 0.18 82 0.11
Central Greece 21 0.08 78 0.10
Western Greece 12 0.05 74 0.10
Peloponnese 35 0.14 82 0.11
Attica 21 0.08 111 0.14
Crete 27 0.11 102 0.13
Summary 249 1 770 1
200 A. Falaras and O. Moschidis

maximum number of inner or outer model links pointing at any latent variable in the
model (Kock and Hadaya 2018). So, this research would need a minimum of
210 statistical units. As the sample included 249 statistical units, it is secured that
the sample is adequate enough to apply the method. The variables of the model are
entrepreneurial culture, business management, bureaucracy, and innovation. As
these are latent variables, so, for each of them, there were speciﬁc evaluation scale
questions (Table 3). Entrepreneurial culture construct was designed based on the
work of Leal-Rodríguez et al. (2017) and was initially represented by six indicators.
Bureaucracy construct was designed based on the work of Nyarku and Oduro (2018)
and was initially represented by the time required to start a business, the time
required to register property, the time required to get electricity, and the time
required to build a warehouse. Business management was initially represented by
immediate customer response; constant pursuit of innovation to all business wrinkle;
teamwork mindset; leadership that systematically pursuits change, inspires, and
shares a vision; and manpower-ﬁt allocation (Peters 1988). Innovation is presented

Table 3 Latent constructs and their indicators

Entrepreneurial culture (EC)
EC1 My organization is characterized by fostering entrepreneurship, innovation, and
assuming risks
EC2 My organization is typiﬁed by promoting individual initiative, risk-taking, innovation,
and uniqueness
EC3 My organization values the acquisition of new resources and the creation of new
challenges as well as appreciating originality and the search for opportunities
EC4 My organization aspires to become leaders in developing unique and innovative
products or services
Bureaucracy (BUR)
BUR1 How limiting do you consider the time required to initiate a branch of your business for
your innovative activities?
BUR2 How limiting do you consider the time required to register corporate property for your
innovative activities?
BUR3 How limiting do you consider the time required to get electricity for a branch of your
business for your innovative activities?
Management (MAN)
MAN1 To what degree do you evaluate that you have a team mindset?
MAN2 To what degree do you evaluate that you inspire and share a vision with your
employees?
MAN3 To what degree would you evaluate yourself as a leader who systematically pursuits
change?
Innovation (INN)
INN1 How innovative was your business in the last three years in terms of technological
innovation?
INN2 How innovative was your business in the last three years in terms of organizational
innovation?
INN3 How innovative was your business in the last three years in terms of marketing
innovation?
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 201

by technological innovation (product and process innovation), organizational, and

marketing innovation (OECD 2005). The use of latent variables, as well as the need
to study the indirect effects in the model, led to the use of partial least squares (PLS)
algorithm and SmartPLS software to process the data (Ringle et al. 2015). However,
by extracting the outer loadings of each indicator after the appliance of the PLS
algorithm, some of them had a value higher than 0.4 but lower than the
recommended 0.7 (Hair et al. 2019; Hair et al. 2022). Due to the fact that the removal
of these indicators increased the other loading values, these indicators were excluded
from the model. Then, the model was redesigned.

3.2 Model Estimation

Firstly, the PLS-algorithm was selecting the “path” weighting scheme for 5000
iterations and a stop criterion of 10-7, while choosing equal indicator weights. In
this way, the path coefﬁcients and the R2 values are extracted. The path coefﬁcients
demonstrate that management has the strongest effect on innovation (0.222)
followed by entrepreneurial culture (0.216) and bureaucracy (-0.191). These three
variables explain 16.9% of the variance of the variable innovation. The extracted R2
values indicate low but acceptable predictive power (Cohen 1988). These results are
demonstrated in the Fig. 1.

4 Results

4.1 Measurement Model Assessment

Next, the measurement model of the latent construct has to be assessed. The model is
assessed for convergent validity and internal consistency reliability. To secure
convergent validity, the outer loading values, and average variance extracted
(AVE) are extracted. Each outer loading value has to be higher than 0.7, and each
AVE value must be higher than 0.5. Then, the model is checked for internal
consistency reliability. So, composite reliability, ρA, and Cronbach’s alpha values
are extracted. Composite reliability and ρA values should be higher than 0.7, while
Cronbach’s alpha values should be between 0.7 and 0.9. The last step taken to finish
the model assessment is the discriminant validity check by using the heterotrait-
monotrait (HTMT) criterion, where each table cell value should be lower than 0.85
(Sarstedt et al. 2017). All the above conditions were met, and convergent validity
and internal consistency reliability were secured (Table 4).
To ensure discriminant validity, the bootstrapping procedure is executed, choos-
ing bias-corrected and accelerated (BCa) bootstrap option for 5000 iterations, at 5%
significance level, and two-tailed test. The results demonstrate that none of the
HTMT confidence intervals include the value 1 (Table 5). So, it is regarded that
202 A. Falaras and O. Moschidis

BUR1 BUR2 BUR3

0.812 0.990 0.773

0.026

BUR –0.191
–0.160
EC1
INN1
0.934
EC2 0.849
0.975 0.169 0.839 INN2
0.216
0.820
EC3 0.853
0.744 INN3
EC INN
EC4

0.198 0.222

0.039

MAN
0.795 0.872 0.736

MAN1 MAN2 MAN3

Fig. 1 Model including path coefﬁcients, R2, and outer loadings

Table 4 Assessment of measurement model

Convergent validity Internal consistency reliability
Latent Loadings AVE Composite reliability ρΑ Cronbach’s alpha
variable Indicators > 0.7 > 0.5 > 0.7 > 0.7 0.7–0.9
EC EC1 0.934 0.763 0.927 0.902 0.892
EC2 0.975
EC3 0.820
EC4 0.744
BUR BUR1 0.812 0.746 0.897 0.865 0.823
BUR2 0.990
BUR3 0.773
MAN MAN1 0.795 0.644 0.844 0.814 0.738
MAN2 0.872
MAN3 0.736
INN INN1 0.849 0.717 0.884 0.808 0.804
INN2 0.839
INN3 0.853
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 203

Table 5 HTMT
BUR EC INN MAN
BUR
EC 0.184
INN 0.287 0.342
MAN 0.080 0.228 0.300

all HTMT values signiﬁcantly differ from 1, completing the discriminant validity
check.

4.2 Structural Model Assessment

Next, the model is checked for collinearity. So, the variance inflation factor (VIF) is
checked. An ideal VIF value is below 3.0. This was the case for all of the extracted
VIFs, so collinearity was not an issue. Then, the bootstrapping procedure is applied
to check if the effects on the model are statistically significant. All of the studied
direct and indirect effects were statistically significant. Also, f 2 is checked on the
PLS algorithm results to extract the effect size among the studied variables. All the
effects are higher than 0.02 but lower than 0.15 which means that the effects are
small.
Then, the model is checked to assess its predictive accuracy. So, PLS predict and
blindfolding are executed (Hair et al. 2019; Hair et al. 2022). On PLS predict the root
mean squared error (RMSE) and mean absolute error (MAE) are checked for each
indicator on the PLS and on the linear model (Table 6). The minority of PLS-SEM
indicators extract higher prediction errors than the linear model. This indicates a
medium predictive relevance of endogenous constructs (Shmueli et al. 2019). Then
blindfolding is executed. If Q2 values are higher than 0, this means that the model
has predictive relevance. Specifically, BUR, MAN, and INN Q2 values are 0.016,
0.019, and 0.109, respectively. However, they are lower than 0.25, which means
than the model’s predictive accuracy is small.
In summary, the results demonstrate that the model has low predictive power and
effect size among the studied variables, but the effects are statistically significant.
Finally, while the direct effect of entrepreneurial culture to innovation is statistically
significant, the same goes for the indirect effects through bureaucracy and manage-
ment. This makes bureaucracy competitive partial mediator and management com-
plementary partial mediator (Nitzl et al. 2016).
The research was conducted with the aim to study the model presented in the case
of Greece. The studied effects may be of low impact, but they are still statistically
significant (Table 7). The results confirmed the significance of entrepreneurial
culture for innovation, in both its direct and its indirect effect. Apart from its
contribution to innovation, entrepreneurial culture is fundamental for the encour-
agement of business management in Greece and therefore the importance of business
204 A. Falaras and O. Moschidis

Table 6 Comparison of RMSE and MAE on PLS and linear model

PLS model Linear model
RMSE MAE RMSE MAE
BUR3 1.286 0.959 1.304 0.970
BUR1 1.269 1.041 1.284 1.053
BUR2 0.994 0.775 1.008 0.786
INN2 1.098 0.947 1.106 0.960
INN3 1.259 1.080 1.265 1.086
INN1 1.287 1.090 1.292 1.090
MAN1 1.001 0.857 1.011 0.861
MAN2 0.992 0.842 0.989 0.843
MAN3 0.942 0.711 0.940 0.731

Table 7 Path coefficients of the structural model and significance testing results
Path 95% Bca confidence Significant f 2 effect
coefficient interval ( p < 0.05)? size
EC → INN 0.216 [0.090, 0.327] Yes 0.053
EC → BUR -0.160 [-0.289, -0.016] Yes 0.026
EC → MAN 0.198 [0.050, 0.313] Yes 0.041
BUR → INN -0.191 [-0.303, -0.063] Yes 0.043
MAN → INN 0.222 [0.096, 0.335] Yes 0.057
EC → BUR → INN 0.035 Ν/Α Yes N/A
EC → MAN → INN 0.049 Ν/Α Yes N/A

management to innovation. Furthermore, it is apparent that the negative effect of

entrepreneurial culture on bureaucracy reduces the negative effect of management
on innovation. The situation of bureaucracy, despite having been slightly improved
in recent years, still acts as a brake on the effort of businesses to innovate. Further-
more, it is apparent that the negative effect of entrepreneurial culture on bureaucracy
reduces the negative effect of bureaucracy on innovation. An additional finding is
the fact that, despite the negative effect of bureaucracy on innovation, the indirect
effect of entrepreneurial culture on innovation remains statistically significant. This
result is the most important finding of this study, as it comes as an answer to the
general public opinion that entrepreneurship in Greece cannot succeed. Thus man-
agement is a complementary partial mediator, while bureaucracy is a competitive
partial mediator (Nitzl et al. 2016).

5 Conclusions and Implications for Theory and Practice

Entrepreneurship in Greece can succeed, but an important requirement is the culti-

vation of the right entrepreneurial culture and therefore effective management. It is
up to entrepreneurs to deal with external business environment obstacles by
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 205

improving their entrepreneurial culture and business management and surpass them.
It is up to society to change opinion about entrepreneurship and think that real wealth
comes from possession of businesses and not employment. Greece needs to be a
country where its population will shift from seeking employment to create business
opportunities. The chosen sector of dairy industry in this study and the successful
appliance of the proposed model highlight this statement.
These results create useful output for scholars and policymakers. This model can
become the basis for further research in academia. Its application on another field of
the economy of Greece or another country would be of interest, so that the necessary
comparisons of results will be made. Also, the model could be increased in com-
plexity by adding more variables that entrepreneurial culture utilize. The same goes
for the barriers of the external business environment. Actually, they operate as a
business filter, where only the fittest survive in every unique business environment.
As far as the state is concerned, the results of the survey about bureaucracy show that
additional actions need to be taken for the improvement of the external business
environment in Greece. Greece’s rank in the Ease of Doing Business Index is 79th, a
fact that reflects that there still margin for improvement (World Bank 2019). The
state has to secure the best possible conditions for entrepreneurship and operate as an
assistant to entrepreneurship in Greece.

References

Adams R, Bessant J, Phelps R (2006) Innovation management measurement: a review. Int J Manag
Rev 8(1):21–47
Bagdadli S, Gianecchini M (2019) Organizational career management practices and objective
career success: a systematic review and framework. Hum Resour Manag Rev 29(3):353–370
Basu R, Bhola P (2022) Modelling interrelationships of quality management, information technol-
ogy and entrepreneurial culture and their impact on performance from Indian IT enabled service
SMEs. BIJ 29(7):2177–2201
Ben Arfi W, Enström R, Sahut JM, Hikkerova L (2019) The significance of knowledge sharing
platforms for open innovation success: a tale of two companies in the dairy industry. J Organ
Chang Manag 32(5):496–516
Bouchard L, Meunier S (2022) Promoting remote workers’ psychological health: effective man-
agement practices during the COVID-19 crisis. Can J Adm Sci 40:1–12
Buccieri D, Javalgi RG, Cavusgil E (2020) International new venture performance: role of inter-
national entrepreneurial culture, ambidextrous innovation, and dynamic marketing capabilities.
Int Bus Rev 29(2):101639
Cabinova V, Fedorcikova R, Jurova N (2021) Innovative methods of evaluating business efficiency:
a comparative study. Mark Manag Innov 6718(2):133–145
Cannas VG, Ciccullo F, Pero M, Cigolini R (2020) Sustainable innovation in the dairy supply
chain: enabling factors for intermodal transportation. Int J Prod Res 58(24):7314–7333
Cao H, Chen Z (2019) The driving effect of internal and external environment on green innovation
strategy-the moderating role of top management’s environmental awareness. Nankai Bus Rev
Int 10(3):342–361
Capelleras JL, Martin-Sanchez V, Rialp J, Shleha W (2018) Entrepreneurs’ export orientation and
growth aspirations: the moderating role of individual human capital. Studies on entrepreneur-
ship. Structural change and industrial dynamics. In: Bosio G, Minola T, Origo F, Tomelleri S
(eds) Rethinking entrepreneurial human capital. Springer, Cham, pp 63–87
206 A. Falaras and O. Moschidis

Chan KA, Pan MJJ, Chua BT, Hu XM, Low MYH (2019) Resources optimisation in new hospital
central kitchen design–a discrete event simulation approach. In: Tan G, Lehmann A, Teo Y, Cai
W (eds) Methods and applications for modeling and simulation of complex systems. Springer,
Singapore, pp 251–259
Chen M, Shen C (2020) The correlation analysis between the service quality of intelligent library
and the behavioral intention of users. Electron Libr 38(1):95–112
Cohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum
Associates, New York
Danish RQ, Asghar J, Ahmad Z, Ali HF (2019) Factors affecting entrepreneurial culture. J Innov
Entrep 8:1
Dimitratos P, Plakoyiannaki E (2003) Theoretical foundations of an international entrepreneurial
culture. J Int Entrep 1(2):187–215
Farinha L, Lopes J, Bagchi-Sen S, Sebastião JR, Oliveira J (2020) Entrepreneurial dynamics and
government policies to boost entrepreneurship performance. Socio Econ Plan Sci 72(1):100950
Fayoul H (1930) Industrial and general administration. Sir Issac Pitman & Sons, London
Frankovic I, Kuhn M, Wrzaczek S (2020) Medical innovation and its diffusion: implications for
economic performance and welfare. J Macroecon 66(4):1
Gerdenitsch C, Sellitsch D, Besser M, Burger S, Stegmann C, Tscheligi M et al (2020) Work
gamification: effects on enjoyment, productivity and the role of leadership. Electron Commer
Res Appl 43(5):1
Ghasemaghaei M, Calic G (2020) Assessing the impact of big data on firm innovation performance:
big data is not always better data. J Bus Res 108(1):147–162
Griffiths MD, Kickul J, Carsrud AL (2009) Government bureaucracy, transactional impediments,
and entrepreneurial intentions. Int Small Bus J 27(5):626–645
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult TM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. SAGE, Los Angeles
Hayton JC, Cacciotti G (2013) Is there an entrepreneurial culture? A review of empirical research.
Entrep Reg Dev 25(9–10):708–731
Hellenic Statistical Authority (2021) Production and sales of manufactured products. https://www.
statistics.gr/en/statistics/-/publication/SIN06/2019. Accessed 14 Jun 2022
Jardim J, Bártolo A, Pinho A (2021) Towards a global entrepreneurial culture: a systematic review
of the effectiveness of entrepreneurship education programs. Educ Sci 11(8):1–18
Karamalis P, Vasilopoulos A (2020) The digital transformation in public sector as a response to
COVID-19 pandemic: The case of Greece. In: Conference: XIV Balkan Conference on Opera-
tional Research (BALCOR 2020), 30 September–3 October, Thessaloniki, Central Macedonia,
Greece
Karanasios K, Lans T (2018) Learning to become an entrepreneur in unfavorable conditions: the
case of new entrants in the context of the Greek debt-crisis. In: Loo S (ed) Multiple dimensions
of teaching and learning for occupational practice. Routledge, London, pp 131–149
Kerrissey MJ, Satterstrom P, Edmondson AC (2020) Into the fray: adaptive approaches to studying
novel teamwork forms. Organ Phycol Rev 10(2):68–86
Kline SJ (2009) An overview of innovation. In: Rosenberg N (ed) Studies on science and the
innovation process. World Scientific, Singapore, pp 173–203
Kock N, Hadaya P (2018) Minimum sample size estimation in PLS-SEM: the inverse square root
and gamma-exponential methods. Inf Syst J 28(1):227–261
Lam L, Nguyen P, Le N, Tran K (2021) The relation among organizational culture, knowledge
management, and innovation capability: its implication for open innovation. J Open Innov
Technol Market Complex 7(1):66
Entrepreneurial Culture, Management, and Innovation of Dairy Industries. . . 207

Lassl W (2020) Toward a functioning organizational (chart) structure. In: Lassl W (ed) The viability
of organizations, vol 3. Springer, Cham, pp 217–241
Leal-Rodríguez AL, Albort-Morant G, Martelo-Landroguez S (2017) Links between entrepreneur-
ial culture, innovation, and performance: the moderating role of family firms. Int Entrep Manag
J 13(3):819–835
Maraveas C, Piromalis D, Arvanitis KG, Bartzanas T, Loukatos D (2022) Applications of IoT for
optimized greenhouse environment and resources management. Comput Electron Agric 198
(July):1
Minguzzi A, Passaro R (2001) The network of relationships between the economic environment
and the entrepreneurial culture in small firms. J Bus Ventur 16(2):181–207
Mio C, Panfilo S, Blundo B (2020) Sustainable development goals and the strategic role of
business: a systematic literature review. Bus Strateg Environ 29(8):3220–3245
Nitzl C, Roldan JL, Cepeda G (2016) Mediation analysis in partial least squares path modelling,
helping researchers discuss more sophisticated models. Ind Manag Data Syst 116(9):1849–1864
Nyarku KM, Oduro S (2018) Effect of legal and regulatory framework on SMEs growth in the
Accra Metropolis of Ghana. Int J Entrepreneurship Innov 19(3):207–217
OECD (2005) Oslo Manual. In: OECD and Eurostat Publication, 3rd ed. https://www.oecd-
ilibrary.org/science-and-technology/the-measurement-of-scientific-and-technological-activi
ties_19900414. Accessed 7 Sep 2018
Papaconstantinou P, Tsagkanos AG, Siriopoulos C (2013) How bureaucracy and corruption affect
economic growth and convergence in the European Union: the case of Greece. Manag Financ
39(9):837–847
Peiró JM, Bayona JA, Caballer A, Di Fabio A (2020) Importance of work characteristics affects job
performance: the mediating role of individual dispositions on the work design-performance
relationships. Personal Individ Differ 157(April):109808
Peters T (1988) Thriving on chaos: handbook for a management revolution, 1st edn. Harper
Perennial, New York
Ramirez-Portilla A, Torres EG (2021) Recognizing the value of unsuccessful innovations: a case
study from the dairy industry in Mexico. In: McMurray A, Muenjohn N, Weerakoon C (eds)
The Palgrave handbook of workplace innovation. Palgrave Macmillan, Cham, pp 407–426
Ringle CM, Wende S, Becker J (2015) SmartPLS 3 [computer software]. SmartPLS GmbH. www.
smartpls.com. Accessed 15 May 2022
Roberts EB (1988) What we’ve learned: managing invention and innovation. Res Technol Manag
31(1):11–29
Saari UA, Herstatt C, Tiwari R, Dedehayir O, Mäkinen SJ (2021) The vegan trend and the
microfoundations of institutional change: a commentary on food producers’ sustainable inno-
vation journeys in Europe. Trends Food Sci Technol 107(7):161–167
Sarstedt M, Ringle CM, Hair JF (2017) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg A (eds) Handbook of market research. Springer, Cham, pp
1–47
Schmiedel T, Müller O, vom Brocke J (2019) Topic modeling as a strategy of inquiry in organi-
zational research: a tutorial with an application example on organizational culture. Organ Res
Methods 22(4):941–968
Schütz F, Heidingsfelder ML, Schraudner M (2019) Co-shaping the future in quadruple helix
innovation systems: uncovering public preferences toward participatory research and innova-
tion. She Ji 5(2):128–146
Shahzad K, Raja U, Hashmi SD (2021) Impact of big five personality traits on authentic leadership.
Leadersh Org Dev J 42(2):208–218
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Sitaridis I, Kitsios F (2019) Entrepreneurship as a career option for information technology
students: critical barriers and the role of motivation. J Knowl Econ 10(3):1133–1167
208 A. Falaras and O. Moschidis

Sørensen JB (2007) Bureaucracy and entrepreneurship: workplace effects on entrepreneurial entry.

Adm Sci Q 52(3):387–412
Stroeva OA, Lavrikova NI, Alekhina TA, Semenova EE, Stepanova MA (2021) Development of
organizational diagrams and governance mechanisms of region’s economy. In: Popkova EG,
Sergi BS (eds) Modern global economic system: evolutional development vs. revolutionary
leap. Springer, Cham, pp 1807–1816
Suddle K, Beugelsdijk S, Wennekers S (2010) Entrepreneurial culture and its effect on the rate of
nascent entrepreneurship. In: Freytag A, Thurik R (eds) Entrepreneurship and culture. Springer,
Heidelberg, pp 227–244
Talukder B, Agnusdei GP, Hipel KW, Dubé L (2021) Multi-indicator supply chain management
framework for food convergent innovation in the dairy business. Sustain Futures 3:100045
Thielsch MT, Röseler S, Kirsch J, Lamers C, Hertel G (2021) Managing pandemics—demands,
resources, and effective behaviors within crisis management teams. Appl Psychol 70(1):
150–187
Thomakis Z, Daskalopoulou I (2021) Entrepreneurial views and rural entrepreneurial potential:
evidence from Greece. J Knowl Econ 13:1611–1634
Torre A, Polge E, Wallet F (2019) Proximities and the role of relational networks in innovation: the
case of the dairy industry in two villages of the “green municipality” of Paragminas in the
eastern Amazon. Reg Sci Policy Pract 11(2):279–294
Vasilopoulos A, Tsitsakis C (2020) Entrepreneurship in Greece. KnE Soc Sci 4(1):71–78
Vlachos V, Mitrakos P, Tsimpida C, Tsitouras A, Bitzenis A (2019) An exploratory survey of the
factors discouraging FDI in Greece. Glob Bus Econ Anthol II(December):184–190
Vlados C (2021) The Stra.Tech.Man scorecard. Int J Bus Adm 12(2):36–63
Vlados C, Chatzinikolaou D (2022) Management conception challenges in the post-COVID-19 era
and the Stra.Tech.Man analysis. In: Machado C, Davim JP (eds) Organizational Management in
Post Pandemic Crisis. Springer, Cham, pp 29–51
World Bank (2019) World development indicators. The World Bank Group. https://databank.
worldbank.org/source/world-development-indicators#. Accessed 13 May 2022
The Inﬂuence of Facebook on Value
Cocreation: Evidence from the Moroccan
Fast-Food Industry

Mohammed Hassouni, Abdellatif Chakor, and Siham Mourad

1 Introduction

The twenty-ﬁrst century has been characterized by a change in economic paradigm.

The economy has switched from a good dominant logic to a service dominant logic
(Vargo and Lusch 2004, 2016). The service dominant logic considers that Goods are
no longer consumed for what they are (their resources), but for what they represent to
the consumer (Vargo and Lusch 2004, 2006, 2016). Service is viewed as “the
application of specialized competences (knowledge and skills) through deeds, pro-
cesses, and performances for the beneﬁt of another entity or the entity itself” (Vargo
and Lusch 2004). This change in orientation implied that value is always jointly
produced by customers and businesses. As a result, it is crucial for businesses to
consider client feedback while developing an offer (Vargo and Lusch 2004, 2016).
Consumers are therefore always Value cocreators.
Moreover, the development of social media has altered how businesses previ-
ously conducted their marketing. Consumers have become more knowledgeable and
demanding than ever before. Thus, companies are likely to listen to them. This
increased closeness between companies and consumers shows no signs of abating.
New ideas have emerged as a result of this change in marketing techniques,

M. Hassouni (✉)
University Ibn Tofail, Institut des Métiers du Sport, Kenitra, Morocco
e-mail: [email protected]
A. Chakor
University Mohammed V, Faculté des Sciences Juridiques Economiques et Sociales, Rabat,
Morocco
e-mail: [email protected]
S. Mourad
Groupe ISCAE, Casablanca, Morocco
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 209

including buzz, electronic word of mouth, and E-reputation. We could define social
media as “a group of internet-based applications that build on the ideological and
technological foundations of Web 2.0 and that allow the creation and exchange of
user generated content” (Kaplan and Haenlein 2010).
Facebook, with almost 3 billion users, is most likely the most widely used social
media platform (We are social and Kepios 2022). Facebook can be characterized as a
social networking website (Kaplan and Haenlein 2010) where advertising can
mostly take the forms of sponsored advertising and Facebook pages (Gaber and
Wright 2014). Advertisers can readily tailor their adverts to appeal to their target
audience. Gender, location, age, and interest are used as the targeting criteria for
sponsored advertising (Gaber and Wright 2014). Facebook pages are free online
locations where brands or businesses may exchange material with their clients. The
latter pages are run by community managers, whose job it is to engage customers in
conversation and keep fans informed about the company’s goods and services.
In Morocco, Facebook is the most widely used social networking platform
(We are social and Kepios 2022). 31 million Moroccans have access to Internet
which represents 84% of the nation’s population. 23 million Moroccans use social
media (We are social and Kepios 2022). 18.9 million people have Facebook
accounts in Morocco (We are social and Kepios 2022). This amount indicates one
of the highest Facebook penetration rates in Africa and is predicted to increase over
the next several years (We are social and Kepios 2022). As a result, Moroccan
consumers are becoming more connected and engaged with brands on social media.
The Moroccan fast-food industry has shown significant growth during the last
two decades (Wahabi 2016). It represents a $1.6 billion industry (Wahabi 2016) and
doesn’t show any sign of slowdown. The most popular brands are McDonald’s and
Pizza hut with 50 points-of-sale across the country (Challenge 2022) in 30 years of
presence. The most followed brands on Facebook are Pizza Hut and Domino’s Pizza
with 32 million likes and 21 million likes (We are social and Kepios 2022). Thus, the
fast-food industry Facebook pages represent active communities where the interac-
tion between consumers and brands can lead to value cocreation.
Therefore, the purpose of this chapter is to identify the factors that influence
customers to cocreate value on the fast-food industry Facebook pages. Explaining
value cocreation on Facebook is crucial for both academics and managers working in
the fast-food industry. Value cocreation on social networks has only been the subject
of a very small number of empirical investigations (Dhaka 2015; Fan and Luo 2020;
Nájera-Sánchez et al. 2020). As a result, studying value cocreation on Facebook in
the fast-food industry would contribute to fill a gap in the value cocreation literature.
Furthermore, the purpose of this research is to give guidance to community man-
agers working in the fast-food industry.
The Influence of Facebook on Value Cocreation: Evidence from the. . . 211

2 Background

Vargo and Lusch (2004) proposed ten principles to summarize the service dominant
logic (see Table 1). The latter ideas were condensed into five central axioms after
extensive discussion with the scientific community (Vargo and Lusch 2016).
According to the first axiom, service is the application of operant resources (skills
and knowledge) to one entity’s benefit (Park and Vargo 2012; Vargo and Lusch
2008, 2016). The second axiom holds that value is always jointly created by two or
more parties, with a single beneficiary. The situations considered in this axiom are
either the one involving a firm and a beneficiary or the one involving many actors
(Vargo and Lusch 2004, 2006, 2016).
All actors are considered resource integrators under the third axiom, and the
beneficiary’s value is considered under the fourth axiom (Vargo and Lusch 2008).
The fifth postulate emphasizes the importance of institutions in coordinating value
cocreation (Vargo and Lusch 2016).
As a result, the service dominant logic has become the foundation of various
cocreation approaches. The shift in emphasis from customers receiving offers to
customers influencing offers is underway and shows no signs of abating. Social
networking platforms such as Facebook have emphasized this tendency even more
by enabling deeper conversations between brands and companies. The subsequent
intimate discussions weren’t feasible earlier. Marketers have been able to collect data
they couldn’t get before.

Table 1 The 10 premises vs. the 5 axioms

FP1 service is the fundamental basis of Axiom 1/FP1 service is the fundamental basis of
exchange. exchange.
FP2 indirect exchange masks the fundamen- Axiom 2/FP6 value is cocreated by multiple
tal basis of exchange. actors, always including the beneficiary.
FP3 goods are a distribution mechanism for Axiom 3/FP9 all social and economic actors are
service provision. resource integrators.
FP4 operant resources are the fundamental Axiom4/FP10 value is always uniquely and
source of competitive advantage. phenomenologically determined by the
beneficiary.
FP5 all economies are service economies. Axiom 5/FP11 value cocreation is coordinated
through actor-generated institutions and institu-
tional arrangements.
FP6 the customer is always a value cocreator.
FP7 the enterprise cannot deliver value, but
only offer value propositions.
FP8 A service-centered view is inherently
customer-oriented and relational.
FP9 all social and economic actors are
resource integrators.
FP10 value is always uniquely and phe-
nomenologically determined by the
beneficiary.
212 M. Hassouni et al.

Prahalad and Ramaswamy (2000) observed a change in consumer role from

passive to active. Customers that interact with the business actively cocreate value
(Galvagno and Dalli 2014; Prahalad and Ramaswamy 2000). Previous studies have
shown that value cocreation is a complex concept with several definitions. For some
researchers, value cocreation is “. . . the joint, collaborative, concurrent, peer-like
process of producing a new value, both materially and symbolically” (Galvagno and
Dalli 2014). Others like Frow et al. (2011) define value cocreation as “An interactive
process, involving at least two willing resource integrating actors, which are engaged
in a particular form(s) of mutually beneficial collaboration, resulting in value
creation for those actors” (Frow et al. 2011). For Mohr and Sarin (2009) cocreation
of value involves “collaborative activities that actively engage customers in the
design and development of new innovations.” Cocreation is becoming more popular
as a practice (Hatch and Schultz 2010; Nájera-Sánchez et al. 2020).
Value cocreation has numerous advantages, including increased consumption,
positive user experiences, product stimulation, and service innovation (Gentile et al.
2007; Payne et al. 2008; Sawhney et al. 2005; Bitner et al. 2008; Morgan et al. 2018).
Customer collaboration determines market offering and requested benefits (Vargo
and Lusch 2004; Anning-Dorson 2018). The service encounter determines the
quality, quantity, and qualities of the offering (Solomon et al. 1985; Bitner et al.
2000; Santos-Vijande et al. 2016). Another benefit would be that value cocreation
enables services recovery (Dong et al. 2008; Alves et al. 2016). This situation
happens when a service failure happens and that the company solves the problem
through dialogue with the frustrated customer.
The idea of cocreation has been treated in various ways, such as in the service
dominant logic of Vargo and Lusch (2008, 2016), where value is always cocreated
by the consumer, or in the service science logic (where current resources integrate
with multiple service systems that result in a system’s well-being (Vargo et al.
2008)); and in the service logic (customers mixing resources supplied by the
company with other resources in everyday activities in value creating processes
(Grönroos 2008; Grönroos and Ravald 2011; Grönroos and Gummerus 2014;
Grönroos et al. 2015)) or other approaches like product development, many-to-
many marketing (Galvagno and Dalli 2014; Mingione and Leoni 2020).
The concept value cocreation used in this chapter is the one where customers
interact with businesses in order to cocreate value (Vargo and Lusch 2008; Galvagno
and Dalli 2014; Koskela-Huotari et al. 2016); resulting in customers giving meaning
to a company’s offering (Grönroos and Ravald 2011; Ind and Coates 2013; Morgan
et al. 2018).
The Internet has enabled a shift in the way people think about cocreation.
Cocreation in real-world settings is different from that in virtual ones (Sawhney
et al. 2005; Füller and Bilgram 2017). Internet affected cocreation of value by
fostering active, reciprocal relationships, experiential learning, and mediated inter-
actions amongst customers. Consumer tribes on the Internet (Cova and Cova 2009;
Ramaswamy and Ozcan 2016) can be sources of innovation. They enable the
development of new products as well as the commercialization of cocreated
products.
The Influence of Facebook on Value Cocreation: Evidence from the. . . 213

Social networking sites facilitate value cocreation and innovation (Sawhney et al.
2005; Törmälä and Saraniemi 2018). Companies can interact with the public quickly
and affordably (Jespersen 2010; Piller et al. 2012; Tjandra et al. 2019). Cocreation in
social media is influenced by Situated creativity (localized creativity in contingent
systems of social interaction) (Potts et al. 2008; Füller and Bilgram 2017) and
Crowd-surfing (the practice of outsourcing activities by a firm to an online commu-
nity or crowd in the form of open call—Whitla 2009). Millions of people follow the
Facebook pages of well-known companies like Coke, Nutella, and Pringles, but
there are also fewer mainstream companies like Marshmallow Peeps and Bacon Salt
that are using the platform to increase customer engagement and spark interest in
their goods (Dunay and Krueger 2011). Coke has a number of Facebook communi-
ties, some of which have millions of followers.
Through Facebook, businesses are actively listening to their customers. For
instance, 93 Facebook groups and 14,000 messages helped Cadbury relaunch a
chocolate bar named WISPA (Poynter 2007). In 2007, the Bar came back to the
shelves. This illustration demonstrates how Facebook may help with value
cocreation.

3 Theoretical Framework

Several models have been suggested to measure value cocreation (Maguire et al.
2007; Füller et al. 2009; Yi and Gong 2013), but in terms of empirical testing, the
DART model continues to be the most suitable (Albinsson et al. 2016; Mukhtar et al.
2012; Payne et al. 2008; Skaržauskaitė 2013; Tanev et al. 2011). Few scholars have
empirically tested this model. This methodology was applied in shop experiences in
Italy by Spena et al. (2012). Mazur and Zaborek (2014) used quantitative method-
ologies to apply the DART model in the Polish manufacturing and service industries.
This model was used in the context of service loyalty by Albinsson et al. (2016).
Taghizadeh et al. (2016) applied this model in the context of innovation strategy and
market performance.
The DART model was developed by Prahalad and Ramaswamy (2004) to
quantify cocreation of value. According to this theoretical framework, the cocreation
of value is determined by four constructs: discussion, access, risk management, and
transparency. The authors deﬁne dialogue as “implies interaction, deep engagement
and the ability and willingness to act on both sides” (Prahalad and Ramaswamy
2004). The ability to access information about goods and services is referred to as
access. Transparency is the frequent dissemination of accurate information. Under-
standing the advantages of a company’s offering (items or services) through dia-
logue, access, and transparency is risk assessment.
The key tenet of this model is that customers exchange knowledge about goods
and services in the market, which serves as a setting for cocreation experiences. The
ability to personalize client experiences is enabled by time, space, and events.
Experiences get personalized through cocreation. Value is based on experiences.
214 M. Hassouni et al.

Furthermore, because we are working in a technological context (Facebook brand

pages), we coupled the DART model with Davis (1989)’ Technology Acceptance
model—TAM. The reason for this selection is that the technology acceptance model
is one of the most commonly used models for technological adoption (at least
424 citations according to Lee et al. 2003). The latter model claims that perceived
usefulness and ease of use inﬂuence technology adoption. The ease-of-use factor
was eliminated since it had a similar meaning than the access factor. As a result, we
ended up with the DARTU model.

3.1 Dialogue and Value Cocreation

Dialogue is the first main construct in the DART model. Dialogue is thought to be
the most important means of engagement and knowledge sharing between a firm and
its customers (Ballantyne 2004; Ballantyne and Varey 2006; Grönroos 2004;
Prahalad and Ramaswamy 2004; Burgdorff 2018). Several studies have shown
that dialogue results in experience cocreation (Prebensen et al. 2013). Other research
demonstrated the necessity of understanding and listening to clients (Prahalad and
Ramaswamy 2004; Dong and Sivakumar 2017). Discussions can also be used to
resolve issues (Hoyer et al. 2010; Alves et al. 2016). As a result, the better the
dialogue, the better the experience cocreation. This is why fast-food companies that
want to provide their customers with a one-of-a-kind experience must use all of their
available resources. As a result, we suggest:
H1: Dialogue influences value cocreation on the fast-food Facebook pages.

3.2 Access and Value Cocreation

Because it makes exchanges more efﬁcient, access is thought to improve customer

experience (Albinsson et al. 2016). Companies must make tools and information
available to customers so they may cocreate valuable experiences (Prahalad and
Ramaswamy 2004; Ramaswamy and Ozcan 2016). Tools like social media accounts
can improve consumer experiences (Füller and Bilgram 2017). A more interactive
relationship between the client and the business is made possible by improved
information access. As a result, we come up with the following assertion:
H2: Access inﬂuences value cocreation on the fast-food Facebook pages.
The Inﬂuence of Facebook on Value Cocreation: Evidence from the. . . 215

3.3 Risk Management and Value Cocreation

Research has shown that participants in the cocreation process must perform risk
assessment (Albinsson et al. 2016; Prahalad and Ramaswamy 2004). Businesses
must explain the advantages and disadvantages of their offers (Prahalad and
Ramaswamy 2004; Hsieh and Chang 2016). Understanding the risk outcomes
enhances decision-making (Kashif et al. 2015). Improved customer relationships
are made possible through risk assessment (Garbarino and Strahilevitz 2004;
Ramaswamy and Ozcan 2016). As a result, we come up with the following assertion:
H3: Risk assessment inﬂuences value cocreation on the fast-food Facebook pages.

3.4 Transparency and Value Cocreation

The ﬁnal component of Prahalad and Ramaswamy (2004)’s paradigm is transpar-

ency. Transparency is still essential for productive communication between a busi-
ness and its clients. Trust, equality, and a better customer experience are all
influenced by transparency (Spena et al. 2012; Anning-Dorson 2018). According
to Garbarino and Strahilevitz (2004), the interactive environment benefits the orga-
nization. Additionally, information symmetry is necessary for transparency (Vargo
and Lusch 2008, 2016). As a result, we support the proposition that:
H4: Transparency influences value cocreation on the fast-food Facebook pages.

3.5 Perceived Usefulness and Value Cocreation

The deﬁnition of perceived usefulness is “a person’s subjective probability that using

a new technology will enhance his or her job performance” (Davis 1989). A
technology’s likelihood of adoption increases with its level of usefulness (Davis
1989; Morosan 2011; Morosan and DeFranco 2016). Various studies have demon-
strated the importance of a technology’s usefulness in inﬂuencing a technological
adoption (Szajna 1996; Aboelmaged 2010; Morosan 2011; Morosan and DeFranco
2016). Facebook’s usefulness plays a part in providing clients with information that
they could require. As a result, we can assume that:
H5: The perceived usefulness inﬂuences value cocreation on the fast-food Facebook
pages.
The model is presented in Fig. 1.
216 M. Hassouni et al.

Dialogue
H1
Access
H2

Risk H3 Value
Assessment Cocreation
H4

Transparency H5

Perceived
Usefulness

Fig. 1 The DARTU model

4 Methodology

The research design was quantitative. The questionnaire method was employed. The
population of this study was one of the Facebook fast-food brands’ community
members in Morocco. Recent studies show that the majority of Moroccans who use
social networking sites are young (below 25), educated, urban residents, and belong
to the Moroccan Middle class (Oukarfi 2013).
As a result, the sample used was convenient. The study took place in a university.
The basis for this choice was the convenience of data collecting as well as the fact
that in our current setting, university students tend to be fast-food consumers, and
follow fast-food brands online. An online questionnaire was distributed to students
in April 2019. 384 responses were collected.
The majority of the indicators were derived from the literature. The questionnaire
is based on the work of Solakis et al. (2017) for the dialogue, access, risk manage-
ment, and transparency variables. The items related to usefulness were derived from
Davis (1989) works. The value cocreation items were inspired from Grisseman and
Stokburger-Sauer’s (2012) works. A 1 to 5 agreement Likert Scale was used where
1 means totally disagree and 5 totally agree. Confirmed bilingual researchers trans-
lated the survey from English to French.
The questionnaire started with a knockout question, indicating the Facebook page
followed by the respondents. The respondents had to choose between McDonald’s,
Burger King, KFC, Pizza hut, Dominos, Tacos de Lyon, and other brands. If the
respondents didn’t follow any brands, they were simply discarded. Afterwards, the
respondents could answer to each of the questionnaire variables. For the dialogue,
access, risk management, transparency, usefulness, and cocreation of value vari-
ables, 4 items were used. The last section of the questionnaire concerned demo-
graphic information.
The Influence of Facebook on Value Cocreation: Evidence from the. . . 217

The PLS-SEM technique was used to test the model, which provides a “robust
framework for estimating causal models with latent variables and systems of simul-
taneous equations with measurement errors” (Henseler et al. 2009). This method is
appropriate for both exploratory and conﬁrmatory research (Lowry and Gaskin
2014). The SmartPLS V.3 software was used.

5 Results

Table 2 shows the main demographic results: 56% of the respondents were female
and 44% male; 60% of the respondents were aged between 18 and 26; 55.2% of the
respondents had a bachelor level. The most followed brands were McDonald’s,
Pizza Hut, and Burger King.
Because of its ability to handle higher order latent constructs and violations of
multivariate normality, a PLS approach was chosen. Furthermore, we used
non-parametric bootstrapping with 250 replications to obtain the estimates’ standard
errors (Hair et al. 2012, 2014; Henseler et al. 2014).
First, we evaluated the model’s quality. Following the advice of Henseler and
Sarstedt (2013), we used a consistent PLS approach and the SRMR criteria rather
than the goodness of ﬁt (GoF). In our chapter, the SRMR (Standardized Root Mean

Table 2 The demographic results

Gender
Men 44%
Women 56%
Age group
18–26 60%
27–35 31%
36–45 6%
>46 3%
Education level
Bachelor’s level 55.2%
Master’s level 28.6%
PhD level 16.1%
Brands followed
McDonald’s 22.4%
Pizza Hut 19.5%
Burger King 15.6%
Dominos 14.1%
Tacos de Lyon 12.2%
KFC 11.7%
Others 4.4%
218 M. Hassouni et al.

0.749
D2
0.817
0.761
D3
0.725

D4 dialogue

0.116
A1 C1
0.710
A2 0.818
0.761
C2
0.761 0.301 0.851
A3 0.680
0.736 0.783
C3
0.787
A4 access
cocreation of C4
0.227 value
GR1

0.730
GR2
0.753
0.754 0.196
GR3
0.662

GR4 risk management

0.167
T1

0.785
T2
0.814
0.804
T3
0.573

T4 transparency

0.831
U2
0.850
0.822
U3
0.659

U4 usefulness

Fig. 2 The overall model

Square Residual) value is 0.075 (< 0.08) indicating a good adjustment (Henseler et al.
2014; Hair et al. 2017). In other words, the difference between observed correlation and
expected correlation is acceptable (Henseler et al. 2014). The coefficient of determi-
nation (R2) was used to assess the strength of the structural model. The model showed a
value of 0.680, which is a moderate explanatory power (see Fig. 2).
Second, each concept’s reliability and convergent validity was estimated.
The convergent validity was measured by the Average Variance Extracted (AVE).
The internal reliability construct was measured by the Dillon Goldstein rho. All the
convergent validity and construct’s reliability conditions are satisfied, as shown in
Table 3 (reliability greater than 0.7 and convergent validity greater than 0.5). The
AVE (average variance extracted) ranged from 0.551 to 0.668. The construct
reliability was measured by the Dillon Goldstein Rho which ranged between 0.694
and 0.829 (see Table 3).
The Influence of Facebook on Value Cocreation: Evidence from the. . . 219

Table 3 Convergent validity and reliability indices

Latent variable Convergent validity (AVE) Reliability (Dillon-Goldstein’s Rho)
Access 0.551 0.729
Value cocreation 0.657 0.829
Dialogue 0.583 0.770
Risk management 0.617 0.694
Transparency 0.668 0.775
Usefulness 0.631 0.825

Table 4 The hypotheses testing

Original Sample Standard t statistics
sample mean deviation (|O/ p
(O) (M) (STDEV) STDEV|) values
H1 Dialogue→Value 0.116 0.120 0.041 2.810 0.005 Confirmed
cocreation
H2 Access→Value 0.301 0.295 0.049 6.172 0.000 Confirmed
cocreation
H3 Risk 0.227 0.228 0.050 4.545 0.000 Confirmed
management→Value
cocreation
H4 0.196 0.200 0.066 2.956 0.003 Confirmed
Transparency→Value
cocreation
H5 0.167 0.165 0.055 3.035 0.003 Confirmed
Usefulness→Value
cocreation

To assess discriminant validity, we used a heterotrait–monotrait (HTMT) corre-

lation ratio that was inferior to 0.9. The discriminant validity conditions were
satisfied (Henseler et al. 2015). Once the accuracy of the measurements has been
confirmed, we can evaluate the model’s structural relationships (see Fig. 2).
The hypotheses results showed that the path coefficient and the t value are all
significant at t > 1.96 and p < 0.05 (see Table 4). The relationship between Dialogue
and value cocreation on the fast-food Facebook pages was confirmed with a t test of
2.810 and a p value of 0.005. Therefore, H1 is accepted. Access shows a significant
relationship with the dependent variable, the t test is = 6.172 and the p value = 0.000.
Consequently, H2 is accepted. Risk assessment influences the value cocreation on
the fast-food Facebook pages, t = 4.545 and p = 0.000. Therefore, H3 is accepted.
The relationship between transparency and cocreation is significant with a p-value of
0.003 and a t test of 2.956. Thus, H4 is accepted. The perceived usefulness shows a
direct relationship with the value cocreation on the fast-food Facebook pages,
t = 3.035 and p = 0.003. As a result, H5 is accepted.
220 M. Hassouni et al.

6 Discussion

Prahalad and Ramaswamy (2004) proposed the DART model to explain the value
cocreation that occurs between businesses and their customers. In this chapter, we
included the perceived usefulness variable as another explanatory variable of the
value cocreation process occurring on the Facebook brand pages of fast-food
restaurants. This choice was made because the literature supported this orientation
(Davis 1989; Schiavone et al. 2014).
The chapter findings completely supported the hypothesis tested. Moroccan
consumers frequently interact with the fast-food companies they like on Facebook.
The latter dialogue can be positive or negative, but it is still a source of information
for improving service and core products. This result is coherent with previous studies
(Mainardes et al. 2017; Maduka 2016; Schiavone et al. 2014). We could explain this
result by the benefits sought by the customer (Kuo and Feng 2013; Solomon et al.
1985; Solomon 2014). Customers dialogue with fast food brands in order to obtain
information, make suggestion, or to solve problems.
The access result proved to be coherent with previous studious regarding value
cocreation (Borges et al. 2016; Ojiaku et al. 2019; Solakis et al. 2021). This could be
explained by the fact that most of the fast-food brands convey the information
needed by Facebook users in order to cocreate value. The content produced on
Facebook satisfies the need for information that Facebook users have.
The risk management result was similar than the one of several studies regarding
the DART Model (Basceanu 2014; Olejniczak and Pliszka 2019). This result could
mean that most of the content available on the Facebook pages of the fast-food
industry enables to manage risk for users. The latter can measure the advantages and
inconvenience of the content they receive.
The transparency result is coherent with other studies about the DART model
(Mainardes et al. 2017; Solakis et al. 2021; Nagarethenam et al. 2018). This could be
explained by the fact the Facebook users can clearly see positive and negative
comment regarding the different Facebook pages which shows that the content is
authentic and not manipulated.
The perceived usefulness result was similar to the one shown in previous studies
(Kim 2018; Hajli 2014; Soltani et al. 2017). This result means that the Facebook
users find it useful to interact with the fast-food brands they follow. This result
proves that Facebook users can benefit from the interaction with their favorite fast-
food brand. This result could also be explained by the potential conversational
marketing Facebook users might get from the brand they follow on Facebook.
The Influence of Facebook on Value Cocreation: Evidence from the. . . 221

7 Conclusion

Value cocreation on Facebook is still a new marketing strategy for companies in the
fast-food industry today. Facebook’s continuous improvement is expected to facil-
itate value cocreation between customers and businesses even more in the future
(Fagerstrøm and Ghinea 2011; Jalonen 2017). As a result, fast-food companies
should enable value cocreation with their customers. This chapter sought to fill a
gap in the literature on both social media marketing and cocreation.
Several conclusions can be drawn from this chapter. At the theoretical level, the
DART and TAM models were associated and evaluated, which is a rare occurrence
in the literature. This chapter is one of the few empirical papers on the topic of value
cocreation (Dhaka 2015). The Model variables allow Community managers to assess
the effectiveness of their Facebook presence at the managerial level.
From a managerial standpoint, implementing the DARTU model could provide
competitive advantages for companies in the fast-food industry. Companies could
develop new products and promotional offers as a result of customer feedback. They
can also solve customer-related problems using this model, which supports the
service recovery paradox theory (Hart et al. 1990) that states that customers are
more satisfied with companies that recover from service failures and admit their
mistakes than with those that do not fail to deliver their services. Understanding the
model would also allow fast-food companies to better target their offerings to
customers because they would know which types of customers love their products
and which don’t, thanks to social media options. Overall, having a strong presence
on a social media site like Facebook is a great way for businesses to get feedback
from their customer base and make operational improvements.
One of the chapter limitations is that it was designed for Moroccan university
students. Other universities or customer types could be considered for further
research. The non-probabilistic nature of the sample makes generalization of the
results difficult.
Another study could be conducted at the corporate level to determine the under-
standing of value cocreation that community Managers working in the fast-food
Sector can have through Facebook. The model proposed in this chapter could also be
applied to other sectors. For example, it would be interesting to test the DARTU
model in the context of banking, healthcare, or another industry.
Value cocreation is still a very context-specific research topic (Wünderlich et al.
2013). Further research could include other context-related variables related to
culture.
Further studies could include antecedents of variables such as dialogue and
relationship moderators to gain a more comprehensive understanding of the value
cocreation phenomenon. The practical outcome would be to recommend strategies
that would result in a better customer experience.
This chapter’s measurements were largely intentional. As a result, it may be
interesting to repeat the study in the same environment with the same participants
several months or years later to see if there is any change in behavior.
222 M. Hassouni et al.

References

Aboelmaged MG (2010) Predicting e-procurement adoption in a developing country: an empirical

integration of technology acceptance model and theory of planned behaviour. Ind Manag Data
Syst 110(3):392–414
Albinsson PA, Perera BY, Sautter PT (2016) DART scale development: diagnosing a Firm’s
readiness for strategic value co-creation. J Mark Theory Pract 24(1):42–58
Alves H, Fernandes C, Raposo M (2016) Value co-creation: concept and contexts of application and
study. J Bus Res 69(5):1626–1633
Anning-Dorson T (2018) Customer involvement capability and service firm performance: the
mediating role of innovation. J Bus Res 86:269–280
Ballantyne D (2004) Dialogue and its role in the development of relationship specific knowledge. J
Bus Ind Mark 19(2):114–123
Ballantyne D, Varey RJ (2006) Introducing a dialogical orientation to the service dominant logic of
marketing. In: Vargo S, Lusch R (eds) The service-dominant logic of marketing: dialogue,
debate, and directions. M.E. Sharpe, Armonk, NY, pp 224–235
Basceanu L-E (2014) Business model innovation and value cocreation based on a single case study
on Liz Claiborne. Thesis, Copenhagen business school
Bitner MJ, Brown SW, Meuter ML (2000) Technology infusion in service encounters. J Acad Mark
Sci 28(1):138–149
Bitner MJ, Ostrom AL, Morgan FN (2008) Service blueprinting: a practical technique for service
innovation. Calif Manag Rev 50(3):66–94
Borges MC, Souto JMM, da Silva AR, Araujo MV, de Meneze DC (2016) Innovation based on
value co-creation: an alternative for Brazilian wine? 39th World Congress of Vine and Wine
BIO Web of Conferences 7:1–5
Burgdorff K (2018) Dialogue as co-creation of value. In: Proceedings of the annual Global Sales
Science Institute Conference. http://gssi.world/wp-content/uploads/2019/05/Proceedings-of-
the-GSSI-Conference-2018-in-Vienna-ISSN-2510-733X.pdf. Accessed 6 Mar 2021
Challenge (2022, Jul 29) Pizza hut dépasse le cap de 50 restaurants au Maroc challenge Challenge.
ma. https://www.challenge.ma/pizza-hut-depasse-le-Cap-de-50-restaurants-aumaroc-113382/.
Accessed 14 Oct 2022
Cova B, Cova V (2009) Faces of the new consumer: a genesis of consumer Govermentality.
Recherche et Applications En Marketing (English Edition) 24(3):81–99
Davis FD (1989) Perceived usefulness, perceived ease of use, and user acceptance of information
technology. MIS Q 13(3):319–339
Dhaka S (2015) Co-creation: literature review and research issues. Int J Res –
GRANTHAALAYAH 3(2):20–37
Dong B, Sivakumar K (2017) Customer participation in services: domain, scope, and boundaries. J
Acad Mark Sci 45:944–965
Dong B, Evans KR, Zou S (2008) The effects of customer participation in cocreated service
recovery. J Acad Mark Sci 36(1):123–137
Dunay P, Krueger R (2011) Facebook marketing for dummies. Wiley and Sons, Hoboken, NJ
Fagerstrøm A, Ghinea G (2011) On the motivating impact of price and online recommendations at
the point of online purchase. Int J Inf Manag 31:103–110
Fan X, Luo Y (2020) Value co-creation: a literature review. Open J Soc Sci 8:89–98
Frow P, Payne A, Storbacka K (2011) Co-creation: a typology and conceptual framework. In:
Proceedings of 2011 Australian and New Zealand Marketing Academy (ANZMAC) Confer-
ence, Perth, Australia, 28 November-1 December, pp. 1–6
Füller J, Bilgram V (2017) The moderating effect of personal features on the consequences of an
enjoyable co-creation experience. J Prod Brand Manag 26:386–401
Füller J, Mühlbacher H, Matzler K, Jawecki G (2009) Consumer empowerment through internet-
based co-creation. J Manag Inf Syst 26(3):71–102
The Influence of Facebook on Value Cocreation: Evidence from the. . . 223

Gaber HR, Wright LT (2014) Fast-food advertising in social media. A case study on Facebook in
Egypt. J Bus Retail Manag Res 9(1):52–63
Galvagno M, Dalli D (2014) Theory of value co-creation: a systematic literature review. Manag
Serv Qual: An International Journal 24(6):643–683
Garbarino E, Strahilevitz M (2004) Gender differences in the perceived risk of buying online and
the effects of receiving a site recommendation. J Bus Res 57(7):768–775
Gentile C, Spiller N, Noci G (2007) How to sustain the customer experience: an overview of
experience components that co-create value with the customer. Eur Manag J 25(5):395–410
Grisseman US, Stokburger-Sauer NE (2012) Customer co-creation of travel services: the role of
company support and customer satisfaction with the co-creation performance. Tour Manag
33(6):1483–1492
Grönroos C (2004) The relationship marketing process: communication, interaction, dialogue,
value. J Bus Ind Mark 19(2):99–113
Grönroos C (2008) Service logic revisited: who creates value? And who co-creates? Eur Bus Rev
20:298–314
Grönroos C, Gummerus J (2014) The service revolution and its marketing implications: service
logic vs service-dominant logic. Manag Serv Qual 24(3):206–229
Grönroos C, Ravald A (2011) Service as business logic: implications for value creation and
marketing. J Serv Manag 22(1):5–22
Grönroos C, Strandvik T, Heinonen K (2015) Value co-creation: critical reflections. In:
Gummerus J, von Koskull C (eds) The Nordic school – service marketing and management
for the future. CERS, Helsinki, pp 69–81
Hair JF, Sarstedt M, Ringle CM, Mena JA (2012) An assessment of the use of partial least squares
structural equation modeling in marketing research. J Acad Mark Sci 40(3):414–433
Hair JF, Sarstedt M, Hopkins L, Kuppelwieser VG (2014) Partial least squares structural equation
modeling (PLS-SEM): an emerging tool in business research. Eur Bus Rev 26(2):106–121
Hair JF, Hollingsworth CL, Randolph AB, Chong AYL (2017) An updated and expanded assess-
ment of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hajli MN (2014) A study of the impact of social media on consumers. Int J Mark Res 56(3):
387–404
Hart CWL, Heskett JL, Sasser WE Jr (1990) The profitable art of service recovery. Harv Bus Rev
68:148–156
Hatch MJ, Schultz M (2010) Toward a theory of brand co-creation with implications for brand
governance. J Brand Manag 17(8):590–604
Henseler J, Sarstedt M (2013) Goodness-of-fit indices for partial least squares path modeling.
Comput Stat 28(2):565–580
Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. In: Sinkovics RR, Ghauri PN (eds) Advances in international market-
ing, vol 20. Emerald Publishing Limited, Bingley, pp 277–319
Henseler J, Dijkstra TK, Sarstedt M, Ringle CM, Diamantopoulos A, Straub DW et al (2014)
Common beliefs and reality about PLS: comments on Rönkkö and Evermann (2013). Organ Res
Methods 17(2):182–209
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Hoyer WD, Chandy R, Dorotic M, Krafft M, Singh SS (2010) Consumer Cocreation in new product
development. J Serv Res 13(3):283–296
Hsieh SH, Chang A (2016) The psychological mechanism of brand co-creation engagement. J
Interact Mark 33:13–26
Ind N, Coates N (2013) The meanings of co-creation. Eur Bus Rev 25(1):86–95
Jalonen H (2017) Social media as a ‘service’ for value co-creation by integrating sponsoring
companies, sports entities and fans. World Acad Sci Eng Technol 11(1):2381–2388
Jespersen KR (2010) User involvement and open innovation: the case of decision maker openness.
Int J Innov Manag 14(3):471–489
224 M. Hassouni et al.

Kaplan AM, Haenlein M (2010) Users of the world, unite! The challenges and opportunities of
social media. Bus Horiz 53(1):59–68
Kashif M, Samsi SZM, Sarifuddin S (2015) Brand equity of Lahore fort as a tourism destination
brand. Revista de Administração de Empresas 55(4):432–443
Kim J (2018) Platform adoption factors in the internet industry. Sustainability 10(9)
Koskela-Huotari K, Edvardsson B, Jonas JM, Sorhammar D, Witell L (2016) Innovation in service
ecosystems-breaking, making, and maintaining institutionalized rules of resource integration. J
Bus Res 69:2964–2971
Kuo Y-F, Feng L-H (2013) Relationships among community interaction characteristics perceived
benefits, community commitment, and oppositional brand loyalty in online brand communities.
Int J Inf Manag 33(6):948–962
Lee Y, Kozar KA, Larsen KRT (2003) The technology acceptance model: past, present, and future.
Commun Assoc Inf Syst 12:752–780
Lowry PB, Gaskin J (2014) Partial Least Squares (PLS) structural equation modeling (SEM) for
building and testing behavioral causal theory: when to choose it and how to use it. IEEE Trans
Prof Commun 57(2):123–146
Maduka OB (2016) Effects of customer value co-creation on customer loyalty in the Nigerian
service industry. Int J Bus Manag 11(12):77–82
Maguire S, Koh SCL, Huang C (2007) Identifying the range of customer listening tools: a logical
pre-cursor to CRM? Ind Manag Data Syst 107(4):567–586
Mainardes EW, Teixeira A, da Silveira Romano PC (2017) Determinants of co-creation in banking
services. Int J Bank Mark 35(2):187–204
Mazur J, Zaborek P (2014) Validating dart model. Int J Manag Econ 44(1):106–125
Mingione M, Leoni L (2020) Blurring B2C and B2B boundaries: corporate brand value co-creation
in B2B2C markets. J Mark Manag 36(1–2):72–99
Mohr JJ, Sarin S (2009) Drucker’s insights on market orientation and innovation: implications for
emerging areas in high-technology marketing. J Acad Mark Sci 37(Spring):85–96
Morgan T, Obal M, Anokhin S (2018) Customer participation and new product performance:
towards the understanding of the mechanisms and key contingencies. Res Policy 47(2):498–510
Morosan C (2011) Customers’ adoption of biometric systems in restaurants: an extension of the
technology acceptance model. J Hosp Market Manag 20(6):661–690
Morosan C, DeFranco A (2016) Co-creating value in hotels using mobile devices: a conceptual
model with empirical validation. Int J Hosp Manag 52:131–142
Mukhtar M, Ismail MN, Yahya Y (2012) A hierarchical classification of cocreation models and
techniques to aid in product or service design. Comput Ind 63(4):289–297
Nagarethenam R, Shamim A, Ghazali Z (2018) The DART perspective on value co-creation
between frontline employees and internal service providers. SHS Web of Conferences ICLM
56(2):1–7
Nájera-Sánchez J-J, Ortiz-de-Urbina-Criado M, Mora-Valentín E-M (2020) Mapping value
co-creation literature in the technology and innovation management field: a bibliographic
coupling analysis. Front Psychol 11:588648
Ojiaku OC, Nwatu BC, Aghara VN (2019) Using the DART model of value cocreation to predict
customer loyalty in pension fund administration in Nigeria. British J Manag Mark Stud 2(3):
13–26
Olejniczak M, Pliszka P (2019) Research on value co-creation with the participation of patients
based on DART model-methodological remarks. Przeglad Epidemiologiczny 73(3):393–402
Oukarfi S (2013) The use of the internet in Morocco: test for measuring the second digital divide. Int
J Innov Appl Stud 2(2):118–130
Park SY, Vargo SL (2012) The service-dominant logic approach to tourism marketing strategy. In:
Tsiotsou RH, Goldsmith RE (eds) Strategic marketing in tourism services, 1st edn. Emerald,
Bingley, pp 231–246
Payne AF, Storbacka K, Frow P (2008) Managing the co-creation of value. J Acad Mark Sci 36(1):
83–96
The Influence of Facebook on Value Cocreation: Evidence from the. . . 225

Piller FT, Vossen A, Ihl C (2012) From social media to social product development: the impact of
social media on co-creation of innovation. Die Undernimming 66(1):7–27
Potts J, Hartley J, Banks J, Burgess J, Cobcroft R, Cunningham S, Montgomery L (2008) Consumer
co-creation and situated creativity. Ind Innov 15(5):459–474
Poynter R (2007) Is Facebook the future of market research? PRWeek (online). http://www.prweek.
com/uk/news/734151/ Accessed 13 Jan 2022
Prahalad CK, Ramaswamy V (2000) Co-opting customer competence. Harv Bus Rev 78(1):79–87
Prahalad CK, Ramaswamy V (2004) Co-creating unique value with customers. Strateg Leadersh
32(3):4–9
Prebensen NK, Vittersø J, Dahl TI (2013) Value co-creation signiﬁcance of tourist resources. Ann
Tour Res 42:240–261
Ramaswamy V, Ozcan K (2016) Brand value co-creation in a digitalized world: an integrative
framework and research implications. Int J Res Mark 33:93–106
Santos-Vijande ML, Lopez-Sanchez JA, Rudd J (2016) Frontline employees’ collaboration in
industrial service innovation: routes of co-creation’s effects on new service performance. J
Acad Mark Sci 44:350–375
Sawhney M, Verona G, Prandelli E (2005) Collaborating to create: the internet as a platform for
customer engagement in product innovation. J Interact Mark 19(4):4–17
Schiavone F, Metallo C, Agrifoglio R (2014) Extending the DART model for social media. Int J
Technol Manag 66(4):271–287
Skaržauskaitė M (2013) Measuring and managing value co-creation process: overview of existing
theoretical models. Soc Technol 3(1):115–129
Solakis K, Peña-Vinces JC, Lopéz-Bonilla JM (2017) DART model from a customer’s perspective:
an exploratory study in the hospitality industry of Greece. Probl Perspect Manag (Open-Access)
15(2-si):537–549
Solakis K, Peña-Vinces J, Lopez-Bonilla JM, Aguado LF (2021) From value co-creation to positive
experiences and customer satisfaction. A customer perspective in the hotel industry. Technol
Econ Dev Econ 27(4):948–969
Solomon MR (2014) Consumer behavior: buying, having, and being. Prentice Hall Englewood
Cliffs, Englewood Cliffs, NJ
Solomon MR, Surprenant C, Czepiel JA, Gutman EG (1985) A role theory perspective on dyadic
interactions: the service encounter. J Mark 49(1):99–111
Soltani M, Jandaghi G, Forouzandeh Shahraki P (2017) Co-creation intention; presenting a model
of antecedents and its impact on attitude toward the product (case study in Shatel company). Iran
J Manag Stud 10(1):143–174
Spena RT, Carida A, Colurcio M, Melia M (2012) Store experience and cocreation: the case of
temporary shop. Int J Retail Distrib Manag 40(1):21–40
Szajna B (1996) Empirical evaluation of the revised technology acceptance model. Manag Sci 42:
85–92
Taghizadeh SK, Jayaraman K, Ismail I, Rahman SA (2016) Scale development and validation for
DART model of value co-creation process on innovation strategy. J Bus Ind Mark 31(1):24–35
Tanev S, Bailetti T, Allen S, Milyakov H, Durchev P, Ruskov P (2011) How do value co-creation
activities relate to the perception of ﬁrms’ innovativeness? J Innov Econ 7(1):131–159
Tjandra NC, Ensor J, Thomson JR (2019) Co-creating with intermediaries: understanding their
power and interest. J Bus Bus Mark 26:319–339
Törmälä M, Saraniemi S (2018) The roles of business partners in corporate brand image
co-creation. J Prod Brand Manag 27(1):29–40
Vargo SL, Lusch RF (2004) Evolving to a new dominant logic for marketing. J Mark 68(1):1–17
Vargo SL, Lusch RF (2006) Service-dominant logic: what it is, what it is not, what it might be. In:
Lush RF, Vargo SL (eds) The service-dominant logic of marketing: dialog, debate, and
directions. M. E. Shape, New York, pp 43–56
Vargo SL, Lusch RF (2008) Service-dominant logic: continuing the evolution. J Acad Mark Sci
36(1):1–10
226 M. Hassouni et al.

Vargo SL, Lusch RF (2016) Institutions and axioms: an extension and update of service-dominant
logic. J Acad Mark Sci 44:5–23
Vargo SL, Maglio PP, Akaka MA (2008) On value and value co-creation: a service systems and
service logic perspective. Eur Manag J 26(3):145–152
Wahabi K (2016) Pourquoi investir dans la restauration rapide? Retrieved from: http://www.2m.ma/
fr/programme/20161220-pourquoi-investir-dans-la-restaurationrapide/
We are social & Kepios (2022, Feb 15) Digital 2022: Morocco; Data Reportal – Global Digital
Insights. DataReportal – Global Digital Insights. https://datarportal.com/reports/digital-2022-
morocco#:~:text=Morocco’s%20internet%20penetration%20rate%20stood,percent)%20
btween%202021%20and%202022 Accessed 14 Oct 2022
Whitla P (2009) Crowdsourcing and its applications in marketing activities. Contemp Manag Res
5(1):15–28
Wünderlich NV, Wangenheim FV, Bitner MJ (2013) High tech and high touch: a framework for
understanding user attitudes and behaviors related to smart interactive services. J Serv Res
16(1):3–20
Yi Y, Gong T (2013) Customer value co-creation behavior: scale development and validation. J Bus
Res 66(9):1279–1284
Part V
Human Resource Management
and Organizational Design and Behavior
Research on Interaction Pattern of Virtual
Team in Crisis Situation

Shen Yuanyanhang, Jiale Wu, and Xiaodan Yu

1 Introduction

With the rapid development of economy and technology, people are facing more and
more frequent and serious disasters and crises. Crisis is a low-probability, high-
impact event with huge pressure and fuzzy uncertainty (Yu et al. 2008). The negative
impact of its continuous development process may spread to other organizations in
the same industry, posing a significant threat to organizations and society (Bies
2013). In the face of the crisis, how to respond to crises quickly and efficiently is an
urgent problem. More and more scholars are committed to studying the causes of the
crisis and how to minimize the damage caused by the crisis (Camillo 2015).
Organizational research has found that teams are increasingly used in crisis
because teams can respond quickly to dynamic, complex crisis situations (Burke
et al. 2006). With the rapid development of information technology, the traditional
teams are no longer dominant. The virtual team gained wide attention and recogni-
tion in research (Han and Beyerlein 2016).
Zellmer-Bruhn et al. (2004) have pointed out that the effectiveness of a team is
not determined by individual actions and discourses, but interaction patterns of team
members during the task. Interaction patterns have gradually become one of the
perspectives of the team effectiveness. Coping with the crisis in the unfamiliar
situation in virtual teams is difficult. The establishment of the interaction pattern

S. Yuanyanhang
School of Information, Renmin University of China, Beijing, China
e-mail: [email protected]
J. Wu · X. Yu (✉)
School of Information Technology & Management, University of International Business and
Economics, Beijing, China
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 229

enables the virtual team members to communicate and communicate more fully and
ensure that the task can be ﬁnished successfully (Espinosa et al. 2015). However,
there is little research in China to explore how the interaction pattern of virtual teams
in a crisis environment affects team effectiveness.
In addition, the crisis will bring emotional ﬂuctuations. Related studies have
shown that emotional responses to existing situations have an impact on the extent
of crisis decision-making (Kaplan et al. 2013). Emotions greatly affect team life, and
it is feasible to use emotion management to regulate team behavior and improve
team effectiveness. However, research on emotions in China is relatively rare. The
previous research pays more attention to the individual’s emotional changes, and the
team-level emotional research is relatively scarce. Also, there are few researches
exploring the mediating role of team emotions between team interaction and team
effectiveness.
Therefore, this paper introduces the mediation of team emotion, explores the
relationship between interaction patterns and team effectiveness in crisis situations,
and the intermediary mechanism of team emotion between the two. According to
this, we could give some suggestions about virtual teams in crises. Our research
questions are:
1. What is the relationship between virtual team members’ interaction patterns, team
emotions, and team effectiveness in a crisis situation?
2. Can team emotions mediate the relationship between interaction patterns and
team effectiveness?
3. How to improve the team performance through management of emotions and
behavioral patterns?

2 Research Model and Hypotheses

2.1 Interaction Pattern and Team Emotion

Interaction pattern is the conventional verbal expression and non-verbal behavior

used for collective action and coordination. Through these verbal or non-verbal
behaviors, teams can share their own emotional states and gradually reach an
emotional integration state by combining the external environment and the shaping
of team norms (Kelly and Barsade 2001), which indicates that the interaction pattern
has an important influence on the formation of team emotional atmosphere.
Spector and Fox (2002) found that the more open the communication, the better
the atmosphere of the team. This is because the stable interaction pattern is condu-
cive to the establishment of a fair and just team atmosphere, from which team
members can gain recognition, feel a sense of accomplishment, and greatly increase
positive emotions, thus creating a good team atmosphere. Levine (2010) found that
during emergencies, sufficient information exchange among team members would
affect the emotional factors within the team. A team that establishes a stable
interaction pattern has stronger flexibility and adaptability can better deal with
Research on Interaction Pattern of Virtual Team in Crisis Situation 231

unconventional or non-programmed risks, and maintain a consistent emotional

atmosphere. Based on the above analysis, we propose the following hypothesis:
H1: In a crisis situation, the stable interaction pattern of virtual team has a positive
effect on the formation of positive team emotions.

2.2 Interaction Pattern and Team Effectiveness

The interaction patterns are quite different, and different behavioral activity patterns
indeed influence team outcomes. Some scholars have conducted empirical studies on
behavioral patterns and linked such differences with team effectiveness (Zellmer-
Bruhn et al. 2004). The interaction pattern of the team is particularly important in the
crisis environment. The team changes the interaction pattern according to the
feedback of the environment and form a unique interaction pattern, which has
stronger adaptability, leading the team to solve problems more efficiently (Jehn
and Mannix 2001).
Studies have found that teams that have established a stable interaction pattern
during a crisis have better team performance, because the smaller the change
frequency of communication mode, the faster the team can make consistent deci-
sions through stable interaction (Daniel et al. 2013). Veil et al. (2011) found that the
establishment of interaction pattern could reduce the frequency of roundtrip com-
munication and save time. Such communication was effective in irregular or emer-
gency situations, and the mode characterized by simple but uncomplicated
interaction predicted higher team efficiency.
There are also researchers (Waller et al. 2004; Stachowski et al. 2009) who
observed team effectiveness and member behavior through simulated crisis exper-
iment. They found that during the simulated crisis, the nuclear team members with
better performance constructed a set of interaction patterns belonging to their own
team, helped the team promote knowledge sharing and interaction, and coordinated
actions to achieve the ultimate goal more efficiently. Grawitch et al. (2003) verified
that in a virtual team, the stable interaction pattern of team members led them to
interpret the situation in a timely, clear, and reasonable way, improved the team’s
response, understanding, information-sharing and decision-making ability, and dem-
onstrated better creativity and efficiency in the face of crisis.
Based on the above, we make the following assumptions:
H2: In a crisis situation, the stable interaction pattern of virtual team has a positive
effect on team effectiveness.
H2a: In a crisis situation, the stable interaction pattern of virtual team has a positive
effect on team performance.
H2b: In a crisis situation, the stable interaction pattern of virtual team has a positive
effect on team satisfaction.
H2c: In a crisis situation, the stable interaction pattern of virtual team has a positive
effect on team commitment.
232 S. Yuanyanhang et al.

2.3 Team Emotion and Team Effectiveness

In the complex and changeable crisis situation, emotion plays a crucial role in
searching for information and dealing with abnormal situations. As one of people’s
basic psychological processes, emotion affects people’s decision-making capability
and behavior. Huy (2011) and Elfenbein (2007) have proved that positive emotions
have a positive impact on team performance. Positive emotions are manifested as a
state of full energy and devotion, leading to positive behaviors.
Through experiments, Barsade et al. (2000) found that the level of emotional
difference between virtual team members was associated with their satisfaction. The
crisis situation is different from the general situation. The virtual team members have
a stronger sense of crisis and sense of urgency due to time constraints, dangerous and
urgent tasks, and the emotional ﬂuctuations generated from it will have a more
prominent impact on the team’s behavior and decision-making. Based on relevant
research results, we propose the following hypotheses:
H3: In a crisis situation, the positive team emotion of virtual team has a positive
effect on team effectiveness.
H3a: In a crisis situation, the positive team emotion of the virtual team has a positive
effect on the team’s performance.
H3b: In a crisis situation, positive team emotions of virtual teams have a positive
effect on team satisfaction.
H3c: In a crisis situation, the positive team emotions of the virtual team have a
positive effect on the team’s commitment.
Combined with relevant literature, existing research results, and emotional mean-
ing construction theory, this paper sorted out the research model of interaction
pattern, team emotion, and team effectiveness, as shown in the Fig. 1.

3 Methodology

3.1 Research Design

This study employed experiments to explore the interaction pattern of virtual teams
in crisis situations, the mechanism between team emotions and team effectiveness,
and further explore the mediating role of positive team emotions.
In order to determine the best time to create a sense of crisis, a pre-experiment was
conducted before the formal experiment. In the pre-experiment, set the experiment
time to 25 min. In the remaining 15 min, the participants were informed that the
Research on Interaction Pattern of Virtual Team in Crisis Situation 233

Team Effectiveness

Interaction Team
H2a(+)
Pattern Performance

H2b(+)

H2c(+)
H1(+)
Team
Satisfaction
H3a(+)

H3b(+)

Team Team
H3c(+)
Emotion Commitment

Fig. 1 Theoretical model

Table 1 Scales
Construct Scale
Crisis Gifford et al. (1979)
Team emotion Qu (2007), “Team Atmosphere Scale”
Interaction pattern Stachowski et al. (2009)
Team effectiveness Jehn and Chatman (2000)

experiment time was only 10 min due to uncontrollable situation, thus creating
pressure and tension to simulate the crisis situation.

3.2 Measurement

This study adopts the mature scale to ensure a certain degree of reliability and
validity. A 5-level scale was adopted in the design of the questionnaire. The scale
selection is shown in the following Table 1.
Cronbach alpha was used to test whether the project had a high consistency,
which is summarized in Table 2. The value of Cronbach alpha for each variable is
above 0.7, indicating a high internal consistency among all factors.
The scale used by previous scholars is selected in this study, and the content
scope is well deﬁned. In addition, the selected scale in this study has undergone a
large number of tests and is representative in the use of relevant constructs. There-
fore, the questionnaire used in this study has certain content validity.
234 S. Yuanyanhang et al.

Table 2 Cronbach’s alpha coefﬁcients of each factor

Dimension Cronbach’s alpha
Interactive pattern 0.89
Team emotion 0.91
Team performance 0.79
Team satisfaction 0.82
Team commitment 0.79
Crisis 0.71

Table 3 Descriptive analysis

Min Max Average Standard deviation Skewness Kurtosis
Crisis 2.60 4.20 3.51 0.414 0.356 0.169
Team emotion 1.29 4.86 4.13 0.810 0.118 0.899
Interactive pattern 2.60 4.60 3.65 0.422 0.688 0.391
Team performance 3.33 5.00 4.28 0.393 0.800 0.626
Team satisfaction 2.75 5.00 4.06 0.385 0.561 1.085
Team commitment 3.00 4.67 3.93 0.390 0.376 0.053

4 Results

4.1 Statistical Description

The purpose of this study is to explore the influence mechanism between the
interaction pattern, team emotion and team performance of virtual teams in crisis
situations. The mean, standard deviation, and kurtosis of each research variable are
shown in Table 3. Likert five-level scale is adopted to measure the data, and the
values of the scale can illustrate the different states of the subjects on the scale.
The results in the table show that the research subjects did have a sense of crisis
during the experiment (crisis mean value 3.51). The experimental data are placed
under the 95% confidence interval, and it can be seen that the skewness and kurtosis
of each variable are within ±1.96, which conforms to the normal distribution, and
subsequent structural equation model analysis can be carried out.

4.2 Preliminary Analysis

Structural equation model (SEM) is a statistical method used to express the relationship
between potential variables and observed variables as well as the relationship between
potential variables. The method of PLS-SEM selected in this study has a better model
predictability than other structural models (Hair et al. 2019; Hair et al. 2022).
Composite reliability (CR) was used to measure the observed values to describe
reliability and consistency of the corresponding latent variables. CR value above 0.7
Research on Interaction Pattern of Virtual Team in Crisis Situation 235

Table 4 Composite reliability (CR)

Variable Composite reliability (CR)
Interaction pattern 0.92
Team emotion 0.93
Team performance 0.88
Team satisfaction 0.88
Team commitment 0.87

Table 5 Convergent validity and discriminant validity

Construct AVE 1 2 3 4 5
1. Interaction pattern 0.610
2. Team emotion 0.643 0.582
3. Team performance 0.701 0.740 0.198
4. Team satisfaction 0.654 0.405 0.652 0.605
5. Team commitment 0.704 0.835 0.891 0.484 0.676

Table 6 Path coefﬁcients

Sample Standard t- p-
Path mean deviation statistic values
Interaction pattern → Team emotion 0.897 0.046 19.826 0.000
Interaction pattern → Team 0.439 0.142 3.093 0.002
performance
Interaction pattern → Team 0.451 0.114 4.070 0.000
satisfaction
Interaction pattern → Team 0.405 0.123 3.327 0.001
commitment
Team emotion → Team performance 0.431 0.146 3.028 0.002
Team emotion → Team satisfaction 0.427 0.104 4.083 0.000
Team emotion → Team commitment 0.485 0.130 3.762 0.000

indicates good internal consistency. According to Table 4, the CR values of the

latent variables are all greater than 0.7, indicating good internal consistency.
Average extraction variance (AVE) and heterotrait–monotrait ratio (HTMT) are
used to judge convergence validity and discriminant validity, respectively. The
establishment of convergence validity requires the average extraction variance
(AVE) to be greater than 0.5. HTMT is used for discriminant validity, which requires
that the HTMT value is less than 0.90. The convergent validity and discriminant
validity of the model measured in this study are shown in Table 5, indicating enough
convergent validity and discriminant validity.
In this study, bootstrapping was used on the path coefficient of the structural
model to test whether the causal relationship assumed by the model was established.
The results of the path coefficient significance test are shown in Table 6.
236 S. Yuanyanhang et al.

As can be seen from Table 6, interaction pattern affects team emotion and the
three dimensions of team effectiveness. Team emotion affects team performance,
team satisfaction, and team commitment. The research model is valid.

5 Conclusion

Based on the team perspective, this study explores the interaction pattern, team
emotion, and team effectiveness of virtual teams in crisis situations, and further
examines the mediating role of team emotion between them. The results indicate that
the interaction pattern has a positive impact on the formation of team positive
emotions and team effectiveness. The positive emotions of the team have a direct
impact on team effectiveness. The interaction pattern is mediated by team emotions
and has an indirect positive effect on team effectiveness. In the future, we would use
mixed methods to obtain more diverse data.

References

Barsade SG, Ward AJ, Turner JDF, Sonnenfeld JA (2000) To your Heart’s content: a model of
affective diversity in top management teams. Adm Sci Q 45(4):802–836
Bies RJ (2013) The delivery of bad news in organizations: a framework for analysis. J Manag 39(1):
136–162
Burke CS, Stagl KC, Salas E, Pierce L, Kendall D (2006) Understanding team adaptation: a
conceptual analysis and model. J Appl Psychol 91(6):1189–1207
Camillo AA (2015) Strategic management and crisis communication interdependence in the global
context: a preliminary investigation. Emerg Economy Stud 1(1):37–49
Daniel S, Agarwal R, Stewart KJ (2013) The effects of diversity in global, distributed collectives: a
study of open source project success. Inf Syst Res 24(2):312–333
Elfenbein HA (2007) Emotion in organizations: a review and theoretical integration. Acad Manag
Ann 1(1):315–386
Espinosa AJ, Nan N, Carmel E (2015) Temporal distance, communication patterns, and task
performance in teams. J Manag Inf Syst 32(1):151–191
Gifford WE, Bobbitt HR, Slocum JW (1979) Message characteristics and perceptions of uncertainty
by organizational decision makers. Acad Manag J 22(3):458–481
Grawitch MJ, Munz DC, Kramer TJ (2003) Effects of member mood states on creative performance
in temporary workgroups. Group Dyn Theory Res Pract 7(1):41–54
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult TM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. SAGE, Los Angeles
Han SJ, Beyerlein M (2016) Framing the effects of multinational cultural diversity on virtual team
processes. Small Group Res 47(4):351–383
Huy QN (2011) How middle managers’ group-focus emotions and social identities influence
strategy implementation. Strateg Manag J 32(13):1387–1410
Jehn KA, Chatman JA (2000) The influence of proportional and perceptual conflict composition on
team performance. Int J Conflict Manag 11(1):56–73
Research on Interaction Pattern of Virtual Team in Crisis Situation 237

Jehn KA, Mannix EA (2001) The dynamic nature of conflict: a longitudinal study of intragroup
conflict and group performance. Acad Manag J 44(2):238–251
Kaplan S, Laport K, Waller MJ (2013) The role of positive affectivity in team effectiveness during
crises. J Organ Behav 34(4):473–491
Kelly JR, Barsade SG (2001) Mood and emotions in small groups and work teams. Organ Behav
Hum Decis Process 86(1):99–130
Levine EL (2010) Emotion and power (as social influence): their impact on organizational citizen-
ship and counterproductive individual and organizational behavior. Hum Resour Manag Rev
20(1):4–17
Qu R (2007) Theory and demonstration of team emotional atmosphere. Economic Science Press,
Beijing
Spector PE, Fox S (2002) An emotion-centered model of voluntary work behaviors: some parallels
between counterproductive work behaviors and organizational citizenship behaviors. Hum
Resour Manag Rev 12(2):269–292
Stachowski AA, Kaplan SA, Waller MJ (2009) The benefits of flexible team interaction during
crises. J Appl Psychol 94(6):1536–1543
Veil SR, Buehner T, Palenchar MJ (2011) A work-in-process literature review: incorporating social
media in risk and crisis communication. J Contingencies Crisis Manage 19(2):110–122
Waller MJ, Gupta N, Giambatista RC (2004) Effects of adaptive behaviors and shared mental
models on control crew performance. Manag Sci 50(11):1534–1544
Yu T, Sengul M, Lester RH (2008) Misery loves company: the spread of negative impacts resulting
from an organizational crisis. Acad Manag Rev 33(2):452–472
Zellmer-Bruhn M, Waller MJ, Ancona D (2004) The effect of temporal entrainment on the ability of
teams to change their routines, research on managing groups and teams. Elsevier Science Press,
Oxford, UK
Predicting the Performance of New Hires:
The Role of Humility, Interpersonal
Understanding, Self-Confidence,
and Flexibility

Debolina Dutta, Chaitali Vedak, and Varghees Joseph

1 Introduction

External recruitment has become a dominant talent management strategy for many
organizations (Gërxhani and Koster 2015). It is increasingly becoming more critical
to study the effect of the alignment of human capital to business objectives and its
return on investments (Zula and Chermack 2007; Becker and Huselid 2006). The
focus on determining a new hire’s performance expands on the human capital theory
(HCT) by demonstrating the value created by human resource (HR) policies and
practices (Zula and Chermack 2007). Due to the imbalance between the rapid rate of
employee recruitment and the shortage of talent, substantial organizational invest-
ments in recruiting and training go unrealized (Selden et al. 2013; Harris et al. 2011).
Therefore, it becomes an essential requirement for new recruits to meet the organi-
zation’s expectations (Cappelli 2019) and deliver on-the-job performance (Breaugh
2014; Gërxhani and Koster 2015; Murphy 2016).
This chapter focuses on this critical yet neglected gap in existing research on
recruitment and recruitment outcomes and contributes in three ways. First, by
employing a longitudinal survey spanning over 16 months, we examine the traits
and virtues of applicants recruited by an organization and the subsequent post-hire
individual performance. Second, previous studies have based their ﬁndings on
experimental design, mainly using student respondents, to determine perceptions
of performance outcomes, which suffer from serious problems of common method
variance. To mitigate this bias, we used real-life applicants and organizational
performance reviews instead of self-reported reviews. Third, we did not limit the
assessment of new hire performance to a few roles, but looked at a spectrum of job

D. Dutta (✉) · C. Vedak · V. Joseph

Indian Institute of Management, Bangalore, India
e-mail: [email protected]; [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 239

roles and seniorities to determine common traits and virtues enabling new hire
performance. This, therefore, increases the generalizability of our ﬁndings.

2 Theoretical Grounding and Hypotheses

We draw on the Human Capital Theory (HCT) to argue that it is the responsibility of
Human Resource Development (HRD) practitioners to assess the impact on perfor-
mance and sustain competitive advantage (Zula and Chermack 2007). HCT suggests
that human capabilities acquired by human resources that are durable traits yield
positive effects on performance in socially valued activities (David 2001; Becker
and Huselid 2006; Subramony et al. 2008). Therefore, human capital frameworks
emerge from the idea that labor inputs are not merely quantitative (Zula and
Chermack 2007).
Both in eastern and western cultures, humility is a multifaceted strength and is
primarily regarded as a universal virtue (Argandona 2015; Li et al. 2021; Morris
et al. 2005; Owens et al. 2011) and a significant body of work substantiates the
positive influence of humble personnel in the workplace at various levels (Owens
and Hekman 2016; Owens et al. 2013; Rego et al. 2019; Sousa and van Dierendonck
2017; Zhang et al. 2017; Vera and Rodriguez-Lopez 2004).
H1: Humility significantly influences the performance of new hire.
Interpersonal understanding (IPU) refers to the individual’s capability to show
empathy and look at things from others’ points of view (Knott and Kayes 2012).
With this capability, employees develop multiple social ties with different colleagues
to improve their performance, and the strength of these ties depends on employees’
level of societal understanding of their workplace (Borgatti et al. 2009).
H2a: Interpersonal understanding significantly influences the performance of new
hire.
H2b: Humility mediates the relationship between Interpersonal understanding and
the performance of new hire.
Self-confidence is an individual’s perception of control over themselves and their
environments based on their past experiences (Mayo et al. 2012), which increases
the individuals’ probability of performing well (Compte and Postlewaite 2004).
H3a: Self-confidence significantly influences the performance of new hire.
H3b: Humility mediates the relationship between Self-confidence and the perfor-
mance of new hire.
Flexibility, on the other hand, is a key requirement for numerous organizations
facing uncertain environments and changing customer demands (Molleman and van
den Beukel 2007). Research demonstrates the importance of this behavior, arguing
that employees must have a wide range of responses to effectively react to different
situations, while preserving their integrity and trustworthiness (Mayo et al. 2012).
Predicting the Performance of New Hires: The Role of Humility,. . . 241

H4a: Flexibility signiﬁcantly inﬂuences the performance of new hire.

H4b: Humility mediates the relationship between flexibility and the performance of
new hire.
Extending this knowledge, we argue that humble employees having sound
interpersonal understanding at the workplace, who maintain self-confidence and
retain flexibility in daily tasks, are critical in improving employee performance.
Therefore, in this chapter, we framed seven hypotheses supporting our claim.

3 Data and Methods

3.1 Research Design

We used a ﬁeld design method with real job applicants spanning over two years to a
large subsidiary of a multinational organization operating in India to understand
applicant trait and their subsequent impact on job performance. The organization
specializes in energy storage and management solutions, including solar energy, has
operated for over 30 years and has sold its products in more than 36 countries. We
captured the new hires’ age, gender, and education from the company records and
the new hires went through a yearly company-wide performance review cycle in the
subsequent year. The HR department initiated capturing employee performance.

3.2 Measures and Control Variables

Well-validated and standardized questionnaires were used to collect data on all

items. Humility of the new hire was captured by a 9-item scale containing items
(Owens et al. 2013). While self-conﬁdence was measured by the three items scale
(Mayo et al. 2012), ﬂexibility was measured by using a 3-item scale (Mayo et al.
2012). IPU was measured using a 3-item scale (Mayo et al. 2012).
Consistent with past research on new hire performance, we controlled for age
(Selden et al. 2013; Ng and Feldman 2008), gender (Liu et al. 2008), education
(Ng and Feldman 2009), and tenure (Kuipers and Giurge 2017) of the new hires.

4 Analysis and Findings

In order to control for common method variance (CMV), we used Harman’s single
factor score, in which all items (measuring latent variables) are loaded into one
common factor where the variance extracted was 40.41% (less than the threshold of
50%), indicating CMV is not a major concern in the analysis (Podsakoff et al. 2012).
242 D. Dutta et al.

In determining discriminant validity, we assessed the heterotrait–monotrait (HTMT)

ratio of correlations among the constructs (Henseler et al. 2015) and the constructs’
composite reliability and discriminant validity were assessed and within limits.
Humility influences new hire performance (β = 0.328, p < 0.01), hence hypoth-
esis 1 was supported. While IPU was significant in predicting new hire performance
(β = -0.258, p < 0.01), supporting Hypothesis 2a, the negative correlation with
performance indicates a suppressor effect of the mediating variable indicating full
mediation (Hair et al. 2014, p. 225). Therefore, hypothesis 2b was also supported.
However, self-confidence (β = 0.029, p > 0.05) and flexibility (β = -0.070,
p > 0.05) did not have a significant effect on new-hire performance. Hence,
hypotheses 3a and 4a were not supported. Other than that, the control variables
such as age, gender, and qualification were not found to have any significant
relationship with a new-hire performance at the workplace: Since the sample pop-
ulation is collected from a specific organization, the control variables of the partic-
ipants may not vary widely enough to have a significant impact on the results.
The bootstrapping procedure with 5000 resampling runs (Hair et al. 2021) was
performed to assess the three mediation hypotheses. The relationship between IPU
and humility was significant (β = 0.392, p < 0.01). Due to the suppressor effect,
humility wholly mediates the effect of IPU on the new-hire performance. Hence,
hypothesis 2b was supported. Flexibility (β = 0.162, p < 0.01) and SC (β = 0.329,
p < 0.01) were significantly influencing humility. However, there was no direct
influence of flexibility and SC on employee performance. We find that humility
wholly mediates the relationship between flexibility and newly hired employees’
performance as well as the relationship between self-confidence and newly hired
employees’ performance. Thus, Hypotheses 3b and 4b were supported.

5 Discussion

This chapter’s primary contribution is to build a model that emphasizes the critical
traits of new hires and their on-job performance, an essential aspect of human capital
that has largely been unexplored. The chapter provides a comprehensive systematic
understanding of the dynamics and impact of the new hire traits and the interplay
with humility and contributes to the body of literature on recruitment by identifying
relevant traits of new hires that link to employee performance. As the findings are
based on field experiment data involving real job applicants across multiple job
families and roles, using real on-job performance data of new hires, it increases the
validity of the findings. The findings indicate that humility wholly mediates the
influence of self-confidence and flexibility on performance. These results are
extremely relevant for practitioners, as it provides guidance on essential traits to
be screened for in new hires.
Predicting the Performance of New Hires: The Role of Humility,. . . 243

6 Implications and Limitations

Despite the important contributions of the chapter, as with most research, it is not
without limitations that should be acknowledged. While this chapter provides
direction toward critical individual traits within applicants that could lead to superior
on-the-job performance, the survey was based on a sample from an Indian MNC, and
the findings may be specific to this particular context. In the interest of parsimony,
we were able to focus on a limited set of traits that could impact the on-the-job
performance of new hires. Research indicates multiple other traits and competencies,
apart from personality traits, which could provide rich direction to practitioners and
help recruitment research. Future studies may explore the influence of additional
constructs on new hire performance.
The implications of our research are significant and multifaceted. By acquiring a
holistic understanding of the traits involved in high-performance organizations can
develop focused training and development programs, improve the onboarding pro-
cess, provide managers with tools and strategies for effectively leading and
mentoring new hires, and improve retention by providing new hires with develop-
ment opportunities that are aligned with their strengths.

Acknowledgments We thank the anonymous reviewers for their valuable feedback and for
helping us to improve our paper.

References

Argandona A (2015) Humility in management. J Bus Ethics 132(1):63–71

Becker BE, Huselid MA (2006) Strategic human resources management: where do we go
from here? J Manag 32(6):898–925
Borgatti SP, Mehra A, Brass DJ, Labianca G (2009) Network analysis in the social sciences.
Science 323(5916):892–895
Breaugh JA (2014) Predicting voluntary turnover from job applicant biodata and other applicant
information. Int J Sel Assess 22(3):321–332
Cappelli P (2019) Your approach to hiring is all wrong. Harv Bus Rev 97(3):48–58
Compte O, Postlewaite A (2004) Conﬁdence-enhanced performance. Am Econ Rev 94(5):
1536–1557
David PA (2001) Knowledge, capabilities and human capital formation in economic growth.
New Zealand Treasury Working Paper, New Zealand, 01/13
Gërxhani K, Koster F (2015) Making the right move. Investigating employers’ recruitment strat-
egies. Pers Rev 44(5):781–800
Hair JJ, Sarstedt M, Ringle CM, Hult GT (2014) A primer on partial least squares structural
equation modeling (PLS-SEM), vol 5. Sage, Thousand Oaks, CA, p 105
Hair JF, Hult GT, Ringle CM, Sarstedt M (2021) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage Publications, Thousand Oaks
Harris JG, Craig E, Light DA (2011) Talent and analytics: new approaches, higher ROI. J Bus
Strateg 32(6):4–14
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
244 D. Dutta et al.

Knott MJ, Kayes CD (2012) Individual contribution to a team: the importance of continuous
adaptive learning. Organ Manag J 9(1):22–33
Kuipers BS, Giurge LM (2017) Does alignment matter? The performance implications of HR roles
connected to organizational strategy. Int J Hum Resour Manag 28(22):3179–3201
Li R, Zhang H, Zhu X, Li N (2021) Does employee humility Foster performance and promotability?
Exploring the mechanisms of LMX and peer network centrality in China. Hum Resour Manag
60(3):399–413
Liu C, Spector PE, Shi L (2008) Use of both qualitative and quantitative approaches to study job
stress in different gender and occupational groups. J Occup Health Psychol 13(4):357–370
Mayo M, Kakarika M, Pastor JC, Brutus S (2012) Aligning or inflating your leadership self-image?
A longitudinal study of responses to peer feedback in MBA teams. Acad Manag Learn Edu
11(4):631–652
Molleman E, van den Beukel A (2007) Worker flexibility and its perceived contribution to
performance: the moderating role of task characteristics. Human Factors and Ergonomics in
Manufacturing & Service Industries 17(2):117–135
Morris AJ, Brotheridge CM, Urbanski JC (2005) Bringing humility to leadership: antecedents and
consequences of leader humility. Hum Relat 58(10):1323–1350
Murphy JP (2016) Quality of hire: data makes the difference. Employ Relat Today 43(2):5–15
Ng TW, Feldman DC (2008) The relationship of age to ten dimensions of job performance. J Appl
Psychol 93(2):392–423
Ng TW, Feldman DC (2009) How broadly does education contribute to job performance? Pers
Psychol 62(1):89–134
Owens BP, Hekman DR (2016) How does leader humility influence team performance? Exploring
the mechanisms of contagion and collective promotion focus. Acad Manag J 59(3):1088–1111
Owens BP, Rowatt WC, Wilkins AL (2011) Exploring the relevance and implications of humility in
organizations. In: Handbook of positive organizational scholarship, vol 1. Oxford University
Press, Oxford, pp 260–272
Owens BP, Johnson MD, Mitchell TR (2013) Expressed humility in organizations: implications for
performance, teams, and leadership. Organ Sci 24(5):1517–1538
Podsakoff PM, MacKenzie SB, Podsakoff NP (2012) Sources of method bias in social science
research and recommendations on how to control it. Annu Rev Psychol 63:539–569
Rego A, Owens B, Yam KC, Bluhm D, Cunha PM, Silard A et al (2019) Leader humility and team
performance: exploring the mediating mechanisms of team PsyCap and task allocation effec-
tiveness. J Manag 45(3):1009–1033
Selden S, Schimmoeller L, Thompson R (2013) The influence of high-performance work systems
on voluntary turnover of new hires in US state governments. Pers Rev 42(3):300–323
Sousa M, van Dierendonck D (2017) Servant leadership and the effect of the interaction between
humility, action, and hierarchical power on follower engagement. J Bus Ethics 141(1):13–25
Subramony M, Krause N, Norton J, Burns GN (2008) The relationship between human resource
investments and organizational performance: a firm-level examination of equilibrium theory. J
Appl Psychol 93(4):778–788
Vera D, Rodriguez-Lopez A (2004) Strategic virtues: humility as a source of competitive advan-
tage. Organ Dyn 33(4):393–408
Zhang H, Ou AY, Tsui AS, Wang H (2017) CEO humility, narcissism and firm innovation: a
paradox perspective on CEO traits. Leadersh Q 28(5):585–604
Zula KJ, Chermack TJ (2007) Intergrative literature review: human capital planning: a review of
literature and implications for human resource development. Hum Resour Dev Rev 6(3):
245–262
Mindful Leadership Under Fire:
A Validation Study of a Hierarchical
Component Model

Stephanie Dygico Gapud and Joseph F. Hair Jr.

1 Introduction

The COVID-19 pandemic created various challenges for organizations, including

supply chain issues, implementation of work arrangements, the safety of employees,
managing financial impact, adapting to changes in consumer behavior, and navigat-
ing government regulations. Some organizations were surprised when things did not
return to normal quickly, and delays in recognizing the paradigm shift led to
decreased efficiency and potential harm (Eastwood 2020; Cecere 2021; Branswell
2022; Raghavan et al. 2021; Sneader and Sternfels 2020; Tucker 2020). High-
reliability organizations, such as those in healthcare and air traffic control, are able
to consistently avoid catastrophic events through a culture of safety, continuous
learning, and mindful leadership. Mindful leadership emerges when team members
participate in a dynamic influencing process during times of uncertainty.
To improve team performance when dealing with uncertainty, organizations
should learn from wildland firefighting teams (Weick and Sutcliffe 2007), which
share authority and responsibility and have a less hierarchical managerial approach
(Gapud 2019). Understanding differences in shared leadership environment and
behavior can help improve leadership strategies and achieve a shared purpose of
safely completing missions. Research has shown that sharing authority with those
who have knowledge, skills, and abilities improves team performance, and a less
hierarchical managerial approach leads to better performance in a hierarchical

S. D. Gapud (✉)
Division of Business, Spring Hill College, Mobile, AL, USA
e-mail: [email protected]
J. F. Hair Jr.
Mitchell College of Business, University of South Alabama, Mobile, AL, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 245

organization (Bernstein 2015; Edmondson 2012; Gapud 2019; Fraher et al. 2016;
Langer 2010; Weick and Sutcliffe 2007). Gapud and Hair prepared a chapter in this
book entitled “When Navigating Uncertainty Lead Mindfully.” Their results pro-
vided relevant components for effectively implementing Auftragstaktik.

1.1 Auftsragstaktik

The concept of decentralizing command in the military by fully trusting in the

capabilities of, and giving leeway to, those on the frontlines to make strategic
decisions based on the command intent is called Auftragstaktik [AT] (Nelsen
1987; Shamir 2010; Wittmann 2017). It is also called mission command. Many
wildland firefighting organizations have adopted the concept of Auftragstaktik and
called it “leader’s intent” (DeGrosky 2005; Ziegler and DeGrosky 2008). Therefore,
the theoretical SEM proposed in this chapter is identified as an Auftragstaktik
[AT] model. The theoretical AT model adopted in the study of wildland firefighting
teams (Gapud 2019) is applied as the benchmark in exploring the use of mindful
leadership within the Gulf Coast City Fire and Rescue organization. The findings
will provide an overview of the similarities and differences in Mindful Leadership
between the wildland and city firefighting teams.

1.2 The Simpliﬁed Version of the Model and Hypotheses

HCM facilitates the creation of a parsimonious model of the hypothesized relation-

ships, as shown in Fig. 1. The model is parsimonious, considering both Mindful
Leadership (Fig. 2) and Organizational Mindfulness are third-order higher-order
constructs (HOCs). Figures 3 and 4 show the full measurement model illustrating
the two third-order higher-order constructs (Mindful Leadership and Organizational
Mindfulness), second-order constructs (Shared Leadership and Leader Behavior),
and their corresponding lower-order constructs (Voice, Social Support, Shared
Purpose, Directive, Transactional, Transformational, Empowering).

1.3 The Hierarchical Component Model of AT

Figure 2 illustrates the collective mindset of wildland ﬁreﬁghting teams where

members are trusting and loyal to each other [Trust] (Podsakoff et al. 1990),
which operates mindfully [Organizational Mindfulness] (Weick and Sutcliffe
2007). Trust and Organizational mindfulness are the antecedent constructs of Mind-
ful Leadership, enabling Team Potency (Guzzo et al. 1993) and ultimately leading to
performance (Gapud 2019). The AT third order HOC Organizational Mindfulness is
Mindful Leadership Under Fire: A Validation Study of a. . . 247

Fig. 1 Simpliﬁed version of the theoretical model with hypothesized relationships. Note: Hypoth-
esized relationships of the constructs are illustrated. H1: Organizational mindfulness + ! Mindful
leadership. H2: Trust + ! Mindful leadership. H3: Mindful leadership + ! Team potency. H4:
Mindful leadership + !Team performance. H5: Team potency + !Team performance

Fig. 2 Hierarchical component model of Mindful leadership (Gapud 2019)

based on the theorizing of Weick and colleagues (Weick et al. 1999; Weick and
Sutcliffe 2007, 2008). Weick and Sutcliffe (2007) described an organization as
mindful when the teams are always prepared to deal with the unexpected [Antici-
pation] and can contain and bounce back from problem situations [Containment].
Resilient performance in the age of uncertainty (Weick and Sutcliffe 2007) relies on
the practice of anticipation as a strategy that enables an organization to be mindful,
which includes being preoccupied with failures [PreOcFail], reluctance to simplify
details [RelucSim], and sensitivity to operations [SensOper]. When preventing the
248 S. D. Gapud and J. F. Hair

Fig. 3 Measurement model

of Auftragstaktik (Wildland
Fireﬁghters)
Mindful Leadership Under Fire: A Validation Study of a. . . 249

Fig. 4 The measurement model of Auftragstaktik (City Fireﬁghters)

250 S. D. Gapud and J. F. Hair

disruptive “unexpected events” which continue to develop despite the anticipation

strategies that the organization has implemented, mindful organizational strategies
rely on two principles—the commitment to resilience [CommitResil] and deference
to expertise [DeferExpert]. The other third-order higher-order construct is Mindful
Leadership, which occurs when there is dynamic influencing within the team,
enabling any member to rise and take the leadership role when the situation calls
for their expertise. The dimensions of Mindful Leadership (Dunoon and Langer
2011; Gapud 2019; Langer and Moldoveanu 2000) are based on the theories of
Shared Leadership (Carson et al. 2007) and Leader Behavior (Manz and Pearce
2017; Pearce et al. 2003; Pearce 2010; Pearce and Conger 2003; Pearce et al. 2014).
Shared leadership is a team property whereby leadership is distributed among team
members rather than focused on one designated leader (Carson et al. 2007). An
organization can be infested with leadership disease when the power is concentrated
in a charismatic singular authority who can easily turn into a corrupt powerful
authoritarian (Manz and Pearce 2017). One cure for leadership disease is shared
leadership (Manz and Pearce 2017).
A reflective–reflective, repeated indicators approach was used to develop and
estimate the model’s HCMs (Organizational Mindfulness and Mindful Leadership)
(Becker et al. 2012; Hair et al. 2018; Sarstedt et al. 2019). The hierarchical compo-
nent model was analyzed using PLS-SEM (Wetzels et al. 2009) following the
paradigm set by Sarstedt et al. (2021) and applying the SmartPLS 4.0.6.8 software
(Ringle et al. 2022).

1.4 Mindful Leadership

Mindful Leadership is a shared leadership approach in which there are dynamic

inﬂuences in all directions within the chain of command. For example, mindful
Leadership was observed in high-reliability organizations (HRO) that safely and
successfully navigate uncertainty as part of their regular routine (Bigley and Roberts
2001; Gapud 2019; Langer 2010; Klein et al. 2006; Pearce et al. 2014; Weick and
Sutcliffe 2007). This approach is similar to Gapud and Hair (2021), who applied
PLS-SEM to assess the predictive characteristics of the Mindful Leadership HCM.
The Mindful Leadership HCM (Fig. 1) was developed using the repeated indi-
cators approach (Becker et al. 2012; Hair et al. 2018; Sarstedt et al. 2019) to estimate
hierarchical component model. Mindful Leadership is modeled as a third-order
construct with a higher level of abstraction than the second-order subcomponents
(Shared Leadership, and Leader Behavior), while simultaneously including several
lower-order constructs (LOCs) (Voice, Shared purpose, Social Support, Directive,
Transactional, Transformational & Empowering), which cover more concrete traits
of this construct. The Mindful Leadership HOC represents an individual’s percep-
tions of shared leadership in teams operating in a vertical/hierarchical organization
(Gapud 2019) as the logical theoretical cause explaining the correlations between the
Mindful Leadership Under Fire: A Validation Study of a. . . 251

leadership behaviors that are being shared (Pearce et al. 2003) and the shared
leadership environment (Carson et al. 2007).

2 Methodology

The model was evaluated using survey results from Alabama Forestry Commission
employees. The same measures were used to gather data from the Gulf Shores Fire
and Rescue Department, Gulf Shores, Alabama. The survey link was distributed to
Gulf Shores Fire and Rescue Department employees (60 employees), and 40 usable
responses were received. The survey involving the Alabama Forestry Commission
was sent to 230 employees, and the number of usable responses was 136. Data from
both organizations were collected using the Qualtrics platform. Attention question
items were used to provide a quality check.

3 Measurement Model Conﬁrmation

Results of the validation of the model of Auftragstaktik in wildland ﬁreﬁghting

teams are compared with that of the city firefighting teams. Kline (2015) noted a
difference between having a model that matches reality and trying to fit a model to
the data. If one organization is meant to be the expert in dealing with the unexpected
(Weick and Sutcliffe 2007), then organizational scholars will have a tool to use in
measuring the organizational mindfulness and the mindful leadership present in any
organization. The measurement model procedure of confirmatory composite analy-
sis (CCA) is followed to evaluate the constructs (Hair et al. 2020).

3.1 Indicator Loadings

The outer model was assessed by examining the relationship between the constructs
and their indicators. The ﬁnal measurement models in Figs. 3 and 4 show that all the
indicators (in yellow boxes) have loadings at or above 0.708 and loaded signiﬁcantly
in their respective constructs, with all p-values <0.001.

3.2 Internal Consistency Reliability

The results confirm all reflectively measured LOCs are reliable and valid (Table 1).
The benchmark model of Auftragstaktik [AT] generated from the wildland fire-
fighters exhibits sufficient levels of reliability, convergent validity, and discriminant
252

Table 1 Comparison of the validity and reliability statistics of City Fire team and Wildland Fire team
Wildland Fireﬁghting Teams City Fireﬁghting Teams
Cronbach’s Composite Average variance extracted Cronbach’s Composite Average variance extracted
Variables α reliability ρc (AVE) α reliability ρc (AVE)
Organizational 0.969 0.939 0.886 0.969 0.939 0.885
mindfulness
Anticipation 0.968 0.937 0.832 0.945 0.889 0.728
PreOcFail 0.753 0.860 0.670 0.679 0.822 0.608
RelucSim 0.929 0.940 0.670 0.855 0.896 0.635
SensOper 0.960 0.960 0.720 0.837 0.885 0.607
Containment 0.964 0.928 0.866 0.966 0.933 0.875
CommitResil 0.687 0.862 0.610 0.687 0.821 0.608
DeferExpert 0.880 0.909 0.625 0.815 0.879 0.647
Mindful leadership 0.957 0.915 0.844 0.850 0.707 0.546
Leader behavior 0.930 0.862 0.621 0.966 0.932 0.777
Directive 0.749 0.860 0.670 0.706 0.838 0.643
Empower 0.853 0.911 0.773 0.866 0.909 0.714
Transform 0.915 0.940 0.797 0.889 0.932 0.820
Transact 0.877 0.942 0.891
Shared leadership 0.975 0.949 0.861 0.917 0.832 0.625
Shared purpose 0.912 0.945 0.850 0.777 0.870 0.691
Social support 0.910 0.944 0.848 0.778 0.870 0.693
Voice 0.930 0.950 0.826 0.878 0.917 0.734
Team performance 0.946 0.961 0.861 0.835 0.901 0.753
Team potency 0.873 0.908 0.664 0.849 0.898 0.687
Trust 0.928 0.949 0.824 0.914 0.940 0.797
Notes: Higher-order construct values are denoted in bold and italics
S. D. Gapud and J. F. Hair
Mindful Leadership Under Fire: A Validation Study of a. . . 253

validity (Franke and Sarstedt 2019; Hair et al. 2022; Legate et al. 2022). Figures 3
and 4 show the measurement models.

3.3 Convergent Validity and Composite Reliability

Convergent validity was assessed based on average variance extracted (AVE) with
the lower limit of 0.50 (Hair et al. 2022). As shown in Table 1, all constructs
exceeded recommended convergent validity guidelines. The lower order constructs’
validity and reliability were evaluated based on established guidelines (Sarstedt et al.
2019). The AVE of PreOcFail, in City Fire, (0.4) was lower than 0.5 threshold, but
the composite reliability (0.81) is higher than 0.6, so the convergent validity of the
construct is minimally acceptable (Fornell and Larcker 1981). Furthermore, the
AVE, composite reliability, and Cronbach’s alpha were computed manually based
on Sarstedt et al. (2021). The AVEs for the second order HOC Leader Behavior with
its four lower-order components (LOCs) and the corresponding loadings were Direct
(0.646), Transact (0.539), Transform (0.962), and Empower (0.923). The third order
HOC Mindful Leadership in the Auftragstaktik model (wildland firefighter) has two
LOCs—Shared (0.959) and Leader Behavior (0.876). All reflective constructs, as
listed in Table 1, exceeded the minimum requirement of 0.70 for composite reliabil-
ity (Hair et al. 2019a).

3.4 Discriminant Validity

Discriminant Validity is the extent to which a construct is empirically distinct from

another construct in the structural model. In this chapter, heterotrait–monotrait
(HTMT) ratio item correlations are applied to assess and determine discriminant
validity. The HTMT (Henseler et al. 2015) is a measure of similarity between latent
variables (Hair et al. 2019b). If the HTMT is smaller than one, discriminant validity
can be established. A threshold of 0.85 reliably distinguishes between those pairs of
latent variables that are discriminant valid and those that are not.
The HTMT is a measure of similarity between latent variables. The lower
threshold of 0.9 (Henseler et al. 2015) is used to assess the HTMT values since
the constructs are conceptually similar. Theoretically, HOCs are related to their
LOCs, therefore it is expected that the HTMT ratios between these constructs will
be high. Moreover, it is also expected that the LOCs dimensions of a higher order
construct will also have high HTMT ratios. For example, the HTMT ratio between
CommitResil and DeferExpert is 0.97. However, minimal discriminant validity is
assumed since theoretically both are expected to be highly related since these two
constructs are the dimensions of the second-higher order HOC Containment.
Tables 2 and 3 illustrate the ratios (boldface) between the HOCs and their respective
LOCs.
Table 2 Discriminant validity–HTMT discriminant validity assessment using the HTMT criterion (Wildland Fireﬁghters)
254

Leader Mindful Organizational

Anticipation CommitResil Containment DeferExpert Direct Empower behavior leadership mindfulness PreOcFail RelucSim SensOper
Anticipation
CommitResil 0.86
Containment 0.89 1.04
DeferExpert 0.85 0.97 1.08
Direct 0.55 0.54 0.45 0.39
Empower 0.70 0.70 0.59 0.53 0.70
Leader behavior 0.74 0.71 0.64 0.58 0.76 1.04
Mindful 0.73 0.61 0.59 0.52 0.65 0.92 0.93
leadership
Organizational 1.03 0.91 0.96 0.92 0.53 0.68 0.72 0.72
mindfulness
PreOcFail 0.98 0.85 0.84 0.78 0.60 0.80 0.81 0.79 0.97
RelucSim 1.04 0.83 0.87 0.83 0.51 0.68 0.72 0.72 1.02 0.96
SensOper 0.98 0.89 0.88 0.84 0.57 0.72 0.74 0.76 0.98 0.93 0.90
Shared 0.67 0.51 0.51 0.44 0.54 0.74 0.75 1.00 0.65 0.71 0.65 0.70
leadership
Shared 0.57 0.43 0.41 0.35 0.60 0.75 0.74 0.95 0.55 0.63 0.56 0.62
_Purpose
Social_Support 0.64 0.54 0.50 0.43 0.51 0.71 0.73 0.97 0.63 0.69 0.64 0.68
Team 0.22 0.26 0.18 0.16 0.09 0.25 0.30 0.38 0.21 0.17 0.23 0.22
performance
Team potency 0.45 0.48 0.35 0.33 0.40 0.54 0.57 0.58 0.43 0.39 0.47 0.42
Transact 0.55 0.45 0.44 0.39 0.53 0.55 0.60 0.60 0.54 0.64 0.55 0.54
Transform 0.72 0.67 0.63 0.58 0.77 0.88 1.04 0.88 0.71 0.77 0.71 0.72
Trust 0.64 0.58 0.52 0.47 0.53 0.80 0.79 0.73 0.62 0.71 0.62 0.65
Voice 0.66 0.47 0.52 0.46 0.43 0.66 0.67 0.93 0.65 0.68 0.65 0.69
S. D. Gapud and J. F. Hair
Shared leadership Shared _Purpose Social_Support Team performance Team potency Transact Transform Trust Voice
Anticipation
CommitResil
Containment
DeferExpert
Direct
Empower
Leader behavior
Mindful leadership
Organizational mindfulness
PreOcFail
RelucSim
SensOper
Shared leadership
Shared _Purpose 0.98
Social_Support 1.01 0.93
Team performance 0.40 0.38 0.43
Team potency 0.54 0.54 0.57 0.76
Transact 0.54 0.53 0.54 0.16 0.24
Mindful Leadership Under Fire: A Validation Study of a. . .

Transform 0.71 0.69 0.71 0.32 0.56 0.59

Trust 0.62 0.57 0.58 0.28 0.41 0.51 0.73
Voice 0.98 0.79 0.87 0.34 0.44 0.47 0.64 0.61
Notes: The numbers in boldface are the HTMT correlation coefﬁcient greater than the 0.90 threshold (Henseler et al. 2015)
255
Table 3 Discriminant validity-HTMT discriminant validity assessment using the HTMT criterion (City Fire)
256

Leader Mindful Organizational Shared

Anticipation CommitResil Containment DeferExpert Directive Empower behavior leadership mindfulness PreOcFail RelucSim leadership
Anticipation
CommitResil 1.00
Containment 0.94 1.11
DeferExpert 0.92 1.02 1.16
Directive 0.43 0.33 0.33 0.35
Empower 0.55 0.44 0.53 0.53 0.92
Leader behavior 0.45 0.35 0.39 0.39 1.04 1.08
Mindful 0.83 0.80 0.73 0.74 0.80 0.88 0.86
leadership
Organizational 1.01 1.09 1.07 1.07 0.35 0.54 0.42 0.77
mindfulness
PreOcFail 1.07 0.98 0.98 0.95 0.50 0.57 0.46 0.85 0.99
RelucSim 0.98 0.84 0.86 0.85 0.54 0.61 0.50 0.84 0.86 1.02
Shared 0.84 0.83 0.72 0.74 0.29 0.33 0.30 0.92 0.78 0.83 0.84
leadership
SensOper 0.94 0.86 0.71 0.69 0.19 0.38 0.33 0.64 0.84 0.63 0.57 0.69
Shared 0.51 0.37 0.36 0.41 0.32 0.27 0.30 0.75 0.44 0.49 0.44 0.90
_Purpose
Social _Support 0.79 1.09 0.83 0.75 0.31 0.48 0.37 0.89 0.81 0.86 0.69 0.88
Team 0.56 0.54 0.44 0.45 0.54 0.47 0.50 0.61 0.48 0.42 0.40 0.48
performance
Team potency 0.63 0.53 0.40 0.39 0.19 0.20 0.20 0.39 0.50 0.47 0.31 0.42
Transact 0.43 0.33 0.41 0.40 0.74 0.85 0.81 0.75 0.41 0.47 0.46 0.29
Transform 0.35 0.25 0.30 0.33 0.92 0.99 1.05 0.82 0.33 0.39 0.39 0.25
Trust 0.41 0.41 0.43 0.46 0.55 0.56 0.59 0.75 0.42 0.40 0.44 0.50
Voice 0.76 0.71 0.62 0.65 0.18 0.22 0.21 0.74 0.67 0.70 0.83 1.01
S. D. Gapud and J. F. Hair
SensOper Shared _Purpose Social _Support Team performance Team potency Transact Transform Trust Voice
Anticipation
CommitResil
Containment
DeferExpert
Directive
Empower
Leader behavior
Mindful leadership
Organizational mindfulness
PreOcFail
RelucSim
Shared leadership
SensOper
Shared _Purpose 0.51
Social _Support 0.66 0.52
Team performance 0.59 0.63 0.49
Team potency 0.75 0.45 0.45 0.71
Transact 0.29 0.32 0.43 0.36 0.18
Transform 0.29 0.35 0.30 0.44 0.18 0.80
Mindful Leadership Under Fire: A Validation Study of a. . .

Trust 0.35 0.52 0.59 0.36 0.24 0.53 0.61

Voice 0.58 0.49 0.61 0.26 0.23 0.15 0.13 0.32
Notes: The numbers in boldface are the HTMT correlation coefﬁcient greater than the 0.90 threshold (Henseler et al. 2015)
257
258 S. D. Gapud and J. F. Hair

HTMT ratios shown in Table 2 indicate discriminant validity among Organiza-

tional Mindfulness, Mindful Leadership, Team Performance, Team Potency, and
Trust. Moreover, all LOCs exhibit discriminant validity from each other following
the 0.90 threshold (Hair et al. 2022). A comparison of the model results follows since
the benchmark (Wildland Fire) provides support for the reliability and validity of the
measurement models. The structural model will now be evaluated (Hair et al. 2020).

4 Assessment of the Structural Model

Assessment of the structural model examines the relationship between constructs as

well as the model’s predictive capabilities (Legate et al. 2022; Manley et al. 2021).
We analyzed the structural model using bootstrapping with 10,000 subsamples and
found that all structural model relationships are significant ( p < 0.05). To establish
that the model does not have any collinearity issues, we examine the variance
inflation factor (VIF). The occurrence of a VIF greater than 3.3 is an indication of
pathological collinearity and also that a model may be contaminated by common
method bias (Hair et al. 2019a). Results indicate that at the construct level, VIFs in
the model ranged from 1.0 to 1.5 (Table 4). The model can be considered free from
method bias since VIFs are below the conservative lower threshold of 3.0 (Hair et al.
2019b) (Table 5).
Figures 5 and 6 show the results of the bootstrapping analysis for each sample
data. The path coefficient numbers in the model are reported with the corresponding
p-values in the parenthesis. In both samples, (Figs. 5 and 6) all relationships between
constructs are statistically significant, except for the path between Mindful Leader-
ship and Performance (Hypothesis 4) in the wildland firefighting teams (Table 6).
Analysis of the bivariate correlation between Mindful Leadership and Team Perfor-
mance reveals there is a significant relationship between the two constructs. There-
fore, the relationship between them in the model is explained by Team Potency. City
firefighting teams produced a different result than that of wildland firefighting teams.
As seen in Fig. 6 and then again in Table 7, H4 is highly significant. Team
Potency is not a full mediator in the model mindset of city firefighting teams.
Leadership, specifically Mindful Leadership, has a highly significant impact in the
Team Performance. There are many ways to interpret this, but we are inclined to
believe that there is so much deference to the authority of those up in the chain of
command among the city fire personnel. City fire usually deals with smaller scales
(one or a few houses on fire) where the leader can easily position oneself to a location
that has the vantage point and still have the most information to make strategic
decisions. This is probably why the components of Mindful Leadership in City Fire
have more comparable loadings—Shared leadership (0.756) and Leader Behavior
(0.722). Since all team members are highly trained to have the knowledge, skills, and
abilities to have high self-efficacy, which as a team translates to team potency—the
chief still has more information regarding factors to adjust the strategy when
environmental factors change.
Table 4 Collinearity statistics-inner variance inflation factors (VIF) values (Wildland Fire)
Leader Mindful Organizational Shared
Anticipation CommitResil Containment DeferExpert Directive Empower behavior leadership mindfulness PreOcFail RelucSim leadership
Anticipation 1 1 1
CommitResil
Containment 1 1
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful 1
leadership
Organizational 1 1 1.5
mindfulness
PreOcFail
RelucSim
Shared
leadership
SensOper
Shared
Mindful Leadership Under Fire: A Validation Study of a. . .

_Purpose
Social _Support
Team
performance
Team potency
Transact
Transform
Trust 1.5
Voice
(continued)
259
Table 4 (continued)
260

SensOper Shared _Purpose Social _Support Team performance Team potency Transact Transform Trust Voice
Anticipation
CommitResil
Containment
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful leadership 1 1.4 1
Organizational mindfulness
PreOcFail
RelucSim
Shared leadership
SensOper 1 1 1
Shared _Purpose
Social _Support
Team performance
Team potency 1.4
Transact
Transform
Trust
Voice
Notes: “If all [factor level] VIFs resulting from a full collinearity test are equal to or lower than 3 (Hair et al. 2019b), the model can be considered free of bias”
S. D. Gapud and J. F. Hair
Table 5 Collinearity statistics-inner variance inﬂation factors (VIF) values (City Fire)
Leader Mindful Organizational Shared
Anticipation CommitResil Containment DeferExpert Directive Empower behavior leadership mindfulness PreOcFail RelucSim leadership
Anticipation 1 1
CommitResil
Containment 1 1
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful 1 1
leadership
Organizational 1 1 1.2
mindfulness
PreOcFail
RelucSim
Shared
leadership
SensOper
Shared
Mindful Leadership Under Fire: A Validation Study of a. . .

_Purpose
Social _Support
Team
performance
Team potency
Transact
Transform
Trust 1.2
Voice
(continued)
261
Table 5 (continued)
262

SensOper Shared _Purpose Social _Support Team performance Team potency Transact Transform Trust Voice
Anticipation 1
CommitResil
Containment
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful leadership 1.1 1
Organizational mindfulness
PreOcFail
RelucSim
Shared leadership 1 1 1
SensOper
Shared _Purpose
Social _Support
Team performance
Team potency 1.1
Transact
Transform
Trust
Voice
Notes: “If all [factor level] VIFs resulting from a full collinearity test are equal to or lower than 3 (Hair et al. 2019b), the model can be considered free of bias”
S. D. Gapud and J. F. Hair
Mindful Leadership Under Fire: A Validation Study of a. . . 263

Fig. 5 Path coefﬁcients and p-values for the structural model of Auftragstaktik (Wildland Fire)

Wildland firefighters, on the other hand, are trained to rely on the newness of the
information brought to the on-going team dynamics as they navigate to find solu-
tions to rapidly changing conditions with various fronts. This is because huge fires
cannot be assessed by one authoritative person alone, even with the help of the eye in
the sky (drones, helicopters, or airplanes). New information for factors in the
sensemaking of the whole team can originate from any member or crew stationed
in a different front. Each member can call for additional resources to be deployed as
they deem necessary and those up in the chain of command perform the support
function to enable each crew to successfully carry out their function. Therefore,
mindfulness in leadership is more shared as seen also in the loadings of the
components. The shared leadership in Wildland Fires loaded higher (0.959) than
the other component Leader Behavior (0.876).

4.1 In-Sample Predictive Power

R2 is the metric for assessing the model’s explanatory power (Legate et al. 2022)
which represents the variance explained in the endogenous constructs (Table 8).
Mindful Leadership in both organizations is predicted by the endogenous latent
variables in the model with a value of above the moderate (0.597) in the Wildland
Fire and close to the substantial threshold (0.706) in the City Fire. This indicates the
constructs preceding Mindful Leadership are providing the variances explaining the
HOC Mindful Leadership construct. Team Potency in City Fire exhibits very limited
predictability by the independent variables in the model. However, the Team
Potency in the teams of wildland ﬁreﬁghters is predicted moderately with an R2
close to the moderate level (0.472).
264

Fig. 6 Path coefﬁcients and p-values for the structural model of Auftragstaktik (City Fire)
S. D. Gapud and J. F. Hair
Mindful Leadership Under Fire: A Validation Study of a. . . 265

Table 6 Path coefﬁcients and results of hypothesis (Wildland Fire)

Original Sample Significance &
sample mean Hypothesis
(O) (M) 2.50% 97.50% decision
Organizational mindfulness 0.43 0.44 0.26 0.63 H1***
! Mindful leadership
Trust ! Mindful leadership 0.44 0.43 0.21 0.62 H2***
Mindful leadership !Team 0.53 0.53 0.40 0.66 H3***
potency
Mindful leadership ! -0.01 -0.01 -0.18 0.17 H4NS
Team performance
Team potency ! Team 0.70 0.70 0.54 0.83 H5***
performance
Anticipation ! PreOcFail 0.84 0.85 0.78 0.90 ***
Anticipation ! RelucSim 0.98 0.98 0.96 0.99 ***
Anticipation ! SensOper 0.91 0.91 0.88 0.94 ***
Containment 0.89 0.89 0.862 0.92 ***
!CommitResil
Containment 0.97 0.97 0.96 0.98 ***
!DeferExpert
Leader behavior !Direct 0.65 0.65 0.52 0.76 ***
Leader behavior 0.92 0.92 0.88 0.95 ***
!Empower
Leader behavior !Transact 0.54 0.54 0.34 0.70 ***
Leader behavior 0.96 0.96 0.95 0.98 ***
!Transform
Mindful leadership 0.88 0.88 0.81 0.92 ***
!Leader behavior
Mindful leadership 0.96 0.96 0.94 0.97 ***
!Shared leadership
Organizational mindfulness 0.99 0.99 0.98 0.99 ***
!Anticipation
Organizational mindfulness 0.89 0.892 0.84 0.93 ***
!Containment
Shared leadership 0.91 0.91 0.87 0.94 ***
!Shared_Purpose
Shared leadership 0.93 0.95 0.91 0.97 ***
!Social_Support
Shared leadership !Voice 0.92 0.92 0.88 0.95 ***
Notes: Critical t-values for a two-tailed test are 1.65 (significance level = 10%), 1.96 (significance
level = 5%). 2.58 (significance level = 1%***) The one-tailed significance of a directional
hypothesis at the 0.05 level is 0.98

4.2 Out-of-Sample Predictive Power

Out-of-sample predictive power of the model was assessed using PLSpredict

(Manley et al. 2021). The method develops k-fold cross-validated prediction errors
266 S. D. Gapud and J. F. Hair

Table 7 Path coefﬁcients and results of hypothesis (City Fire)

Significance &
Original Sample Hypothesis
sample (O) mean (M) 2.50% 97.50% decision
Organizational mindfulness 0.52 0.52 0.33 0.67 H1***
! Mindful leadership
Trust !Mindful leadership 0.49 0.50 0.28 0.68 H2***
Mindful leadership !Team 0.32 0.33 0.16 0.49 H3***
potency
Mindful leadership ! Team 0.30 0.38 0.18 0.57 H4***
performance
Team potency !Team 0.48 0.48 0.23 0.66 H5***
performance
Anticipation ! PreOcFail 0.87 0.88 0.83 0.91 ***
Anticipation !RelucSim 0.88 0.88 0.83 0.92 ***
Anticipation ! SensOper 0.80 0.80 0.66 0.91 ***
Containment ! 0.89 0.89 0.86 0.92 ***
CommitResil
Containment ! 0.98 0.98 0.96 0.98 ***
DeferExpert
Leader behavior!Directive 0.85 0.85 0.74 0.93 ***
Leader behavior ! 0.90 0.95 0.92 0.97 ***
Empower
Leader behavior ! Transact 0.76 0.77 0.66 0.88 ***
Leader behavior ! 0.95 0.94 0.91 0.97 ***
Transform
Mindful leadership 0.72 0.70 0.25 0.89 ***
!Leader behavior
Mindful leadership 0.76 0.76 0.52 0.92 ***
!Shared leadership
Organizational mindfulness 0.92 0.92 0.88 0.96 ***
! Anticipation
Organizational mindfulness 0.96 0.96 0.94 0.98 ***
! Containment
Shared leadership ! 0.72 0.72 0.58 0.83 ***
Shared_Purpose
Shared leadership ! 0.76 0.76 0.63 0.86 ***
Social_Support
Shared leadership ! Voice 0.89 0.89 0.82 0.93 ***
Notes: Critical t-values for a two-tailed test are 1.65 (significance level = 10%), 1.96 (significance
level = 5%). 2.58 (significance level = 1%***) The one-tailed significance of a directional
hypothesis at the 0.05 level is 0.98

to evaluate the root mean squared error (RMSE), the mean absolute error (MAE),
and the mean absolute percentage error (MAPE) to assess the predictive performance
of their PLS path model (Shmueli et al. 2019; Stone 1974). We used the default
10 repetitions. To assess the predictive power of the model, we have to compare the
Mindful Leadership Under Fire: A Validation Study of a. . . 267

Table 8 R2 values of endogenous latent variables in the path model

R2 adjusted
Wildland Fire City Fire
Mindful leadership 0.597 0.706
Team potency 0.472 0.091
Team performance 0.280 0.475
Notes: The R2 can be substantial (0.75), moderate (0.50) or weak (0.25) (Henseler et al. 2009; Hair
et al. 2011)

Table 9 Prediction summary of latent variables

Wildland Fire City Fire
Variables Q2predict RMSE MAE Q2predict RMSE MAE
Mindful leadership 0.578 0.663 0.486 0.662 0.590 0.469
Leader behavior 0.600 0.644 0.484 0.226 0.922 0.584
Directive 0.250 0.907 0.651 0.087 1.001 0.688
Empower 0.521 0.705 0.525 0.305 0.870 0.605
Transact 0.248 0.883 0.672 0.251 0.888 0.742
Transform 0.545 0.687 0.521 0.191 0.938 0.644
Shared leadership 0.824 0.425 0.320 0.441 0.771 0.612
Shared_Purpose 0.315 0.849 0.629 0.199 0.938 0.699
Social_Support 0.376 0.804 0.596 0.433 0.777 0.556
Voice 0.419 0.774 0.582 0.262 0.899 0.687
Team performance 0.047 0.995 0.735 0.168 0.940 0.721
Team potency 0.165 0.934 0.718 0.119 1.022 0.712
Anticipation 0.976 0.157 0.113 0.838 0.409 0.325
CommitResil 0.649 0.606 0.449 0.768 0.498 0.398
Containment 0.793 0.463 0.335 0.920 0.292 0.226
DeferExpert 0.718 0.539 0.397 0.874 0.364 0.290
PreOcFail 0.688 0.567 0.430 0.666 0.597 0.489
RelucSim 0.929 0.269 0.207 0.621 0.631 0.501
SensOper 0.435 0.767 0.564 0.518 0.716 0.520

results with two benchmarks. Table 9 shows all relevant variables have positive Q2
value for both populations being studied. Thus, the prediction error of the PLS-SEM
results is smaller than the LM prediction errors, indicating the Auftragstaktik PLS
model achieves better predictive performance. We now proceed to the second
benchmark.
We will focus on the Mindful Leadership variables since we are interested in
validating it as a third-order HOC. Moreover, included in Table 10 are the variables
for the two independent constructs Team Potency and Team Performance, to see if
the proposed Mindful Leadership construct is indeed predicting both. We examine
the results of PLS prediction using the data from our benchmark population to see if
the PLS-SEM outcomes have a lower prediction error than the simple linear regres-
sion model (LM) prediction outcomes.
268 S. D. Gapud and J. F. Hair

Table 10 PLS-SEM vs. linear regression model (LM) prediction results

PLS-SEM LM
Variables Q2predict RMSE MAE RMSE MAE
LBDir1 0.196 0.987 0.719 1.054 0.819
LBDir2 0.256 0.824 0.536 0.972 0.667
LBDir3 0.022 1.171 0.937 1.265 0.981
LBEM5 0.402 0.962 0.679 1.115 0.794
LBEM5 0.397 0.967 0.681 1.115 0.794
LBEM5 0.412 0.954 0.675 1.115 0.794
LBEmp1 0.368 1.052 0.813 1.156 0.900
LBEmp4 0.427 1.172 0.903 1.331 1.016
LBTranS1 0.214 1.238 0.967 1.345 1.075
LBTranS2 0.229 1.225 0.924 1.333 1.018
LBTransF1 0.426 1.071 0.798 1.214 0.904
LBTransF1 0.424 1.073 0.800 1.214 0.904
LBTransF1 0.429 1.069 0.797 1.214 0.904
LBTransF2 0.364 1.153 0.864 1.310 0.977
LBTransF2 0.368 1.150 0.859 1.310 0.977
LBTransF3 0.368 1.076 0.884 1.169 0.937
LBTransF3 0.369 1.076 0.884 1.169 0.937
LBTransF3 0.367 1.077 0.885 1.169 0.937
LBTransF4 0.564 0.952 0.746 1.073 0.818
LBTransF4 0.569 0.947 0.743 1.073 0.818
LBTransF4 0.557 0.960 0.749 1.073 0.818
SL10V 0.368 1.257 0.988 1.474 1.119
SL10V 0.369 1.256 0.989 1.474 1.119
SL10V 0.369 1.256 0.988 1.474 1.119
SL1Purp 0.274 1.159 0.895 1.356 1.027
SL1Purp 0.278 1.156 0.892 1.356 1.027
SL1Purp 0.275 1.158 0.894 1.356 1.027
SL2Purp 0.268 1.128 0.844 1.306 0.930
SL2Purp 0.271 1.126 0.842 1.306 0.930
SL2Purp 0.273 1.125 0.841 1.306 0.930
SL3Purp 0.224 1.228 0.931 1.397 1.067
SL3Purp 0.224 1.228 0.931 1.397 1.067
SL3Purp 0.227 1.226 0.929 1.397 1.067
SL4Soc 0.289 1.233 0.939 1.515 1.130
SL4Soc 0.291 1.231 0.938 1.515 1.130
SL4Soc 0.292 1.230 0.938 1.515 1.130
SL5Soc 0.329 1.156 0.867 1.337 1.007
SL5Soc 0.328 1.157 0.868 1.337 1.007
SL5Soc 0.329 1.156 0.867 1.337 1.007
SL6Soc 0.317 1.196 0.919 1.391 1.039
SL6Soc 0.318 1.195 0.919 1.391 1.039
(continued)
Mindful Leadership Under Fire: A Validation Study of a. . . 269

Table 10 (continued)
PLS-SEM LM
Variables Q2predict RMSE MAE RMSE MAE
SL6Soc 0.317 1.196 0.919 1.391 1.039
SL7V 0.377 1.173 0.904 1.414 1.080
SL7V 0.376 1.173 0.903 1.414 1.080
SL7V 0.378 1.171 0.907 1.414 1.080
SL8V 0.309 1.339 1.036 1.552 1.194
SL8V 0.310 1.339 1.032 1.552 1.194
SL8V 0.309 1.340 1.029 1.552 1.194
SL9V 0.293 1.332 0.960 1.524 1.094
SL9V 0.296 1.330 0.957 1.524 1.094
SL9V 0.296 1.329 0.952 1.524 1.094
TPerf1 0.039 0.719 0.525 0.834 0.635
TPerf2 0.029 0.721 0.522 0.828 0.622
TPerf3 0.035 0.717 0.525 0.807 0.622
TPerf4 0.057 0.756 0.552 0.877 0.661
TPot1 0.113 1.123 0.867 1.281 0.963
TPot2 0.084 0.734 0.553 0.836 0.648
TPot3 0.115 0.997 0.773 1.121 0.853
TPot4 0.107 0.901 0.685 0.999 0.780
TPot5 0.113 1.021 0.788 1.092 0.838
Notes: Some manifest variables appeared twice or thrice because we used the reﬂective–reﬂective
technique in developing the third-order higher-order construct Mindful Leadership

Table 10 shows that all the manifest variables measuring the components of
Mindful Leadership, Team Potency, and Team Performance in the PLS-SEM have
smaller root-mean-square error (RMSE) and mean absolute error (MAE) values as
compared to that of a linear regression model (LM).

4.3 Importance-Performance Map Analysis (IPMA)

Now that we have determined that the Auftragstaktik Model has predictive power,
we examine how the constructs developed using Wildﬁre data perform using the
City Fire data. To do so, we apply the Importance Performance Map Analysis to
obtain information on how to improve Mindful Leadership in each organization. At
the construct level, the Wildland Fire data indicates the Alabama Forestry Commis-
sion needs to focus on developing Trust if the organization leaders want to improve
mindfulness in how their teams operate. In contrast, the results of the City Fire data
are different, and indicate Gulf Shores City Fire and Rescue employees need to learn
more Organizational Mindfulness concepts (Figs. 7 and 8).
Examining the individual measured items (Figs. 9 and 10), however, both
organizations’ Mindful Leadership is highly impacted by Trust items. Trust indeed
270 S. D. Gapud and J. F. Hair

Fig. 7 Importance-performance map analysis for mindful leadership (Wildland Fire)

is the glue that binds leaders and members together. High on the importance scale for
both sample populations is Trust3 which states that “I have complete faith in the
integrity of my manager/supervisor.” DE3 and RSI4 variables would yield an
increase of about 60% in the Mindful Leadership of City firefighters. DE3 measure
states, “People in this organization value expertise and experience over hierarchical
rank.” RSI4, on the other hand states that “People in this organization are not shot
down for surfacing information that could interrupt operations.” Improvement in
these items requires some organizational change in terms of “the way things are done
in this organization.”
Culture of an organization is difficult to change. For Gulf Shores Fire and Rescue
to effect a change in how the employees perceive leadership, the organization must
implement a new communication process during operations. A new way or proce-
dure for team members to bring new ideas and be assured that they will be heard and
not shut down while the fire is in progress. Those on the frontlines have experience
and expertise (DE3) which should be recognized without regard for the position he
or she occupies within the chain of command.
Mindful Leadership Under Fire: A Validation Study of a. . . 271

Fig. 8 Importance-performance map analysis for mindful leadership (City Fire)

5 Discussion

This chapter presents a study that confirmed that the previously established second
order HOC Shared Leadership and second-order HOC Leader Behavior are the
hierarchical component dimensions of third-order HOC Mindful Leadership. The
study found that HOC Mindful Leadership can take on different bandwidths in an
organization, depending on the extent to which shared leadership (voice, social
support, shared purpose) and leadership behavior (directive, empowering, transac-
tional, transformational) are shared.
Comparing the results of the Alabama Forestry Commission (Wildland Fire) with
that of the Gulf Shores Fire and Rescue Department (City Fire) gave researchers a
glimpse of how a model performs differently in each organization. The results were
shared with Chief Sealy and his deputy. The deputy stated that since the organization
introduced shared leadership, there has been a change in how people see leaders’
culture. However, there are still some holdouts, and not all employees are ready to be
empowered. Some are still learning how to voice their opinion and participate in
strategy formulation when engaging in a mission. He suggested a similar survey be
conducted to determine if the leadership perceptions have changed.
The model of AT that we derived from the wildland firefighting teams has
predictive power and can be used to help managers detect areas to improve
272 S. D. Gapud and J. F. Hair

Fig. 9 Importance-performance map for mindful leadership (Wildland Fire)

Fig. 10 Importance-performance map for mindful leadership (City Fire)

Mindful Leadership Under Fire: A Validation Study of a. . . 273

mindfulness in an organization when combined with the IPMA. AT has many facets
and can detect the presence of organizational sensemaking and adaptive capability.
Moreover, the AT model draws its capability from previously well-theorized con-
structs. Mission command or Auftragstaktik [AT] is a competence-based view
strategy for a business operating in a volatile and uncertain environment (Nelsen
1987; Shamir 2010; Wittmann 2017). Furthermore, the AT model behaves differ-
ently in two hierarchical organizations mainly because of differences in organiza-
tional processes.
The major theoretical implication of this chapter is the validation of a Structural
Equation Model (SEM) that illustrates how a process leadership style varies in
different organizations. The study presented an interesting case by sampling two
distinct ﬁreﬁghting units that utilize situational awareness to create and implement
strategy and change tactics at a moment’s notice as the complexity analysis changes.
However, they have different approaches to leadership, which impact how they
perceive team and organizational performance. Importance performance map anal-
ysis clearly shows how Trust and Organizational Mindfulness differences impact
Mindful Leadership. Moreover, within an organization, differences in mindset and
trust may occur.

6 Conclusion

The study’s ﬁndings have important implications for practice, particularly for
hierarchical organizations wanting to be adaptive in times of uncertainty. While
such organizations should maintain their structure for stability, they should adopt a
communication strategy that allows recognition of ideas and expertise to ﬂow from
all directions. The AT model can help identify areas of training needed to improve
organizational outcomes. Combined with advanced analytical tools, it can aid
organizations and scholars in achieving this goal. Future studies should classify
employees based on various combinations of antecedents to cater to their training
needs. Finally, the study validates the theoretical SEM and illustrates how a process
leadership style varies in different organizations. Future research should continue to
expand the understanding and application of Mindful Leadership and the model
of AT.

References

Becker JM, Klein K, Wetzels M (2012) Hierarchical latent variable models in PLS-SEM: guidelines
for using reﬂective-formative type models. Long Range Plan 45(5–6):359–394
Bernstein E (2015) Leadership and teaming. September 2013 (Revised August 2015). Harvard
Business School
Bigley GA, Roberts KH (2001) The incident command system: high-reliability organizing for
complex and volatile task environments. Acad Manag J 44(6):1281–1299
274 S. D. Gapud and J. F. Hair

Branswell H (2022, Dec 27) Three years on, the pandemic — and our response — have been jolting.
Here’s what even the experts didn’t see coming. STAT. https://www.statnews.com/2022/12/27/
covid19-pandemic-what-most-surprised-experts/. Accessed 27 Dec 2022
Carson JB, Tesluk PE, Marrone JA (2007) Shared leadership in teams: an investigation of
antecedent conditions and performance. Acad Manag J 50(5):1217–1234
Cecere L (2021, Dec 1) The assumptions underlying the global supply chain are faltering in the face
of pandemic headwinds. Forbes Business. https://www.forbes.com/sites/loracecere/2021/12/01/
the-assumptions-underlying-the-global-supply-chain-are-faltering-in-the-face-of-pandemic-
headwinds/?sh=658ff953ebce Accessed 11 Mar 2022
DeGrosky MT (2005) Improving after-action review (AAR) practice. In: Butler BW, Alexander
ME (eds) Eight international wildland firefighter safety summit: 10 years later. The International
Association of Wildland Fire, April, Missoula, MT, pp 26–28
Dunoon D, Langer E (2011) Mindfulness and leadership: opening up to possibilities. Integral
Leadersh Rev 11(5):1–15
Eastwood B (2020, Jun 1) What happens to industry and employment after COVID-19? Ideas made
to matter. Digital Economy MIT Management Sloan School https://mitsloan.mit.edu/ideas-
made-to-matter/what-happens-to-industry-and-employment-after-covid-19 Accessed
10 Jun 2021
Edmondson AC (2012) Teaming: how organizations learn, innovate, and compete in the knowledge
economy. John Wiley & Sons, New York
Fornell C, Larcker DF (1981) Evaluating structural equation models with unobservable variables
and measurement error. J Mark Res 18(1):39–50
Fraher AL, Grint K, Branicki L (2016, August) Sustaining mindfulness in high-reliability organi-
zations: a multi-modal study of US navy SEALs. In: 76th Annual Meeting of the Academy of
Management, 5–9 August, Anaheim, United States
Franke G, Sarstedt M (2019) Heuristics versus statistics in discriminant validity testing: a compar-
ison of four procedures. Internet Res 29(3):430–447
Gapud SD (2019) Leadership on fire: mindfully shared. Doctoral Dissertation, University of South
Alabama
Gapud S, Hair JF (2021) Predicting the leadership mindset of wildland firefighters. J Strateg Int
Stud 15(1):32–51
Guzzo RA, Yost PR, Campbell RJ, Shea GP (1993) Potency in groups: articulating a construct. Br J
Soc Psychol 32(1):87–106
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Sarstedt M, Ringle CM, Gudergan SP (2018) Advanced issues in partial least squares
structural equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019a) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Sarstedt M, Ringle CM (2019b) Rethinking some of the rethinking of partial least squares.
Eur J Mark 53(4):566–584
Hair JF, Howard MC, Nitzl C (2020) Assessing measurement model quality in PLS-SEM using
confirmatory composite analysis. J Bus Res 109:101–110
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. In: Sinkovics RR, Ghauri PN (eds) New challenges to international
marketing (advances in international marketing), vol 20. Emerald Group Publishing Limited,
Bingley, pp 277–319
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Klein KJ, Ziegert JC, Knight AP, Xiao Y (2006) Dynamic delegation: shared, hierarchical, and
deindividualized leadership in extreme action teams. Adm Sci Q 51(4):590–621
Mindful Leadership Under Fire: A Validation Study of a. . . 275

Kline RB (2015) Principles and practice of structural equation modeling, 4th edn. The Guilford
press, New York
Langer E (2010) A call for mindful leadership. Harv Bus Rev 28:60
Langer EJ, Moldoveanu M (2000) Mindfulness research and the future. J Soc Issues 56(1):129–139
Legate AE, Hair JF, Chretien JL, Risher JJ (2022) PLS-SEM: prediction-oriented solutions for
HRD researchers. Hum Resour Dev Q. (in press)
Manley SC, Hair JF, Williams RI, McDowell WC (2021) Essential new PLS-SEM analysis
methods for your entrepreneurship analytical toolbox. Int Entrep Manag J 17(4):1805–1825
Manz CC, Pearce CL (2017) Twisted leadership: how to engage the full talents of everyone in your
organization. Maven House Press, Palmyra, VA
Nelsen JT (1987) Auftragstaktik: a case for decentralized battle. Parameters 17(1):21–34
Pearce CL (2010) Leading knowledge workers: beyond the era of command and control. In: Pearce
CL, Maciariello JA, Yamawaki H (eds) The Drucker difference: what the world's greatest
management thinker means to today’s business leaders. McGraw-Hill, New York, pp 36–45
Pearce CL, Conger JA (2003) Shared leadership: reframing the hows and whys of leadership. Sage,
Thousand Oaks, CA
Pearce CL, Sims HP Jr, Cox JF, Ball G, Schnell E, Smith KA, Trevino L (2003) Transactors,
transformers and beyond: a multi-method development of a theoretical typology of leadership. J
Manag Dev 22(4):273–307
Pearce CL, Manz CC, Sims HP (2014) Share, don't take the lead: leadership lessons from
21 vanguard organizations. Information Age Publishing, Charlotte, North Carolina
Podsakoff PM, MacKenzie SB, Moorman RH, Fetter R (1990) Transformational leader behaviors
and their effects on followers’ trust in leader, satisfaction, and organizational citizenship
behaviors. Leadersh Q 1(2):107–142
Raghavan A, Demircioglu MA, Orazgaliyev S (2021) COVID-19 and the new normal of organi-
zations and employees: an overview. Sustainability 13(21):11942
Ringle CM, Wende S, Becker J-M (2022) SmartPLS 4. SmartPLS GmbH, Oststeinbek. http://www.
smartpls.com
Sarstedt M, Hair JF, Cheah J-H, Becker J-M, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J 27(3):197–211
Sarstedt M, Ringle CM, Hair JF (2021) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg AE (eds) Handbook of market research. Springer Interna-
tional Publishing, Cham, pp 1–47
Shamir E (2010) The long and winding road: the US Army managerial approach to command and
the adoption of mission command (Auftragstaktik). J Strateg Stud 33(5):645–672
Shmueli G, Sarstedt M, Hair J, Cheah J, Tang H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Sneader K, Sternfels B (2020) From surviving to thriving: reimagining the pos-COVID-19 return.
McKinsey & Company. https://www.mckinsey.com/featured-insights/future-of-work/from-
surviving-to-thriving-reimagining-the-post-covid-19-return. Accessed 16 Jun 2021
Stone M (1974) Cross-Validatory choice and assessment of statistical predictions. J R Stat Soc
36(2):111–147
Tucker, H. (2020, May 3). Coronavirus bankruptcy tracker, these major companies are failing amid
the shutdown. Forbes https://www.forbes.com/sites/hanktucker/2020/05/03/coronavirus-bank
ruptcy-tracker-these-major-companies-are-failing-amid-the-shutdown/?sh=3410dcb73425
Accessed 6 Mar 2021
Weick KE, Sutcliffe KM (2007) Managing the unexpected: resilient performance in an age of
uncertainty, 2nd edn. Jossey-Bass, John Wiley & Sons, San Francisco
Weick KE, Sutcliffe KM (2008) Organizing for higher reliability: lessons learned from wildland
ﬁreﬁghters. Fire Manage Today 68(2):14–19
276 S. D. Gapud and J. F. Hair

Weick KE, Sutcliffe KM, Obstfeld D (1999) Organizing for high reliability: processes of collective
mindfulness. Res Organ Behav 1:81–123
Wetzels M, Odekerken-Schroder G, van Oppen C (2009) Using PLS path modeling for assessing
hierarchical construct models: guidelines and empirical illustration. MIS Q 33(1):177–195
Wittmann J (2017) Auftragstaktik for business organizations in volatile and uncertain environ-
ments: a competence-based view. IAFOR J Bus Manage 2(2):29–49
Ziegler JA, DeGrosky MT (2008) Managing the meaning of leadership: leadership as ‘communi-
cating intent’ in wildland ﬁreﬁghting. Leadership 4(3):271–297
When Navigating Uncertainty Lead
Mindfully

Stephanie Dygico Gapud and Joseph F. Hair Jr.

1 Introduction

When unexpected events occur, they can have a significant impact on the function-
ing of the organization and the well-being of its employees. Depending on the nature
of the event, it can cause disruptions to operations, financial losses, and damage to
the organization’s reputation. High reliability organizations like nuclear power
plants, wildland firefighters, air traffic control, naval aircraft carriers, and emergency
response agencies are trained to safely achieve their mission of dealing with the
unexpected as part of their normal work (Barker and Weick 1993; Rochlin et al.
1987; Weick 1987; Wildavsky 1988; Roberts 1990; Grabowski and Roberts 2011;
Roberts et al. 2005; LaPorte 1996). It sounds paradoxical, but we argue that a high
degree of uncertainty is necessary for an organization to effectively adapt to envi-
ronmental pressure and change (Sashkin 2018).
In this chapter, we will present a structural equation model to illustrate the
relationship between the constructs found when teaming (Edmondson 2012) and
how teams in high reliability organizations [HRO] (Bigley and Roberts 2001; Weick
and Roberts 1993; Weick and Sutcliffe 2001, 2007; LaPorte and Consolini 1991)
mindfully share the lead (Pearce et al. 2003) to improve their team potency (Guzzo
et al. 1993; Koopmann and Ancona 2015; Zaccaro et al. 2001) and their individual
leadership mindset as they achieve their perceived goals.

© Springer Nature Switzerland AG 2023 277

1.1 Background

We will focus our analysis on the Alabama Forestry Commission (AFC), a 98-year-
old organization that is smooth sailing; the duties and the expectations are set, there
are protocols to follow, and members are comfortable with each other as they go
about their work routine. However, when called to action, like when a wildfire is
reported to AFC, the tempo in the workplace changes, their training kicks in, routine
work is set aside, and team members communicate more often to create a shared
situational awareness of the complexity of the unexpected event. The initial com-
plexity analysis will determine the degree of the unknowns, which in turn determines
the necessary strategy and expertise required to put out the fire. Moreover, the
organization’s members form attack teams or crews to mitigate the event. The size
of the team and team membership is fluid.
Moreover, collaboration with other agencies, if necessary, is formed and forged
on the fly. Each team member is expected to voice their opinion and take the lead
when the situation calls for one’s knowledge or expertise. All team members bring
their expertise and are trained to innovate, lead, and support others, much like
members of a jazz band (Pavlovich 2003; Bernstein 2015).

1.2 Mindful Leadership Deﬁned

We deﬁne this dynamic inﬂuencing process in the members of a team navigating

actively changing internal and external environmental factors while achieving the
goal of safely, completing a shared purpose as Mindful Leadership. Mindful Lead-
ership requires individual mindfulness or being actively present in the moment.
Leading mindfully in times of uncertainty means studying or learning about the
situation as it unfolds in an organizational context (Bennis and Nanus 1985).
Moreover, it is the job of these leaders to turn an ambiguous situation into an
uncertain situation.
A situation is ambiguous when there are plenty of unknowns. A situation is
uncertain when the cause and effect of the unexpected event are known, and
therefore a solution that will meet the complexity and volatility of the situation
can possibly be forged. When leaders lead mindfully, they create a work environ-
ment that is open, transparent, and encourages collaboration, which can promote a
culture of mindfulness within the team. Furthermore, in today’s volatile, uncertain,
complex, and ambiguous (VUCA) world, one cannot afford to be on autopilot or be
in a state of mindlessness. Additionally, leading mindfully involves embracing
uncertainty and being comfortable with ambiguity (Langer 2000; Langer and
Moldoveanu 2000; Bennis and Nanus 1985). This means being able to make
decisions, take action in the face of uncertainty, and adapt to changing circumstances
and new information (Edmondson 2012).
When Navigating Uncertainty Lead Mindfully 279

1.3 Mindful Leadership is Sharing the Lead

When teaming (Edmondson 2012), the mindful leadership process is not concerned
about position and roles. All team members alternate, taking the lead and following
because there is no designated leader. Moreover, when there is an interactive,
ongoing mutual inﬂuencing process, everyone is expected to rise to the occasion
when one’s knowledge and ability are needed by the novel situation in an atmo-
sphere where there is social support because both the individual and the group
outcomes matter (Bernstein 2015; Pearce et al. 2022). In ﬂuid, cross-functional,
global, multilocation, and virtual teams, all members alternate between leading and
following, depending on whose expertise is being called for by the situation.
Therefore, sharing the lead can be one aspect of creating a mindful organization,
as it can promote a culture of collaboration, empowerment, and shared responsibility
(Pearce 2010; Pearce et al. 2003, 2014; Pearce and Manz 2005; Manz and Pearce
2017). During teaming, members focus on analyzing the complexity and eventually
accomplishing the mission. The team achieves its mission by not focusing on the
top-down hierarchical form of leadership, by distributing decision-making power
more evenly throughout the organization, and by encouraging multiple leaders or
teams to take ownership of different aspects of the organization. Having mindful
leadership can foster a sense of shared purpose and values among all organization
members and align the organization’s goals and actions with those values (Bernstein
2015; Langer 2010; Manz and Pearce 2017; Carson et al. 2007).

1.4 Organizational Mindfulness: The Mindset of Mindful

Leaders

Individual mindfulness “is a state of mind or mode of practice that permits the
questioning of expectations, knowledge, and adequacy of routines in complex and
not fully predictable social, technological, and physical settings” (Levinthal and
Rerup 2006; Vogus and Welbourne 2003; Weick and Sutcliffe 2007). Moreover, a
mindful leader takes on the mindset called organizational mindfulness (Weick and
Sutcliffe 2007). According to Langer (2000), being mindful is the simple act of
drawing distinctions which leads to greater sensitivity to context and perspectives
and ultimately to greater control of outcomes. Navigating a changing organizational
landscape riddled with uncertainties requires mindful organizing (Ray et al. 2011;
Vogus 2011) and taking on the managerial mindset of organizational mindfulness
(Weick and Sutcliffe 2007). Following the philosophizing on organizational mind-
fulness of Weick and his coauthors, Eastburn (2018) provided clear examples of
mindfulness and mindlessness among bank managers. His ﬁndings clearly show that
managers of community banking institutions, who led mindfully and actively paid
attention to the environmental signals that might have an impact on the
280 S. D. Gapud and J. F. Hair

organizational performance, were able to develop strategies to create small wins to

weather the financial crisis of 2007–2009.
Organizational mindlessness could lead to a disastrous outcome and is exposed
or observed when a consequential event occurs which was not expected. Several
statistics have shown that driving to and from home is where most accidents happen
(Harms et al. 2021). Mindless or autopilot behavior occurs when navigating in
familiar surroundings or is fixated on a routine (Langer 2000). Routine per se does
not render mindlessness in an organization, but rather how the actors apply the
routine in the context of the situation (Levinthal and Rerup 2006). If an organiza-
tional response to why something is required to be done is “that is the way we’ve
always done it” then maybe it is time to review the protocol and streamline the
process to calibrate mindfulness. Organizational mindlessness can result from inertia
due to the tendency to preserve the status quo. One example of this is not wanting to
embrace the change that the CEO has sensed and advocated—e.g., causing Kodak to
miss out on the new market created by the digitalization of image technology
(Gavetti et al. 2005). Mindlessness, while tongues of fire are nipping at one’s
boots, can cause errors and possibly fatality. To avoid organizational mindlessness,
one must be comfortable in making decisions and sharing knowledge in times of
ambiguity and uncertainty by using their mistakes as learning experiences by
encouraging others to join in the search for new ideas as they constantly enhance
their understanding of the developing unexpected by testing their ideas and theories
against those colleagues and outside experts (Bennis and Nanus 1985; Langer 2000;
Langer and Moldoveanu 2000; Weick and Sutcliffe 2007).
Organizational mindfulness as a leadership or managerial mindset requires an
assessment of the relevance of the decision-making protocols. Having organizational
mindfulness as a mindset means each protocol and organizational process should be
reviewed as to its applicability based on the new framework required by the ensuing
event (Weick 1995; Weick et al. 2005). It refers to the extent to which an organiza-
tion captures discriminatory detail about emerging threats and creates a capability to
swiftly act in response to these details (Weick et al. 1999; Weick and Sutcliffe 2001;
Weick and Sutcliffe 2007). The five principles of organizational mindfulness—
preoccupation with failure, reluctance to simplify, sensitivity to operations, commit-
ment to resilience, and deference to expertise—work together to enable this capa-
bility. The following definitions are taken from the philosophizing of Weick and his
coauthors (Weick et al. 1999; Weick and Sutcliffe 2001; Weick and Sutcliffe 2007)
on sensemaking and organizational mindfulness.
The Principle of Anticipation in organizational mindfulness refers to the ability of
an organization to prepare for and anticipate potential problems or threats before
they occur. Anticipation aims to prevent small, unexpected outcomes from worsen-
ing. This principle has three dimensions: preoccupation with failure, reluctance to
simplify information, and sensitivity to operation. According to Weick and his
coauthors, highly reliable organizations invest in resources and activities to develop
contingencies, imagining worst-case scenarios and detecting hazards early in their
development. These three principles involve paying attention to the present
When Navigating Uncertainty Lead Mindfully 281

organizational situation and learning from past failures in order to prevent them from
happening again in the future.
Preoccupation with failure (PreOcFail) means having organizational practices
that ensure ongoing attention to small signals and potential sources of failures and
preparing for them. This principle helps organizations to be able to anticipate and
respond to emerging threats.
Reluctance to simplify (RelucSim) means being wary of oversimplifying complex
situations and recognizing the nuances of different contexts. This principle helps
organizations to be able to capture discriminatory detail about emerging threats.
Sensitivity to operations (SensOper) means being aware of the current state of
affairs and being able to detect changes in that state. This principle helps organiza-
tions to be able to detect emerging threats.
The Principle of Containment is about managing and containing the impact of
problems or threats when they occur. This principle’s two dimensions are the
principle of commitment to resilience and deference to expertise, as both involve
utilizing the knowledge and expertise available to the team and the ability to bounce
back from and recover from unexpected events.
Commitment to resilience (ComtResil) means being able to bounce back from
adversity and adapt to changing circumstances. This principle helps organizations to
be able to act in response to emerging threats swiftly.
Deference to expertise (DefExper) means recognizing the expertise of others and
being open to learning from them. This principle helps organizations to be able to
capitalize on the expertise of others in order to respond to emerging threats.
Together, anticipation and containment form a crucial part of organizational
mindfulness, helping organizations to be proactive in their approach to risk man-
agement and to be able to handle unexpected challenges effectively. These principles
allow for the creation of a unique capability for an organization to swiftly act in
response to discriminatory details about emerging threats, which is the core of
organizational mindfulness. Current wildland firefighter literatures were published
using qualitative (Dether and Black 2006; Thomas et al. 2007; Weick and Sutcliffe
2007) and quantitative (Black and McBride 2013) data from federal employees in
the western part of the United States. According to Black and McBride (2013),
Weick used the wildland firefighting teams operating in the western part of the
United States as his muse in fine-tuning his theory on organizational mindfulness.
The measure of Organizational Mindfulness was used in the study of two other
organizations. Ray et al. (2011) explained mindful organizing in various levels of
hierarchy in business schools. Eastburn (2018) proved that organizational mindful-
ness as a mindset differentiated bank CEO’s performance during the financial crisis
in 2008. In this chapter, we will use the model to test whether organizational
mindfulness exists among wildland firefighters in the southeastern region of the
country. In a way, this is a validation study of the hierarchical construct of Organi-
zational Mindfulness.
Teams navigating uncertainty trust each other, take on an organizational mind-
fulness mindset, and lead each other mindfully by sharing the lead in an environment
where voice is encouraged. There is social support as the team pursues its shared
282 S. D. Gapud and J. F. Hair

purpose, which makes the team potent and therefore achieves excellent performance.
This is the ﬁrst time in literature to explicate and present quantitative analysis on the
relationships of constructs relevant to leadership in teaming, organizational mind-
fulness, HRO, team potency, and team performance.

2 Methodology

Mixed method was used to put in proper context the results of PLS-SEM analysis.
The survey respondents were employees of the AFC. The link to the survey was
distributed to all the employees via e-mail and the survey was accessed using the
Qualtrics platform. The number of usable responses was 136. Attention question
items were included to provide a quality check. Moreover, all the AFC’s County
Managers, Work Unit Managers, and Program Directors were interviewed via zoom
to provide insight and qualitative explanation to the results of the quantitative
analysis.
In this chapter, we used hierarchical component modeling (HCM). The HCM
facilitates broadening the scope of the study and simplifying the structural model
relationships (Becker et al. 2012; Hair et al. 2018; Wetzels et al. 2009). The model
used two third order higher-order constructs (HOC)—namely, Organizational Mind-
fulness and Mindful Leadership. Organizational Mindfulness as a hierarchical com-
ponent model was ﬁrst used by Eastburn (2018). Moreover, we proposed the model
of Mindful Leadership as a third-order HOC based on our understanding of theo-
rizing of Langer (2000), Pearce et al. (2003), and Carson et al. (2007). Figure 1
shows the expanded version of the model.

Fig. 1 The theoretical model with hypothesized relationships. Notes: Hypothesized relationships
of the constructs: H1: Organizational mindfulness + →Mindful leadership. H2: Trust + →Mindful
leadership. H3: Mindful leadership + →Team potency. H4: Mindful leadership + →Team perfor-
mance. H5: Team potency + →Team performance
When Navigating Uncertainty Lead Mindfully 283

2.1 Sample Characteristics

All ﬁeld employees at Alabama Forestry Commission [AFC] are trained to be

leaders and to rise to the occasion when their knowledge and skills are required by
the situation. The firefighter’s organization is composed of team members who are
trained to be decision-makers and to lead a crew (chainsaw crew, burn crew,
firefighting crew, etc.). Therefore, the HCM measurement approach was consistent
with Pearce and Conger (2003), Avolio et al. (1996), Ensley et al. (2006), and Pearce
and Sims (2002).

2.2 The Theoretical Model

The theoretical model of how to lead mindfully is depicted in Fig. 1, along with its
antecedents (trust and organizational mindfulness) and the outcome variable of
perceived team performance, with a potential mediated team potency relationship.
Based on relevant literature, we posit that high-reliability organizations (HROs) will
positively inﬂuence the ﬁrm’s level of team performance capability to identify new
information, which results in improved team performance.

2.2.1 Mindful Leadership

Shared leadership can be measured using various approaches (Wassenaar and Pearce
2017). Analyses were performed at the individual level to capture the individuals’
perceptions of shared leadership. In an organization operating in a high-risk envi-
ronment, where the size and boundary of teams change as dictated by the mission,
and mistakes can very well cause fatality, the sense of team achievement and level of
performance are present in each individual’s cognition.
The Mindful Leadership is presented in the model as a third-order HOC (Fig. 2)
which incorporates Leader Behavior (Pearce et al. 2003), and shared leadership team
environment (Carson et al. 2007) as its dimensions. Moreover, both these dimen-
sions are also HOC, each with their own set of lower-order constructs (LOC). The
lower order constructs [LOCs] represent the more concrete traits of the construct
(e.g., voice, empowerment, deference to expertise, etc.), thereby creating a more
parsimonious model (Becker et al. 2012; Hair et al. 2018; Ringle et al. 2012; Sarstedt
et al. 2019; Schaubroeck et al. 2011; Wetzels et al. 2009). The proposed model
builds on Mindful Leadership (Dunoon and Langer 2011), team-shared leadership,
and leadership behavior (directive, transactional, empowering, and transformational)
(Ensley et al. 2006; Pearce 2010; Pearce et al. 2003) that emanates from the team
members (Carson et al. 2007) within a hierarchical organization.
The following are sample questions to measure each lower order constructs:
[Voice] People in this team are encouraged to speak up to test assumptions about
284 S. D. Gapud and J. F. Hair

Fig. 2 Hierarchical component model of Mindful Leadership (Gapud 2019)

issues under discussion; [SocSup] The members of my team give encouragement to

team members who seem frustrated; [ShPurp] The members of my team devise
action plans and time schedules that allow for meeting our team’s goals (Pearce
et al. 2003). For the shared leadership dimensions, we used the measure forwarded
by Carson et al. (2007). The following are the lower order constructs and the
corresponding sample survey questions: [Direct] Our leader issues instructions
and commands; [Transact] Our leader dispenses contingent personal rewards;
[Transform] Our leader encourages us to challenge status quo; [Empower] We
are encouraged to participate in teamwork.

2.2.2 Organizational Mindfulness

Team Mindful Leadership is driven by Organizational Mindfulness (Weick and

Sutcliffe 2007) (Fig. 3). Likewise, Organizational Mindfulness in this study is
measured at three levels of abstraction. Organizational Mindfulness is a third order
construct which was theorized based on the principles of Anticipation and Contain-
ment (Langer 1989; Weick and Sutcliffe 2007). The Anticipation and Containment
constructs were previously utilized in a SmartPLS analysis of Organizational Mind-
fulness in banks as the higher second order constructs (Eastburn 2018). The LOCs of
Anticipation include: preoccupation with failure, resistance to simplify information,
and sensitivity to operation. Sample survey items included: [PreOcFail] PWF8—
People in this organization are rewarded if they spot problems, mistakes, errors, or
failures; [RelucSim] RSI8-When something unexpected happens at work, team
members are encouraged to conduct a complete analysis of the situation rather
When Navigating Uncertainty Lead Mindfully 285

Fig. 3 A hierarchical component model of Organizational Mindfulness

than advocate their view; [SensOper] S01—During an average day in our organi-
zation, people come into enough contact with each other to build a clear picture of
the current situation (Weick and Sutcliffe 2007; Ray et al. 2011; Eastburn 2018).
Containment has two LOCs: deference to expertise and commitment to resilience.
Sample survey items included: [ComtResil] CTR2—In this organization, resources
are continually devoted to training and retraining people in their areas of expertise;
[DeferExper] DTR3—People in this organization value expertise and experience
over hierarchical rank.

2.2.3 Trust and Loyalty Toward the Leaders

Trust is quantiﬁed using the scale that measures trust in and loyalty to the leader
(Podsakoff et al. 1990). Trust Items asked respondents to provide responses using a
7-point Likert scale statement such as: I feel quite conﬁdent that my leader will
always try to treat me fairly.

2.2.4 Team Potency

Team Potency is measured following the scale developed for group potency. The
scale is composed of ﬁve items with sample statements such as: My team has
conﬁdence in its abilities to perform at high levels (Guzzo et al. 1993).
286 S. D. Gapud and J. F. Hair

2.2.5 Performance

Team Performance in this study is a measure of perceived Team Performance. The

AFC is a government agency; as such, the expectations are laid down at the
beginning of each ﬁscal year. Both the employee and supervisor sign the document
detailing the duties and objectives to accomplish which are based on the workload
assigned to the work unit by the Regional Forester. Each region’s share of accom-
plishment is set by the State Forester. This study was conducted at the beginning of
the ﬁscal year, and therefore the team’s accomplishment and performance for the
previous year is being assessed at the same time. Following the rating used to assess
performance, the respondents were asked if they performed within, above, below, or
far-exceeding the standard. Sample item asked if: My team reached its goal.

2.3 Qualitative Data

Organizations navigating ambiguity and uncertainty that want to be resilient have

much to learn from wildland firefighters. The investigation featured in this chapter
dove deep into the team dynamics of a 95-year-old hierarchical organization: the
Alabama Forestry Commission [AFC] (Gapud 2019). We conducted one-on-one
Zoom interviews to collect qualitative data to help explain the context of the
measured constructs when an organization is teaming.
We prepared several questions; however, when asked regarding the individual’s
ability to influence on-site decision-making while doing fire-related activities, all our
interviewees recalled an incident when their system (the whole state) was overloaded
continuously for several months. In 2016, the State of Alabama, Georgia, and
Tennessee were hit with a long drought from May to November in that year, causing
huge wildfire occurrences in various cities (National Weather Service 2016). All the
wildland firefighting teams were working extra hours. New wildfires were being
called in while they were working on one. Teams from the southern portion of
Alabama were fighting fires with local crew in the north. Knowing the local terrain is
key to effective wildland firefighting. The strategy of fighting fire in the wild
includes plowing using a dozer or manually by raking, with the objective of
exposing the mineral soil to cut the continuity of fuel (i.e., forest-floor litter like
twigs, leaves, branches, grass, bushes, etc.). Local firefighters are aware of the
hazards found in their own neck of the woods. Mobile, Alabama firefighters are
aware that there are pipes buried in the forest that carry fuel or natural gas. North
Alabama is known for quarries; therefore, a steep hill might very well have a sudden
drop, which is dangerous when you are driving heavy equipment like a plow or a
dozer. Accurate situational awareness, therefore, should include continued commu-
nication with local crew, especially if new to the area. AFC owns two airplanes
which serve as the eye in the sky to warn ground crews of dangers in their area.
These airplanes, combined with drones, were very important in assessing the
When Navigating Uncertainty Lead Mindfully 287

intensity and the speed of the spread of fire, especially when the fire was travelling
through the canopy. Moreover, the fire was spotted a quarter mile from a flaming
front, igniting yet another fire. This is one of the rare moments in the history of AFC
that a helicopter was commissioned to drop buckets of water and fire-retardant
chemicals to quench forest fires to stop the spread to expensive vacation homes in
the mountainous forested area. Due to steep terrain, these areas were hard to protect
using either the tractor or by hand crews.
The situation AFC encountered illustrated how resiliently the teams functioned,
maneuvered, and delivered, even as their margins were narrowing when the system
was overwhelmed with fire (Gapud 2019). Teams were being forged on the fly as the
central command struggled to find fresh legs to help with the fight to put out fire with
many battle fronts. Huge fires were crossing state line and leadership was being
transferred from a firefighting team from one state to another. Some private forest-
products companies sent their employees to volunteer and help fight wildfires that
were spreading close to their forest property which border their production plant.
Each wildland firefighter was aware of the order which fire to prioritize—people,
structure, private properties, natural forest. This made sure not to let fire go near
subdivisions where there are lots of lives and private structure that could be
damaged. Image 1 shows the location of all the fires that were reported and put
out during the fire season of 2016. Normally, huge named fires occur in the southern
portion of the state. However, that season, because of a long period of drought, the
northern portion of the state caught many named fires. In Alabama, a fire gets a name
when the damage reaches more than 100 acres.
One country manager painfully recalled that he was unable to oblige to the call of
a town official to a fire because he was currently at another fire. The system was
overwhelmed, and the fire behavior everywhere was uncontrollable, reaching up to
the crowns of the trees and travelling fast. The Northeast Regional Forester recalled
that they were just triaging fires to determine which one they could control. He was
mindful of his priorities—to eliminate fire without endangering the lives of his fire
crew team, while protecting the general public, life, property, occupied places,
people’s homes, improvement structures, vehicles, and prioritizing forest plantations
over natural stands. His complexity analysis made him realize that the AFC did not
have the capability to safely fight a 50–60-foot fire, traveling at one point 30–60
mph, in concert with the wind velocity during that day. He immediately informed the
authorities, announcing that the fire was out of control and that areas with people
needed to be evacuated. He further commented that, “the rain luckily came when we
were in a coal-mine area and the fire was heading to the offices. That was our team’s
wildfire number seven for the day.”
The teams’ internal dynamics were conceptualized and specified to create a model
representing the relationships between the constructs. Organizational scholars and
decision-makers will improve their understanding of how hierarchy (Leavitt 2005)
and shared leadership in teams go hand-in-glove (Pearce et al. 2014; Pearce 2010).
Moreover, the relationship between self-leadership and vertical leaders appears to be
critical in enabling knowledge workers to concentrate on achieving their mission
(Pearce and Manz 2005).
288 S. D. Gapud and J. F. Hair

Image 1 Map of ﬁres. Source: National Weather Service (2016). Central Alabama Year in Review.
National Oceanic and Atmospheric Administration [NOAA] on April 5, 2019, from: https://www.
weather.gov/bmx/climo_2016review

3 SEM Analysis Results

3.1 Measurement Model Conﬁrmation

The reflective model was evaluated using the Confirmatory Composite Analysis
(CCA) following the guidelines of Hair et al. (2020). The outer model was assessed
by examining the relationship between the constructs and their indicators. The
results show all of the reflectively measured constructs’ measures are reliable and
valid (Table 2). Internal consistency reliability, convergent validity, and
When Navigating Uncertainty Lead Mindfully 289

Table 1 Abbreviations used in the theoretical and in the measurement model

Constructs Measured variables
abbreviation Deﬁnition Abbreviation Deﬁnition
DefExper Deference to expertise DTE6 Deference to expertise 6
Direct Directive LBDir3 Leadership behavior Directive1
Empower Empowering LBEmp5 Leadership behavior empowering
5
PreOcFail Preoccupation to failure LBTranS1 Leadership behavior transac-
tional 1
RelucSim Reluctance to simplify LBTransF4 Leadership behavior transforma-
information tional 4
SensOper Sensitivity to operation PF1 Preoccupation with failure 1
ShPurp Shared purpose RS2 Reluctance to simplify informa-
tion 2
SocSup Social support SL2Purp Shared leadership 2 shared
purpose
Transact Transactional SL5Soc Shared leadership 5 social
support
Transform Transformational SL9V Shared leadership 9 voice
Voice Voice SO5 Sensitivity to operation 5
TPerf2 Team performance 2
TPot5 Team potency 5
TRUST4 Trust 4

discriminant validity were assessed for the reflective constructs. The model exhibits
sufficient levels of reliability (>0.85), convergent validity, and discriminant validity
(Hair et al. 2022; Henseler et al. 2015; Franke and Sarstedt 2019). Convergent
validity was assessed initially through indicator reliability. The rule of thumb is
that standardized indicator outer loadings must be 0.708 or higher (Hair et al. 2014).
Table 1 lists the definition of the abbreviations of the constructs in the theoretical
model.

3.1.1 Loadings

Figure 4 shows the ﬁnal measurement model. Most of the retained items have outer
loadings above the threshold of 0.708 except for PWF1 (0.67) in the higher-order
Organizational Mindfulness construct. These items were retained to support content
validity (Hair et al. 2019; Hair et al. 2022), and all were very close to the minimum
threshold.
290 S. D. Gapud and J. F. Hair

Fig. 4 Measurement
models showing the path
coefﬁcients and R2 values
When Navigating Uncertainty Lead Mindfully 291

Table 2 Validity and reliability of construct scores

Cronbach’s Composite Average variance extracted
α reliability ρc (AVE)
Organizational 0.961 0.964 0.589
_Mindfulness
Anticipation 0.952 0.958 0.618
PreOcFail 0.753 0.858 0.669
SensOper 0.891 0.915 0.606
RelucSim 0.929 0.942 0.672
Containment 0.901 0.922 0.628
ComtResil 0.787 0.862 0.610
DefExper 0.880 0.909 0.625
Mindful_Leadership 0.960 0.964 0.627
Leader_Behavior 0.929 0.943 0.703
Direct 0.749 0.856 0.667
Empower 0.853 0.911 0.773
Transform 0.915 0.940 0.797
Transact 0.877 0.942 0.891
Shared_Leadership 0.957 0.963 0.722
Voice 0.930 0.950 0.826
ShPurp 0.912 0.945 0.850
SocSup 0.910 0.944 0.848
Team_Potency 0.873 0.908 0.664
Team performance 0.946 0.961 0.861
Trust 0.928 0.949 0.824
Notes: Higher-order construct values are in bold and italics. LOCs are listed below the second-order
HOC

3.1.2 Convergent Validity

Convergent validity was also assessed by examining whether average variance

extracted (AVE) was above the lower limit of 0.50 (Fornell and Larcker 1981;
Hair et al. 2022). Table 2 shows all items of all constructs in the model achieved
above average convergent validity. The lowest AVE was exhibited by Organiza-
tional Mindfulness (0.59), but was still well above the minimum 0.50 AVE threshold
(Hair et al. 2022). This indicates the set of variables for each construct is consistent
in what they are intended to measure (Hair et al. 2022).
Composite reliability [CR] was also examined. All the reﬂective constructs in the
model exceeded the minimum requirement of 0.70 for composite reliability.
292 S. D. Gapud and J. F. Hair

3.1.3 Discriminant Validity

The HTMT criterion analysis results (Table 3) indicate discriminant validity among
Organizational Mindfulness, Mindful Leadership, Team Performance, Team
Potency, and Trust. Discriminant validity exists between the second order HOCs
Anticipation and Containment and other reﬂectively measured constructs, including
Mindful Leadership, Team Performance, Team Potency, and Trust; and the LOCs
Voice, ShPurp (shared purpose), SocSup (social support), Direct (directive), Trans-
act (transactional), Transform, (Transformational), and Empower (Empowering).
Discriminant validity was also found between all the second order HOCs, including
Leader Behavior, Shared Leadership, Commitment to Resilience, Preoccupation
with Failure, Sensitivity to Operation, and Reluctance to Simplify; and other reﬂec-
tively measured constructs, including Team Performance, Team Potency, and Trust.
At the same time, however, discriminant validity is not present between Leader
Behavior, Shared Leadership, and their higher order construct (HOC) Mindful
Leadership. Similarly, discriminant validity is not present between the third order
HOC Organizational Mindfulness, the higher second order constructs Anticipation
and Containment, and all the LOCs in the hierarchical component model, including
Commitment to Resilience, Preoccupation to Failure, Sensitivity to Operation, and
Reluctance to Simplify. These results were expected, however, as the measurement
model of the LOCs associated with HOCs are theoretically related to each other
(Hair et al. 2022). Since the results provide support for the measure’s reliability and
validity, the structural model can now be evaluated.

3.2 Structural Model Assessment

The assessment of the structural model examines the relationship between constructs
as well as the model’s explanatory capabilities (Hair et al. 2022). The first step is to
establish that the model does not have any collinearity issues. To measure the
increase in regression coefficient due to collinearity, a variance inflation factor
[VIF] analysis was performed for each set of predictor constructs in the model.
Results show that all VIFs were above 0.20 and well below 5.0, which indicates that
the model does not exhibit collinearity problems (Hair et al. 2014). The presence of
common method variance [CMV] bias can also be determined by evaluating the
variance inflation factors [VIF] in the construct level. In a study using a Monte Carlo
simulation Kock (2015) illustrated how VIF is a valid measure of common
method bias.
Results indicate that at the construct level, VIFs in the model ranged from 1.0 to
1.53. Since VIFs are below the 3.3 threshold, the model can be considered free from
method bias (Kock 2015; Hair et al. 2020) (Fig. 5 and Table 4).
All hypothesized relationships (see Table 5 and Fig. 4) were accepted except for
the Mindful Leadership to Team Performance path. The antecedents Organizational
Table 3 Discriminant validity assessment using the heterotrait–monotrait (HTMT) criterion
Leader Mindful Organizational
Anticipation CommitResil Containment DeferExpert Direct Empower behavior leadership mindfulness PreOcFail RelucSim SensOper
Anticipation
CommitResil 0.86
Containment 0.89 1.04
DeferExpert 0.85 0.97 1.08
Direct 0.55 0.54 0.45 0.39
Empower 0.70 0.70 0.59 0.53 0.70
Leader behavior 0.74 0.71 0.64 0.58 0.76 1.04
Mindful 0.73 0.61 0.59 0.52 0.65 0.92 0.93
leadership
Organizational 1.03 0.91 0.96 0.92 0.53 0.68 0.72 0.72
mindfulness
PreOcFail 0.98 0.85 0.84 0.78 0.60 0.80 0.81 0.79 0.97
When Navigating Uncertainty Lead Mindfully

RelucSim 1.04 0.83 0.87 0.83 0.51 0.68 0.72 0.72 1.02 0.96
SensOper 0.98 0.89 0.88 0.84 0.57 0.72 0.74 0.76 0.98 0.93 0.90
Shared 0.67 0.51 0.51 0.44 0.54 0.74 0.75 1.00 0.65 0.71 0.65 0.70
leadership
Shared_Purpose 0.57 0.43 0.41 0.35 0.60 0.75 0.74 0.95 0.55 0.63 0.56 0.62
Social_Support 0.64 0.54 0.50 0.43 0.51 0.71 0.73 0.97 0.63 0.69 0.64 0.68
Team 0.22 0.26 0.18 0.16 0.09 0.25 0.30 0.38 0.21 0.17 0.23 0.22
performance
Team potency 0.45 0.48 0.35 0.33 0.40 0.54 0.57 0.58 0.43 0.39 0.47 0.42
Transact 0.55 0.45 0.44 0.39 0.53 0.55 0.60 0.60 0.54 0.64 0.55 0.54
Transform 0.72 0.67 0.63 0.58 0.77 0.88 1.04 0.88 0.71 0.77 0.71 0.72
Trust 0.64 0.58 0.52 0.47 0.53 0.80 0.79 0.73 0.62 0.71 0.62 0.65
Voice 0.66 0.47 0.52 0.46 0.43 0.66 0.67 0.93 0.65 0.68 0.65 0.69
(continued)
293
Table 3 (continued)
294

Shared leadership Shared _Purpose Social_Support Team performance Team potency Transact Transform Trust Voice
Anticipation
CommitResil
Containment
DeferExpert
Direct
Empower
Leader behavior
Mindful leadership
Organizational mindfulness
PreOcFail
RelucSim
SensOper
Shared leadership
Shared_Purpose 0.98
Social_Support 1.01 0.93
Team performance 0.40 0.38 0.43
Team potency 0.54 0.54 0.57 0.76
Transact 0.54 0.53 0.54 0.16 0.24
Transform 0.71 0.69 0.71 0.32 0.56 0.59
Trust 0.62 0.57 0.58 0.28 0.41 0.51 0.73
Voice 0.98 0.79 0.87 0.34 0.44 0.47 0.64 0.61
Notes: *Numbers in bold are HTMT correlation coefﬁcients greater than the 0.90 threshold
S. D. Gapud and J. F. Hair
When Navigating Uncertainty Lead Mindfully 295

Fig. 5 Path coefficients and p-values for the structural model. Notes: The figure shows the results
of the bootstrapping analysis. Numbers in parentheses are the corresponding p-values of the path
coefficients. All relationships between constructs are statistically significant except for the path
between Mindful leadership and Team performance which has a p-value above 0.05

Mindfulness and Trust were positively related to Mindful Leadership. The relation-
ship with Mindful Leadership and Team Potency is highly significant (1% level).
The relationship between Mindful Leadership and Team Performance in the model
was not positive and not statistically significant, which led us to conclude that Team
Potency fully mediates the relationship between Mindful Leadership and the per-
ceived performance. Mediation occurs when a third variable intervenes or governs
the relationship between two other related constructs (Hair et al. 2017). This is
consistent with results of previous studies which have shown that Team Potency is
the explanatory variable between leadership and performance, or between trust and
performance (Bligh et al. 2006; Monteiro and Vieira 2016; Schaubroeck et al. 2011;
Sivasubramaniam et al. 2002).
To test the bivariate relationship between Mindful Leadership and Team Perfor-
mance we conducted mediation post hoc analysis. Post hoc analysis consists of
examining the data after the experiment has concluded, looking specifically for
patterns that have not been specified a priori. Keeping in mind that Mindful
Leadership is the higher order construct composed of Shared Leadership and Leader
Behavior as its dimensions, it is important to determine the bivariate correlation
between Mindful Leadership and Team Performance. Doing so will determine
whether the path correlation coefficient result is due mainly to the multivariate
relationship in the model.
The latent variable scores from the PLS-SEM analysis were used in the compu-
tation of the correlation between the three dependent variables. Bivariate correlation
analysis will isolate the impact of other variables in the model on the correlation
Table 4 Collinearity statistics—inner variance inflation factors (VIF) values
296

Leader Mindful Organizational Shared

Anticipation CommitResil Containment DeferExpert Directive Empower behavior leadership mindfulness PreOcFail RelucSim leadership
Anticipation 1 1 1
CommitResil
Containment 1 1
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful 1
leadership
Organizational 1 1 1.5
mindfulness
PreOcFail
RelucSim
Shared leadership
SensOper
Shared_Purpose
Social_Support
Team
performance
Team potency
Transact
Transform
Trust 1.5
Voice
S. D. Gapud and J. F. Hair
SensOper Shared _Purpose Social _Support Team performance Team potency Transact Transform Trust Voice
Anticipation
CommitResil
Containment
DeferExpert
Directive
Empower
Leader behavior 1 1
Mindful leadership 1 1.4 1
Organizational mindfulness
PreOcFail
RelucSim
Shared leadership
SensOper 1 1 1
Shared_Purpose
Social_Support
When Navigating Uncertainty Lead Mindfully

Team performance
Team potency 1.4
Transact
Transform
Trust
Voice
297
298

Table 5 Path coefﬁcients and results of hypotheses

Original Sample Standard
sample mean deviation t-statistics Hypothesis decision/
(O) (M) (STDEV) (|O/STDEV|) 2.50% 97.50% Statistical significance
Direct effects
Organizational mindfulness→Mindful 0.43 0.44 0.10 4.41 0.253 0.628 H1***
leadership
Trust→Mindful leadership 0.44 0.44 0.11 4.08 0.202 0.624 H2***
Mindful leadership→Team potency 0.53 0.53 0.07 8.07 0.384 0.647 H3***
Mindful leadership→Team performance 0 -0.01 0.09 0.05 -0.180 0.181 H4 NS
Team potency→Team performance 0.70 0.70 0.07 9.64 0.556 0.845 H5***
Indirect effects
Organizational mindfulness→Mindful 0 0 0.04 0.05NS -0.085 0.074 NS
leadership→Team performance
Trust→Mindful leadership→Team performance 0 0 0.04 0.05NS -0.086 0.088 NS
Organizational mindfulness→Mindful 0.16 0.17 0.05 3.07*** 0.079 0.285 ***
leadership→Team potency→Team performance
Mindful leadership→Team potency→Team 0.37 0.38 0.07 5.53*** 0.242 0.507 ***
performance
Trust→Mindful leadership→Team 0.16 0.16 0.04 3.66*** 0.084 0.264 ***
potency→Team performance
Notes: Critical t-values for a two-tailed test are 1.65 (significance level = 10%*), 1.96 (significance level = 5%**), 2.58 (significance level = 1%***). The
one-tailed significance of a directional hypothesis at the 0.05 level is 0.98
S. D. Gapud and J. F. Hair
When Navigating Uncertainty Lead Mindfully 299

between Mindful Leadership and Team Performance. Results of the bivariate corre-
lation show that Mindful Leadership is signiﬁcantly and positively correlated with
Team Performance. This suggests that in the multivariate analysis, the relationship
between Mindful Leadership and Team Performance is the effect of other variables
in the model—in this situation, the relationship is fully mediated by Team Potency.

3.3 In-Sample Prediction Results

To assess the model’s in-sample fit, we first consider the R2 (Table 6). The coeffi-
cient of determination R2, is a measure of in-sample prediction of all endogenous
constructs (Shmueli et al. 2019). We find that all the endogenous constructs have
high R2 values way above the large (0.2592) predictive accuracy threshold for the
field of psychology (Cohen 1992). R2 results confirm the ability of the model to
explain the endogenous construct using in-sample data (the Alabama Forestry
Commission fire-fighting teams) which should not be used to infer to the population
(results for firefighting teams from other organizations) (Rigdon 2012; Sarstedt et al.
2014; Shmueli et al. 2019).
The effect size ( f2) is the second measure of the predictive validity. It is an
estimate of the predictive ability of each independent construct in the model. The
effect size is also considered an in-sample predictive metric. For more detailed
information on predictive validity and its derivations and application, please see
“Assessing measurement model quality in PLS-SEM using confirmatory composite
analysis” (Hair et al. 2020). As seen in Table 7, in-sample effect sizes of the
constructs are close to large (e.g., the effect of Organizational Mindfulness, 0.31

Table 6 R2 values of endogenous latent variables in the path model

R2 adjusted Psychology R2 rules of thumb Marketing R2 rules of thumb
Mindful 0.60 High Moderately high
leadership
Team performance 0.47 High Moderate
Team potency 0.28 High Above the weak threshold
Notes: Cohen (1992) Classiﬁed the strength of the predictive accuracy (R2 values) as either small
(0.0196), medium (0.1304) and large (0.2592)

Table 7 Within-sample effect size: f 2 values of constructs

Mindful leadership Team performance Team potency
Mindful leadership 0.00 0.40
Organizational mindfulness 0.31
Team potency 0.67
Trust 0.32
Notes: Rules of thumb for assessing f 2 values follows Cohen (1992)— small (0.02), medium (0.15),
and large (0.35)
300 S. D. Gapud and J. F. Hair

and Trust, 0.32 on Mindful Leadership) and above the large threshold (e.g., the effect
of Mindful leadership, 0.40 on Team Potency, and the effect of Team Potency, 0.67
on Team Performance).

3.4 Out-of-Sample Predictive Power

To analyze the out-of-sample predictive power of the model, we use the PLS-predict
function of the SmartPLS 4.0 software (Ringle et al. 2022) developed by Shmueli
et al. (2016). The method uses training and holdout samples to generate and evaluate
predictions from PLS path model estimations. Moreover, Hair et al. (2022) provides
an overview of how the SmartPLS implementation of the PLS predict algorithm
enables researchers to obtain cross-validated prediction errors and prediction error
summary statistics (Shmueli et al. 2019). The statistics include the root mean squared
error [RMSE], the mean absolute error [MAE], and the mean absolute percentage
error [MAPE] (Shmueli et al. 2019). The results assess the predictive performance of
the PLS path model for the manifest variables [MV or indicators] and the latent
variables [LV or constructs]. Following the rules of thumb for running PLSpredict
(Shmueli et al. 2019) we used 10 folds (i.e., k = 10), and 10 repetitions (r = 10).
Results show that all the manifest variables in the model have positive Q2predict. To
assess the results of a PLS path model, its predictive performance is compared
against two benchmarks: the Q2predict value, and the LM [linear model] approach.
The Q2predict value compares the prediction errors of the PLS path model to the
errors when using mean values. If the Q2predict value is positive, the PLS-SEM model
has better predictive performance than the mean values. The LM approach regresses
all exogenous indicator variables on each endogenous indicator variable to generate
predictions. The PLS-SEM results should produce lower prediction errors for RMSE
and MAE than LM. Note that the LM prediction error results as reported by the
PLS-SEM software is only available for the indicator variables, and not for the latent
variables. The PLS-Predict analysis speciﬁed Mindful leadership as the key target
construct. PLS-LV residual analysis results shows a near-symmetrical, bimodal
graph. Therefore, we use RMSE as the criterion to compare the predictive perfor-
mance of alternative PLS path models (Shmueli et al. 2019). It is interesting to note
that in the PLS Model (see Table 8), the manifest variables belonging to Team
Potency and Performance has positive Q2predict, while in the LM, all of the men-
tioned MVs have negative Q2predict values. This indicates Team Potency and Team
Performance are also predicted by the model. Following the rule of thumb in
analyzing PLS-Predict (Shmueli et al. 2019), since the RMSE of PLS-SEM < LM
in a majority of the manifest variables, we can say that the model has moderate out-
of-sample predictive capability (Manley et al. 2021). Furthermore, Table 8 shows the
Q2predict values of the antecedent and outcome constructs of Mindful Leadership are
all positive. Therefore, the model can be used to predict similar relationships of
constructs to exist in other hierarchical organizations where leadership is mindfully
shared (Fig. 6).
When Navigating Uncertainty Lead Mindfully 301

Table 8 Prediction summary RMSE MAE Q2predict

of the lower order antecedent
Anticipation 0.359 0.279 0.902
constructs and the dimension
constructs of Mindful ComtResil 0.531 0.398 0.646
Leadership Containment 0.488 0.374 0.770
DefExper 0.525 0.401 0.710
Direct 0.451 0.322 0.185
Empower 0.486 0.358 0.543
Leader behavior 0.484 0.362 0.619
Mindful leadership 0.593 0.437 0.577
PreOcFail 0.533 0.407 0.683
RelucSim 0.393 0.309 0.874
SensOper 0.478 0.363 0.786
ShPurp 0.674 0.504 0.299
Shared leadership 0.651 0.479 0.436
SocSup 0.640 0.470 0.366
Transact 0.359 0.274 0.205
Transform 0.489 0.367 0.567
Voice 0.592 0.436 0.416

Fig. 6 PLS LV prediction residuals of Mindful Leadership

Figure 7 shows that the PLS is underpredicting the MV LBEmp4 (Our leader
engages us in participative goal setting) as evident by the skewedness of the
302 S. D. Gapud and J. F. Hair

Fig. 7 Side-by-side comparison of the manifest variable LBEmp4 prediction residuals generated in
PLS and LM Analysis
When Navigating Uncertainty Lead Mindfully 303

residuals to the right (Shmueli et al. 2019). Furthermore, the RMSE of LBEmp4 is
smaller in the PLS model (1.17) as compared to the LM (1.30).
Three metrics (R2 and f 2 effect size of endogenous variables [in-sample predic-
tion], and PLSpredict [out-of-sample prediction]) were applied to assess the predic-
tive validity of the model. Clearly, the results are a concrete step toward
understanding the leadership behavior of teams mindfully sharing the lead in the
context of HRO when they are teaming.

3.5 Importance Performance Map Analysis

The Importance-Performance Map Analysis [IPMA] at the construct and at the

indicator levels was conducted to extend the PLS-SEM results reporting. At the
construct level, the IPMA allows the identiﬁcation of the predecessor construct that
has a relatively high importance (strong total effect), but also low performance (low
average latent score), so that improvements can be implemented. For example, in our
model, we propose that Team Potency explains the relationship between Mindful
Leadership and performance (Hair et al. 2018). Conducting the IPMA by targeting
Team Potency, we can determine which among the predecessor constructs (Organi-
zational Mindfulness, Trust, and Mindful Leadership) in the model will have to be
focused on in training so we can improve the Team Potency (Fig. 8).

Fig. 8 The Importance-performance map analysis of the constructs

304 S. D. Gapud and J. F. Hair

Fig. 9 The Importance-performance map analysis of the indicators

Figure 9 shows that at the indicator level, improving Trust2 and Trust3 will also
have impact on improving Team potency. Moreover, as seen in Fig. 8, Mindful
Leadership has the strongest impact (highest total effect, 0.30–0.78) on Team
Potency. Moreover, it has low performance (25–42 latent variable score). This
connotes that any managerial action to improve Mindful Leadership will have high
impact on improving the Team Potency.

4 Discussion

4.1 Summary of Findings

This study began with the assumption that organizations navigating ambiguity and
uncertainty and want to be resilient have much to learn from wildland firefighters.
Our findings demonstrated how resiliently the teams functioned, maneuvered, and
delivered, even as their margins were narrowing when the system was overwhelmed
with fire. Our approach assessed a cross-section of the teams’ internal dynamics to
develop a model and test the construct relationships. We believe decision-makers
reading this manuscript will have a meaningful example of how hierarchy (Leavitt
2005) and shared leadership in teams go hand-in-glove (Pearce et al. 2014; Pearce
When Navigating Uncertainty Lead Mindfully 305

2010). Moreover, our research has exemplified and clarified that the self-leadership
of the vertical leader is critical in enabling a team of knowledge workers to focus on
achieving their mission (Pearce and Manz 2005). Moreover, Mindful Leadership, as
an emerging higher-order construct proposed here, further explains leadership
behavior typically emanating from team members. Furthermore, we proposed an
expansion of a theory on teaming, suggesting that teaming can occur when a non-
high-reliability organization (nonHRO) or regular teams within organizations face a
challenge or unexpected situation.
All hypothesized relationships were accepted except for the path between Mind-
ful Leadership and Team Performance. In this study, as antecedents, Organizational
Mindfulness and Trust are positively related to Mindful Leadership. The relationship
between Mindful Leadership and Team Performance in the model is not positive and
not statistically significant, which led us to conclude that Team Potency fully
mediates the relationship between Mindful Leadership and the perceived perfor-
mance. The qualitative portion of the study provided the examples of team dynamics
exemplifying the relationship of the constructs related to HROs. Previous study on
HRO focused on naval air traffic control teams exercising mission command, or
Auftragstaktik [AT]. Rochlin et al. (1987) illustrated how the characteristics of HRO
and AT overlap and intersect. The model in this study encapsulates the structure of
constructs related to AT and HRO at the same time.
The hierarchical component model, Mindful Leadership, proposed in this study is
flexible and adaptive because it is able to capture differences in the leadership
behavior exhibited by three clusters of people, based on how they perceived their
team performance. This study of Mindful Leadership is a concrete step toward the
understanding of the leadership behavior in teams mindfully sharing the lead in the
context of HRO. As Bennis (2012, p. 544) stated in his crucibles of authentic
leadership, understanding leaders is important because “leaders wield power.”
Only when we are mindful that leadership emanates from all members of the team
do we begin to realize how to empower them and make the team potent, adaptive,
and agile at any time.
To claim that the final retained SEM model in this study has attained “statistical
beauty, it has to (1) make sense; (2) differentiate between known and unknown; and
(3) set conditions for posing new questions” (Kline 2015, p. 22).

4.2 The Model Makes Sense

The model derived in this study is based on previous studies on prominent leadership
(Pearce and colleagues) and organization scholars (Weick and colleagues), com-
bined with the ﬁrsthand data and personal experience of this author in working with
the organization in focus (Alabama Forestry Commission). Kline (2015, p. 68)
quoted Bollen’s (1989) argument that “if a model is consistent with reality, then
the set of data is consistent with the model. But if the set of data is consistent with the
model, this does not imply that the model corresponds to reality.” Therefore, the
306 S. D. Gapud and J. F. Hair

model and clustering procedure should be tested in other HROs to determine if

similar clusters of people coexist in the population, and if the model has similar
third-order abstraction results, given other sets of data. Moreover, it is prudent to
suggest that the model and the series of analysis (clustering and IPMA) be tested
using other types of organizations, possibly other high reliability organizations, or
any branch of the military.
Mindful Leadership is a process leadership style, therefore it is important to note
that just like the mindset of organizational mindfulness is not a static state. Like
individual mindfulness, constant updates of situational awareness is necessary.
Moreover, environmental scanning results can differ at the individual and organiza-
tional level. We mentioned that leaders rise in a team based on their knowledge,
skills, and abilities. The strategy of small wins in the development of sensitivity to
operation encourage team members to speak up (Weick and Sutcliffe 2007). As the
authors further explain, just because you see it, don’t assume that someone else sees
it too.

5 Conclusions

Mindful leadership is an emerging construct on leadership which views leadership

not as a position of authority or a hierarchical role, but as an unfolding social process.
It is an individual’s perspective of shared leadership of team members working in a
high-risk and a rapidly changing context environment. Mindful leadership, when
present in an organization, could pave the way to the unique strategy that the
company will derive from being mindful of the present and the organization’s
context in history. It is the process leadership that occurs when teaming. Moreover,
when leadership is shared and authority is allowed to migrate to those organizational
members in the frontlines, the vertical leaders could be freed of the minutiae of
operations to have more time and energy to pay attention to the bigger picture and
make sound strategic decisions (Klein et al. 2006; Klein 1998). Therefore, adopting
“mindful leadership” that incorporates a higher level of collective mindfulness and
shared leadership constructs can positively inﬂuence a ﬁrm’s performance and
reliability to manage the unexpected. In a sense, the organization would be smarter
and better able to adapt to their environment and manage challenges.

6 Implications for Theory and Practice

6.1 Implications to Other Hierarchical Organizations

Other hierarchical organizations wanting to be agile do not need to dissolve their

organizational structure. As illustrated in this study, the hierarchy provides stability
to the organization while the adapting, sensing, and growth occurred in the areas of
When Navigating Uncertainty Lead Mindfully 307

the organization actively engaged where the rubber meets the road. Providing
leeway to those on the frontlines, in terms of making decisions in deploying
resources to where they are needed to “put out the ﬁre,” is a strategy that a company
can implement. The company can take on the strategy of anticipation when it has
more accurate situational awareness. When faced with shifting variables affecting
the organization, a strategy of containment is in order. The latter requires reliance on
history and organizational learning (Weick and Sutcliffe 2008; Bennis and Nanus
1985).
This ability of the teams to deploy resources at a moment’s notice as a
competence-based-view strategy for business organizations operating in volatile
and uncertain environments is what military science calls Auftragstaktik (Wittmann
2017). The constructs studied in this chapter are similar to the constructs (i.e., trust,
anticipation, containment, leadership, mindfulness, situational awareness, common
vision, social support, deference to expertise, empowerment of subordinates, etc.)
discussed by military science authors (Nelsen 1987; Shamir 2010; Wittmann 2017)
when explicating Auftragstaktik. The model in this chapter encompasses the com-
plexity of the concepts relevant to decentralization of command. Future application
of the model in military science studies should be explored.

6.2 Finding Evidence of Mindful Leadership in Practice

Be mindful, stupid! To exemplify the strategy of small wins (Weick and Sutcliffe
2007), we will use the famous quote that is a result of environmental scanning and
complexity analysis that has defined the rise to power of Bill Clinton—the economy,
stupid. Note that it was not Clinton who coined the phrase, it was actually the
campaign strategist James Carville (a member of his campaign team) who has keenly
summarized the messaging that succinctly communicates the accuracy of the
nation’s economic situational awareness and sensitivity to operation. The note was
meant to remind his campaign team to be mindful and stay focused on the messag-
ing. However, this simple note was akin to a smoldering material that turned into a
wildfire that most political analysts credit Clinton’s rise to the highest office in the
country (Bennett 2013; Christiansen 2018). The act of putting out the sign to remind
the campaign staffers is sensitivity to operation. As a strategy in managing a high-
stake uncertain event like a presidential campaign, one ought to develop sensitivity
to operation—this means spending time on the front end of operation and writing the
three-point messages for the team is a way of letting people know what they are
doing and why they are doing it so that the message STICC (situation, task, intent,
concern, and calibrate) (Klein 1998). The phrase “It’s the economy, stupid” became
the mantra that had brought Clinton a decisive victory in the electoral college of
thirty-two states to Bush’s eighteen (Christiansen 2018). It was the accuracy in
messaging that tapped into the preoccupation of the population with the ensuing
economic failure; the voters needed somebody who could simplify and understand
the situation they were in.
308 S. D. Gapud and J. F. Hair

In the Mindful leadership of Brian Chesky, everyone belongs. As mentioned in

the introduction of this chapter, when an unexpected event impacts an organization,
teaming happens: The pace of work quickens, communication with the team
becomes more frequent like when a fast-moving wildfire is nipping at your boots.
When the COVID-19 pandemic hit, Brian Chesky was leading a huge global firm
that was barely breaking even. How he turned his company to a profitable company
that successfully offered its initial public offering in the midst of the pandemic is
proof that a high degree of uncertainty sets a mindful leader apart. In his interview
with James Manyika (2021), a senior partner emeritus at McKinsey & Company, he
said,
One of the first lessons I learned is that in a crisis, you have to move much more quickly.
People see crises as defining moments, so they try to slow down their decision making, but
instead, you have to speed up. You also have to increase the amount of communication you
do. The faster things change, the more you need to communicate the change.

Consistent with Bennis (2015), to be a mindful leader, one should be comfortable

with operating in times of ambiguity and uncertainty. Similarly, Brian Chesky is
mindfully aware of his influence as a leader as he navigated uncertain times during
the pandemic.
The hardest thing to manage in a crisis isn’t your company—it’s your own thoughts. As a
leader, you could think all is doomed. You can ask, “Why me?” You can get paralyzed. Or
you can tell yourself, “This is my defining moment, and it will leave indelible marks.” Then
you can be optimistic—not with blind optimism but optimism rooted in facts that give you
hope of getting out of the situation. If you can project that confidence, abide by your
principles, and act quickly, then you can navigate out of the crisis like a captain on a ship
that is starting to take on water.

We could continue on how successful Airbnb is since the pandemic served like a
fire that forged a new look for the company. Their organizational mindfulness during
the crisis brought forth the best in them, as Brian Chesky showed his heart in
supporting his team by providing the most generous severance package. Company
supported emphasizing respect and recognition of the talents and abilities of the team
members that he had to painfully let go. Reading the letter should give any person
studying leadership as a process a clear understanding of how mindful leadership
works. In this letter, Brian successfully made known the business theory on how his
shared leadership environment works, how he made time to listen to the voice of
each member, and how he considered all employees as continuing co-owners even
though he had to let them go—which, in turn, encourages their continued support in
the purpose of the corporation. He exemplified leadership behavior to his team by
providing a clear direction in the care of the social, emotional, mental, financial
needs of people he had to let go. He empowered them by providing for all the tools
they would require—from resume writing, coaching, and creating an outward facing
marketing site to highlight the capabilities of the people that would be let go. They
were also allowed to keep their company-issued laptops. The letter, he sent to the
whole company—can be read at https://news.airbnb.com/a-message-from-co-
founder-and-ceo-brian-chesky/.
When Navigating Uncertainty Lead Mindfully 309

Moreover, we mentioned that teaming happens when an unexpected occurs. In

the words of Brian Chesky, “A crisis is a spotlight. All hands are on deck. It’s a
moment to demonstrate your values and lead by example.”
Lastly, the founder letter that Chesky wrote in the IPO is a testament of his
sensitivity to operation and preoccupation with failure. He started writing honestly
about the recent 20% reduction of the workforce and a lengthy take on the impact of
the uncertainty that everyone is experiencing brought about by the pandemic—but
then he appealed to the prospective shareholders by showing his commitment to
resilience through his mindful crafting of his business theory. “The best thing for
shareholders is for society to want us to exist. And society will want us to exist if they
think that as Airbnb beneﬁts, they beneﬁt.” He explained that “The rules of business
have changed. This idea that the job of a corporation is solely to serve shareholders
may have been correct once but talk to anyone under the age of 30—they care about
the quality of the products, but they also care about what the company stands for.”
As Bennis (2012, p. 544) stated in his crucibles of authentic leadership, under-
standing leaders is important because “leaders wield power.” Therefore, only when
we are mindful that leadership emanates from all members of the team do we begin
to realize how to empower them and make the team potent, adaptive, and agile at any
time. But then who is in your team? Airbnb’s founders believe that all stakeholders
are part of their team. And so, we want to end this discussion on mindful leadership
with a forward-facing note from Brian Chesky to exemplify how a leader with the
mindset of organizational mindfulness views the future: “I think capitalism needs a
little bit of an evolution. We need to adapt to meet the needs and challenges in
society, and the next generation of entrepreneurs needs role models. The company is
13 years old. You don’t raise your 13-year-old to be a good 14-year-old. You raise
your 13-year-old to be a great adult with a long career over many decades. That is
how we are thinking about Airbnb.”

7 Future Studies

In this chapter, we explicated that just as the teams of wildland firefighters illustrate,
hierarchy and sharing the lead go hand-in-glove. However, what is the behavioral
mechanism that actually explains the phenomenon? Auftragstaktik or mission com-
mand is a way of flattening bureaucracy by sharing the lead to those on the frontlines.
Our results indicate that through mindful leadership organizations can be adaptive
and flexible in times of uncertainty when teaming. Why would providing decision-
making authority that is usually above one’s paygrade make Auftragstaktik work
and therefore make the organization adapt to a developing scenario in the work-
place? We suggest that the very positive act of the legitimate hierarchical authority in
sharing some of the decision-making responsibility to the lower-rank member of the
organization provides the unlocking mechanism that broadens the repertoire of
action available to the lower-rank team members, much like Fredrickson’s broaden
and build theory (Fredrickson 2001).
310 S. D. Gapud and J. F. Hair

However, mindfulness suggests questioning the absolutes; therefore, this

unlocking mechanism within the leader/follower relationship has to be researched,
measured, and analyzed to contribute knowledge toward greater understanding of
the shared leadership theory.

References

Avolio BJ, Jung DI, Murry W, Sivasubramaniam N (1996) Building highly developed teams:
focusing on shared leadership process, efficacy, trust, and performance. In: Beyerlein MM,
Johnson DA, Beyerlein ST (eds) Advances in interdisciplinary studies of work teams: team
leadership, vol 3. Elsevier Science/JAI Press, US, pp 173–209
Barker JR, Weick KE (1993) Organizing for high reliability: processes of collective mindfulness.
Organ Dyn 22(2):36–52
Becker JM, Klein K, Wetzels M (2012) Hierarchical latent variable models in PLS-SEM: guidelines
for using reflective-formative type models. Long Range Plan 45(5–6):359–394
Bennett AJ (2013) The race for the white house from Reagan to Clinton: reforming old systems,
building new coalitions. Palgrave Macmillan, New York
Bennis W (2012) The crucibles of authentic leadership. In: Antonakis J, Day DV (eds) The nature of
leadership. Sage, Thousand Oaks, CA, pp 543–555
Bennis W (2015) Managing the dream: leadership in the 21st century. Antioch Rev 73(2):364–370
Bennis W, Nanus B (1985) The strategies for taking charge. Leaders. Harper, New York
Bernstein E (2015) Leadership and teaming. September 2013 (Revised August 2015), US: Harvard
Business School
Bigley GA, Roberts KH (2001) The incident command system: high reliability organizing for
complex and volatile task environments. Acad Manag J 44(6):1281–1299
Black AE, McBride BB (2013) Assessing high reliability practices in wildland fire management: an
exploration and benchmarking of organizational culture (pp. 55), (Res. Note RMRS-RN-55),
Fort Collins, CO: US Department of Agriculture, Forest Service, Rocky Mountain Research
Station
Bligh MC, Pearce CL, Kohles JC (2006) The importance of self-and shared leadership in team
based knowledge work: a meso-level model of leadership dynamics. J Manag Psychol 21(4):
296–318
Bollen KA (1989) A new incremental fit index for general structural equation models. Sociol
Methods Res 17(3):303–316
Carson JB, Tesluk PE, Marrone JA (2007) Shared leadership in teams: an investigation of
antecedent conditions and performance. Acad Manag J 50(5):1217–1234
Christiansen P (2018) It’s the economy, stupid!: 1992. In: Orchestrating public opinion: how music
persuades in television political ads for US presidential campaigns, 1952–2016. Amsterdam
University Press, Amsterdam, pp 129–137
Cohen J (1992) A power primer. Psychol Bull 112(1):155–159
Dether D, Black A (2006) Learning from escaped prescribed fires-lessons for high reliability. Fire
Manage Today 66(4):50–56
Dunoon D, Langer E (2011) Mindfulness and leadership: opening up to possibilities. Integral
Leadersh Rev 11(5):1–15
Eastburn RW (2018) Organizational mindfulness in banking: a discriminating factor for firm
performance. J Manag Issues 30(1):122–147
Edmondson AC (2012) Teaming: how organizations learn, innovate, and compete in the knowledge
economy. John Wiley & Sons, San Francisco
When Navigating Uncertainty Lead Mindfully 311

Ensley MD, Hmieleski KM, Pearce CL (2006) The importance of vertical and shared leadership
within new venture top management teams: implications for the performance of startups.
Leadersh Q 17(3):217–231
Fornell C, Larcker DF (1981) Evaluating structural equation models with unobservable variables
and measurement error. J Mark Res 18(1):39–50
Franke G, Sarstedt M (2019) Heuristics versus statistics in discriminant validity testing: a compar-
ison of four procedures. Internet Res 29(3):430–447
Fredrickson BL (2001) The role of positive emotions in positive psychology: the broaden-and-build
theory of positive emotions. Am Psychol 56(3):218–226
Gapud SD (2019) Leadership on fire: mindfully shared. Doctoral Dissertation, University of South
Alabama
Gavetti G, Henderson R, Giorgi S (2005) Kodak and the digital revolution. Harvard Business
School Case 705-448, November 2004. (Revised November 2005), US: Harvard Business
School Publishing
Grabowski M, Roberts K (2011) High reliability virtual organizations: co-adaptive technology and
organizational structures in tsunami warning systems. ACM Trans Comput Hum Interact 18(4):
1–23
Guzzo RA, Yost PR, Campbell RJ, Shea GP (1993) Potency in groups: articulating a construct. Br J
Soc Psychol 32(1):87–106
Hair JF, Sarstedt M, Hopkins L, Kuppelwieser VG (2014) Partial least squares structural equation
modeling (PLS-SEM) an emerging tool in business research. Eur Bus Rev 26(2):106–121
Hair JF, Hult GTM, Ringle CM, Sarstedt M, Thiele KO (2017) Mirror, mirror on the wall: a
comparative evaluation of composite-based structural equation modeling methods. J Acad Mark
Sci 45:616–632
Hair JF, Sarstedt M, Ringle CM, Gudergan SP (2018) Advanced issues in partial least squares
structural equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Howard MC, Nitzl C (2020) Assessing measurement model quality in PLS-SEM using
confirmatory composite analysis. J Bus Res 109:101–110
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Harms IM, Burdett BR, Charlton SG (2021) The role of route familiarity in traffic participants’
behaviour and transport psychology research: a systematic review. Transp Res Interdiscip
Perspect 9:100331
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Klein G (1998) Sources of power: how to make decisions. MIT Press, Cambridge, Mass
Klein KJ, Ziegert JC, Knight AP, Xiao Y (2006) Dynamic delegation: shared, hierarchical, and
deindividualized leadership in extreme action teams. Adm Sci Q 51(4):590–621
Kline RB (2015) Principles and practice of structural equation modeling, 4th edn. The Guilford
Press, NY
Kock N (2015) Common method bias in PLS-SEM: a full collinearity assessment approach. Int J
e-Collab 11(4):1–10
Koopmann CJ, Ancona LM (2015) Team potency: a multilevel construct for teams. Acad Manag J
58(2):431–452
Langer EJ (1989) Minding matters: the consequences of mindlessness–mindfulness. Adv Exp Soc
Psychol 22:137–173
Langer EJ (2000) Mindful learning. Curr Dir Psychol Sci 9(6):220–223
Langer EJ, Moldoveanu M (2000) Mindfulness research and the future. J Soc Issues 56(1):129–139
LaPorte TR (1996) High reliability organizations: unlikely, demanding and at risk. J Contingencies
Crisis Manage 4(2):60–71
312 S. D. Gapud and J. F. Hair

LaPorte TR, Consolini PM (1991) Working in practice but not in theory: theoretical challenges of
high-reliability organizations. J Public Adm Res Theory: J-PART 1(1):19–48
Leavitt HJ (2005) Top down: why hierarchies are here to stay and how to manage them more
effectively. Harvard Business School Press, Boston
Levinthal D, Rerup C (2006) Crossing an apparent chasm: bridging mindful and less-mindful
perspectives on organizational learning. Organ Sci 17(4):502–513
Manley SC, Hair JF, Williams RI, McDowell WC (2021) Essential new PLS-SEM analysis
methods for your entrepreneurship analytical toolbox. Int Entrep Manag J 17:1805–1825
Manyika J (2021) The 21st-century corporation: a conversation with Brian Chesky of Airbnb.
Strategy & Corporate Finance McKinsey & Company. https://www.mckinsey.com/capabilities/
strategy-and-corporate-ﬁnance/our-insights/the-21st-century-corporation-a-conversation-with-
brian-chesky-of-airbnb. Accessed 4 Nov 2021
Manz CC, Pearce CL (2017) Twisted leadership: how to engage the full talents of everyone in your
organization. Maven House Press, Palmyra, VA
Monteiro RB, Vieira VA (2016) Team potency and its impact on performance via self-efﬁcacy and
adaptability. BAR-Braz Adm Rev 13:98–119
National Weather Service (2016) Central Alabama year in review. National Oceanic & Atmo-
spheric Administration [NOAA] https://wwwweathergov/bmx/climo_2016review Accessed
5 Apr 2019
Nelsen JT (1987) Auftragstaktik: a case for decentralized battle. Parameters 17(3):21–34
Pavlovich K (2003) All that jazz. Long Range Plan 36(5):441–458
Pearce CL (2010) Leading knowledge workers: beyond the era of command and control. In: Pearce
CL, Maciariello JA, Yamawaki H (eds) The Drucker difference: what the world's greatest
management thinker means to today's business leaders. McGraw-Hill, NY, pp 36–45
Pearce CL, Conger JA (2003) Shared leadership: reframing the hows and whys of leadership. Sage,
Thousand Oaks, CA
Pearce CL, Manz CC (2005) The new silver bullets of leadership: the importance of self- and shared
leadership in knowledge work. Organ Dyn 34(2):130–140
Pearce CL, Sims HP Jr (2002) Vertical versus shared leadership as predictors of the effectiveness of
change management teams: an examination of aversive, directive, transactional, transforma-
tional, and empowering leader behaviors. Group Dyn Theory Res Pract 6(2):172–197
Pearce CL, Sims HP Jr, Cox JF, Ball G, Schnell E, Smith KA, Trevino L (2003) Transactors,
transformers and beyond: a multi-method development of a theoretical typology of leadership. J
Manag Dev 22(4):273–307
Pearce CL, Manz CC, Sims HP (2014) Share, don’t take the lead: leadership lessons from
21 vanguard organizations. Information Age Publishing, Charlotte, North Carolina
Pearce CL, Houghton JD, Manz CC, Dillon PJ, Fugate M, Wassenaar CL (2022) Time for a
group hug? Toward a theory of shared emotional leadership in and of family business. J Fam
Bus Strat 100549. (in press):100549
Podsakoff PM, MacKenzie SB, Moorman RH, Fetter R (1990) Transformational leader behaviors
and their effects on followers’ trust in leader, satisfaction, and organizational citizenship
behaviors. Leadersh Q 1(2):107–142
Ray JL, Baker LT, Plowman DA (2011) Organizational mindfulness in business schools. Acad
Manage Learn Educ 10(2):188–203
Rigdon EE (2012) Rethinking partial least squares path modeling: in praise of simple methods.
Long Range Plan 45(5–6):341–358
Ringle CM, Sarstedt M, Straub DW (2012) A critical look at the use of PLS-SEM in MIS quarterly.
MIS Quarterly 36(1):iii–xiv
Ringle CM, Wende S, Becker J-M (2022) SmartPLS 4. SmartPLS GmbH, Oststeinbek. http://www.
smartpls.com
Roberts KH (1990) Some characteristics of high reliability organizations. Organ Sci 1:160–177
Roberts KH, Madsen P, Desai V, Van Stralen D (2005) A case of the birth and death of a high
reliability healthcare organization. Qual Saf Health Care 14:216–220
When Navigating Uncertainty Lead Mindfully 313

Rochlin GI, LaPorte TR, Roberts KH (1987) The self-designing high-reliability organization:
aircraft carrier flight operations at sea. Nav War Coll Rev 40(4):76–92
Sarstedt M, Ringle CM, Smith D, Reams R, Hair JF Jr (2014) Partial least squares structural
equation modeling (PLS-SEM): a useful tool for family business researchers. J Fam Bus Strat
5(1):105–115
Sarstedt M, Hair JF Jr, Cheah J-H, Becker J-M, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J 27(3):197–211
Sashkin M (2018) A model of leadership for the 21st century [Lecture notes]. Mitchell College of
Business, Mobile, AL
Schaubroeck J, Lam SSK, Peng AC (2011) Cognition-based and affect-based trust as mediators of
leader behavior influences on team performance. J Appl Psychol 96(4):863–871
Shamir E (2010) The long and winding road: the US Army managerial approach to command and
the adoption of mission command (Auftragstaktik). J Strateg Stud 33(5):645–672
Shmueli G, Ray S, Estrada JMV, Chatla SB (2016) The elephant in the room: predictive perfor-
mance of PLS models. J Bus Res 69(10):4552–4564
Shmueli G, Sarstedt M, Hair JF, Cheah J, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Sivasubramaniam N, Murry WD, Avolio BJ, Jung DI (2002) A longitudinal model of the effects of
team leadership and group potency on group performance. Group Org Manag 27(1):66–96
Thomas D, Black A, Dether D, Hetts K, Dueitt M (2007) The jungle prescribed fire review: an
experiment in learning. Wildland Fire Lessons Learned Center, Tucson, AZ, pp 1–42
Vogus TJ (2011) Mindful organizing: establishing and extending the foundations of highly reliable
performance. In: Cameron K, Spreitzer G (eds) Handbook of positive organizational scholar-
ship. Oxford University Press, Oxford
Vogus TJ, Welbourne TM (2003) Structuring for high reliability: HR practices and mindful
processes in reliability-seeking organizations. J Organ Behav 24:877–903
Wassenaar CL, Pearce CL (2017) The nature of shared leadership. In: Antonakis J, Day DV (eds)
The nature of leadership. Sage, Thousand Oaks, CA, pp 167–188
Weick KE (1987) Organization culture as a source of high reliability. Calif Manage Rev 29:116–
136
Weick KE (1995) Sensemaking in organizations (foundations for organizational science). Sage
Publications, Thousand Oaks, LA
Weick KE, Roberts KH (1993) Collective mind in organizations: heedful interrelating on flight
decks. Adm Sci Q 38:357–381
Weick KE, Sutcliffe KM (2001) Managing the unexpected: assuring high performance in an age of
complexity. Jossey-Bass, San Francisco
Weick KE, Sutcliffe KM (2007) Managing the unexpected: resilient performance in an age of
uncertainty, 2nd edn. Jossey-Bass, US
Weick KE, Sutcliffe KM (2008) Organizing for higher reliability: lessons learned from wildland
firefighters. Fire Manage Today 68(2):14–19
Weick KE, Sutcliffe KM, Obstfeld D (1999) Organizing for high reliability: processes of collective
mindfulness. Res Organ Behav 21:81–123
Weick KE, Sutcliffe KM, Obstfeld D (2005) Organizing and the process of sensemaking. Organ Sci
16(4):409–421
Wetzels M, Odekerken-Schroder G, van Oppen C (2009) Using PLS path modeling for assessing
hierarchical construct models: guidelines and empirical illustration. MIS Q 33(1):177–195
Wildavsky A (1988) Searching for safety. Transactions publishers. New Brunswick, NJ
Wittmann J (2017) Auftragstaktik for business organizations in volatile and uncertain environ-
ments: a competence-based view. IAFOR J Bus Manag 2:29–49
Zaccaro, S. J., Rentsch, J. K., & Marks, S. C. (2001). The role of team potency in team performance:
a meta-analysis. J Appl Psychol, 86(5), 989–1004. https://news.airbnb.com/a-message-from-co-
founder-and-ceo-brian-chesky/ Accessed 10 June 2021
The Interplay Between Push Factors
and Transformational Leadership
in Influencing Interorganizational Labor
Mobility in Public Sector

Rosemary Massae, Deusdedit A. Rwehumbiza, and John J. Sanga

1 Introduction

Labor mobility facilitates productive employer–employee matches (Akgündüz et al.

2019). However, in the current world, labor mobility is considered one of the
problems that managers in all organizations have to take into consideration, espe-
cially when talented employees are to leave (Linhartová and Urbancová 2012).
Organizations spend massive resources to make their employees pleased to work
effectively and efficiently. Nevertheless, the workers still feel dissatisfied and leave
their current organization and join a new one (Chowdhury 2016).
Job mobility measures the number of jobs that change throughout an employee’s
career, including intra- and inter-organizational changes (Hall 1996). On the other
hand, according to Agarwal et al. (2020), labor mobility is the process of an
employee(s) moving across different jobs until she/he finds the one with the right
fit. Studies show that different types of labor mobility exist. Studies show that labor
turnover is widespread, labor mobility and interorganizational labor mobility (ILM),
in particular, remains relatively underexplored in management research (Agarwal
et al. 2020; Steenackers and Guerry 2016). ILM is also known as turnover intention
(TOI) (de Luis Carnicer et al. 2004; Wynen et al. 2013). However, from the human
resources management position, ILM is theoretically expensive and may have
negative consequences for an organization, such as that of labor turnover. The
costs of employee ILM often exceed 100% of the yearly remuneration for the

R. Massae (✉)
Institute of Finance Management, Dar es Salaam, Tanzania
e-mail: [email protected]
D. A. Rwehumbiza · J. J. Sanga
University of Dar es Salaam Business School, Dar es Salaam, Tanzania

© Springer Nature Switzerland AG 2023 315

vacated position due to the recruitment, selection, and training of new employees
(Bryant and Allen 2013).
The push-pull model invented from the human migration philosophy is a basis for
understanding ILM (Haldorai et al. 2019). Push factors have been considered
because they consist of psychological machineries for leaving the organization
such as job satisfaction and organizational commitment that outweigh the economic
factors for leaving (Griffeth et al. 2000). Hence, this chapter’s ﬁrst and second
objectives are to test the direct effect of push factors on ILM in the public sector.
The increasing ILM in the public sector can be facilitated by the provisions of the
Public Service Management and Employment Policy 1999 as amended in 2008, that
there shall be free movement of labor both within the public service and between the
public service and the private sector (URT 1999). The provisions of this policy are
contrary to the labor retention theories which consider a high rate of labor mobility to
be an unhealthy condition for organizations.
Few studies have considered the role leaders can play on voluntary turnover
dynamics, and maximum of these have been carried out on developed countries such
as the United States of America and Europe (Rubenstein et al. 2018; Akgündüz et al.
2019). Further research into the impact of transformational leadership (TL) on
employees ILM is worth to be undertaken, to enable generalizability (Sun and
Wang 2017; Tse et al. 2013). Consequently, this chapter’s third objective is to test
the interaction effect of TL and push factors on ILM, which has never existed, hence
bridging up the gap.
Tse et al. (2013) show that Social Exchange Theory (SET) can help to explain
how and why TL should be considered a necessary “pull-to-stay” factor, deterring
employees from forming an intention to leave.
Existing studies disclose that, even though determinant factors for leaving or
staying are applicable in a broader setting, cultural, customs, traditional, and insti-
tutional differences will exist. Hence, the results from developed countries cannot be
automatically generalized to non-developed ones (Agarwal et al. 2020; Sousa-Poza
and Henneberger 2004). Hence, this chapter has contributed to the existing labor
mobility model by examining the moderating effect of TL on the relationship
between coworkers’ relationship and work–family conﬂict and ILM in the public
sector.

2 Theoretical Framework and Hypotheses

2.1 Social Exchange Theory (SET)

The theoretical foundation of this study is grounded in Social Exchange Theory

(SET) under the “basic rules and norms of exchange” within the social relationship,
which is known as reciprocity (Blau 1964). Cropanzano and Mitchell (2005) assert
that the core principle of SET is that the association between two social beings
depends on how each of these beings respects social rules and customs of exchange.
The Interplay Between Push Factors and Transformational Leadership. . . 317

Scholars such as Tse et al. (2013) assert that interpersonal trust, mutual loyalty,
emotional identiﬁcation, and continuing reciprocal actions from social exchange
relationships become a strong “pull-to-stay” factor, discouraging employees from
undertaking ILM. Despite its usefulness in explaining the research variables, SET
has been criticized for lacking sufﬁcient theoretical precision (Cropanzano et al.
2017). Hence, Herzberg’s Two-factor Theory complemented it.

2.2 Herzberg’s Two-Factor Theory

Herzberg’s Two-Factor Theory is one of the most signiﬁcant theories in studying

turnover intention and labor mobility in general (Chiat and Panatik 2019). This
theory recognizes that employees have two categories of needs affecting job satis-
faction: hygiene and motivation. Motivation factors lead to job satisfaction and
hence are called motivators; they include achievement, recognition, the work itself,
responsibility, advancement, and the possibility of growth (Alshmemri et al. 2017).
The hygiene factors are conditions surrounding the job. They include company
policy and administration practices, quality supervision, interpersonal relationship,
physical working conditions, salary, status and job security, whose negative aspects
lead to dissatisfaction (Alshmemri et al. 2017). This chapter assumes that quality
leadership behavior and good coworkers’ relationship will reduce ILM for hygiene
factors.

2.3 Research Model and Hypotheses

The conceptual model shows the relationship between independent variables

(coworkers’ relationship, work–family conﬂict) and TL, a moderating variable,
and ILM, the dependent variable.

2.3.1 Interorganizational Labor Mobility and Push Factors

ILM is a type of employee turnover based on the mobility of employees to a new

organization called external turnover (Mbah and Ikemefuna 2012). On the other
hand, the TOI is addressed as an indicator of ILM and a substitute for actual turnover
behavior (Liljegren and Ekberg 2009). Hence, in this chapter, ILM was used
interchangeably with TOI and was considered to be employees’ TOI, which refers
to a conscious and deliberate willfulness of Government public employees to seek
other alternatives within public service organizations (Tett and Meyer 1993).
Scholars such as Chowdhury (2016) and Haldorai et al. (2019) found that the
most signiﬁcant factors for employees to quit a job are the push factors. These
include small salary, less fringe beneﬁts, social status, working environment, lack of
318 R. Massae et al.

recognition, career development, work–family conﬂict, and interpersonal tension.

The studies above have considered the direct relationship between push factors and
ILM. However, this chapter has considered the moderating effect of TL on the
relationship between push factors and ILM, which is considered as an extension of
the existing ILM model.

Coworkers’ Relationship and Interorganizational Labor Mobility

Kim et al. (2013) reveal that coworkers’ relationship negatively correlated with
turnover intention in China and Korea. Wong et al. (2017) assert that Gen Y
employees emphasize the necessity of friendly interpersonal relationship among
coworkers as an indispensable factor for enduring in the hospitality business in
China. Consistent with Herzberg et al. (1959), who posit that the work conditions,
such as poor relationships with coworkers, cause employees’ dissatisfaction
resulting in turnover intention. Based on this discussion, the following hypothesis
is formulated:
H1: Coworkers’ relationship has a negative inﬂuence on interorganizational labor
mobility.

Work-Family Conﬂict and Interorganizational Labor Mobility

Work–family conflict occurs when there is interference between work and family life
experiences, for example, rigid work hours, work overload, job stress, interpersonal
conflict, prevalent travel, career changes, uncooperative supervisor or lack of orga-
nizational support (Greenhaus et al. 1989). Yildiz et al. (2021) indicate a positive
correlation between work–family conflict and intention to leave. Herzberg et al.
(1959) posit that if the organization provides its employees with good work–family
policies and sound human resource practices, it will enhance their job satisfaction
and intention to stay. The opposite is true when there is an imbalance between work
and family activities. With the above in mind, the following hypothesis is expressed:
H2: Work–family conflict has a positive influence on interorganizational labor
mobility.

The Moderating Role of Transformational Leadership

A number of leadership models have been developed forecasting the relationships

between leadership and work-related consequences in the public sector context
(Moon and Park 2019; Trottier et al. 2008). However, TL is considered more
important than transactional leadership (Moon and Park 2019).
This is due to the fact that TL motivates and encourages workers to outdo
expectations and surpass their own interests for the sake of an organization (Bass
The Interplay Between Push Factors and Transformational Leadership. . . 319

1985). Studies have considered TL as a pull-to-stay factor as a result of social

network within an organization from SET (Tse et al. 2013). Previous studies have
shown that TL as a moderator can lessen the impact of both pull and push to leave
factors. However, in contrast to the current chapter, those pull and push factors were
not explicitly outlined (Waldman et al. 2015). Moreover, TL was found to relate
negatively to employees exit behavior (Moon and Park 2019). From the above
arguments, the following hypotheses are postulated:
H3: Transformational leadership moderates the influence of push factors on
interorganizational labor mobility.
H3a: The higher (lower) the transformational leadership, the weaker (stronger) the
influence of coworkers’ relationships on ILM.
H3b: The higher (lower) the transformational leadership, the weaker (stronger) the
influence of work–family conflict on ILM.

3 Research Methodology

A positivist research philosophy was adopted because the researcher observed and
measured social realities to arrive at causal explanation and prediction as a contri-
bution (Saunders et al. 2016). The commonly used approach to theory testing in
positivism is a deductive approach associated with highly structured quantitative
data analysis methods (Creswell 2012). Thus, a survey strategy was used to collect
quantitative data (Saunders et al. 2016).

3.1 Area of Study and Population

The survey was done in the Tanzania mainland. Given the data from the office of the
Treasury Registrar, there were 27 government agencies with 8856 employees during
the research time (www.estabs.go.tz). However, most of these executive agencies’
human resource management practices are currently centralized. However, this was
contrary to the rationale of their establishment, leading to a decreased sense of
autonomy, which is expected to decrease morale of not only leaders but also
employees. Furthermore, the justification for selecting the public sector, was also
based on the provision of Public Service Management and Employment Policy of
1999 as amended 2008. This policy is conflicting the labor retention theories, which
posit the necessity of employee retention, to avoid wastage of both human resources
and drop in productivity. Geographically, 78% of these executive agencies’ head
offices are located in Dar es Salaam, with the rest located in the Coastal region
(11%), Morogoro (7%) and Dodoma (4%).
320 R. Massae et al.

3.2 The Sample Size

The sample size of 383 employees was derived with the help of Yamane (1973)’‘s
formula, n = N / [1 + N (e) 2]. Hence, eight agencies with more than 200 employees
situated in Dar es Salaam city were used to draw the sample. Speciﬁcally, large
organizations were preferred for investigation because they have a formal human
resource (HR) department that inﬂuences HR practices (Michael 2009). This chapter
adopts probability simple random sampling technique (Saunders et al. 2016). The
list was obtained from the HR department as follows: Tanzania Public Service
College (TPSC) = 227, Tanzania Institute of Accountancy (TIA) = 222, Tanzania
Airport Authority (TAA) = 768, Tanzania Building Agency (TBA) = 358, Weight
and Measure Agency (WMA) = 236, Tanzania Medicine & Medical Device
Authority (TMDA) = 266, Tanzania National Roads Agency (TANROAD) = 686,
and Government Procurement Service Agency (GPSA) = 266. Thereafter, the
sample was selected proportionally as follows: the sample size assumed was
n = 460, the total population was N = 3029 divided into 8 agencies, such that
N1 = 227 for (TPSC), N2 = 222 for (TIA), N3 = 768 for (TAA), N4 = 358 for
(TBA), N5 = 236 for (WMA), N6 = 266 for (TMDA), N7 = 686 for (TANROADS)
and N8 = 266 for (GPSA). Then, the sample size was obtained as follows: for
N1 = 227, we had P1 = 227/3029 and hence n1 = n*P1 = 460 (227/3029) = 35;
n2 = n*P2 = 460 (222/3029) = 34; n3 = n*P3 = (768/3029) = 117;
n4 = n*P4 = 460 (358/3029) = 54; n5 = n*P5 = 460 (236/3029) = 36;
n6 = n*P6 = 460 (266/3029) = 40; n7 = n*P7 = 460 (686/3029) = 104;
n8 = n*P8 = 460 (266/3029) = 40.

3.3 Data Collection

This chapter mainly focused on the primary data collected using a survey tool
(Creswell 2012). The researchers conducted the pilot study from March to April
2021. The data were thoroughly analyzed, and the researchers were able to measure
the quality of the instrument through data validity and reliability (Saunders et al.
2016).
During the large-scale survey, a drop-and-pick method was used to administer the
questionnaires due to its efficiency in comparison to others such as phone and the
Internet (Jackson-Smith et al. 2016). The participating agencies were contacted first
through physical addresses and telephone numbers. After that, the questionnaire was
dropped off to each respondent in their respective department. The filled-up ques-
tionnaires were later on gathered.
Out of 460 surveys incorporated a buffer for the risk of non-response, which is
estimated to be 20% for management studies in Tanzania (Goodluck 2009). There-
fore, 389 were collected, making a response rate of 84.6%. All received responses
were examined for missing values. Responses with a more significant number of
The Interplay Between Push Factors and Transformational Leadership. . . 321

missing values (more than 15%) were dropped out as per recommendation from Hair
et al. (2010). Speciﬁcally, 56 questionnaires were dropped out due to their having a
lot of missing values. Hence, 333 were retained for data analysis, which had three
missing values accounting for only 0.9% of the total data set. Hence the few missing
values were retained and given a code of (- 99) (Hair et al. 2017).

3.4 Measurement of Variables

The coworkers’ relationship was measured by six items that were adapted from Tang
(1998) and Hain and Francis (2004). The work–family conflict was measured by six
items that were adapted from Carlson et al. (2000). Six adapted items measured TL
adapted from Jensen et al. (2019). Furthermore, six items were used to measure TOI
by adapting Walsh et al. (1985) and Jung and Yoon (2013).
The two independent and moderator variables were operationalized based on
existing validated scales using seven-point Likert scales ranging from 1 = (strongly
disagree) to 7 = (strongly agree). However, the TOI was operationalized based on
validated five-point Likert scale ranging from 1 = (strongly disagree) to 5 = (strongly
agree). The use of a five-point Likert scale for the dependent variable and a seven-
point Likert scale for the independent variable, was among the routine therapy taken
to avoid Common Method Variance (CMV) (Podsakoff et al. 2003). All variables
were captured using multi-item indicators to avoid estimation bias (Hair et al. 2017),
(see Table 1).

Table 1 The operationalization and measurement of research variables

Variable Definition Indicators Measure Reference
Interorganizational The government employees’ 06 Ordinal Walsh et al.
labor mobility/ turnover intention which is (1985), Jung
turnover intention mindful and deliberate willful- and Yoon
ness to seek for another alterna- (2013)
tive within public service
Coworkers’ Involves integration, workgroup 06 Ordinal Tang (1998),
relationships cohesion, social capital, and the Hain and
primary group that assists with Francis
job-related problems (2004)
Work–family The form of interfunctions con- 06 Ordinal Carlson et al.
conflict flict whereby the responsibilities (2000)
burdens from the work and family
spheres are mutually mismatched
in some respect
Transformational The type of a leader who looks for 06 Ordinal Jensen et al.
leadership potential motives in followers (2019)
seeks to sat-
isfy higher demands, and
engages the full person of the
follower
322 R. Massae et al.

3.5 Data Analysis Method

Survey questionnaire was used to gather the data; however, it introduces the risk of
CMV as severe methodological challenge in public management studies (Jakobsen
and Jensen 2015). CMV is the systematic variance share among the variables
introduced to the measures by the measurement method rather than the measure’s
theoretical constructs (Podsakoff et al. 2012). Tehseen et al. (2017) recommend
using procedural and statistical remedies to test and control CMV. For example, the
researchers adopted procedural remedies, including avoiding common response
scale properties for different items, protecting the anonymity of the respondents
and avoiding item ambiguity.
Furthermore, researchers tested variance inflation factor (VIF) using a full col-
linearity test as a statistical remedy. CMV can be successfully recognized by using
full collinearity test, with a model that passes standard convergent and discriminant
validity test (Kock 2015). Using SmartPLS 3.2.7, all VIF values were lower than 3.3,
demonstrating that the model was free from CMV, as recommended by (Kock 2015).
Table 2 below indicates that all VIF are below 3.3.
Partial least squares structural equation modeling (PLS-SEM) was the preferred
method because the primary purpose of the research objective is the prediction and
explanation of the target endogenous latent variable.
PLS-SEM, unlike other methods, depends on composites variables; additionally,
it can analyze both reflective and formative measurement models without identifi-
cation problems (Hair et al. 2011).
SmartPLS 3.2.7 was used to calculate two PLS path models in this chapter
(Ringle et al. 2015). Evaluation of PLS-SEM results encompasses two stages: The
first one is examining the measurement model. If the evaluation supports the
measurement model quality, the researchers proceed with the structural model
evaluation (Sarstedt et al. 2017). Hence, the structural models were primarily
determined based on the following predictive capabilities: the significance of the
path coefficient (β) which represents the strength of the relationship; the coefficient
of determination (R2); effect size f 2 which represents the magnitude of the hypoth-
esized relationship; predictive relevance (Q2); and PLSpredict (Q2predict) (Hair et al.
2017; Cohen 1988). The R2 denotes the overall structural model’s in-sample pre-
dictive power, and PLSpredict represents the structural model’s out-of-sample
predictive power (Shmueli et al. 2016).

Table 2 Multicollinearity assessment results

Latent Variable VIF Multicollinearity problem (VIF > 3)
Turnover intention
Coworkers relationship 1.030 No
Transformational leadership 1.000 No
Work–family conﬂict 1.030 No
Notes: VIF means Variance Inﬂation Factor
The Interplay Between Push Factors and Transformational Leadership. . . 323

4 Results

4.1 Demographic Characteristics

The results indicated that most respondents were males, 209 (62.8%), whereas
females were 124 (37.2%). This reflects the working nature of public service
organizations. It could be attributed to the fact that the nature of work in most of
these agencies is masculine because it consists of engineers and technicians of
various professional backgrounds. In terms of age, the study consists of employees
of various age groups as follows: 20–30 years were 79 (23.7%); 31–40 were
152 (45.6%); 41–50 were 69 (20.7%); whereas those above 50 years of age were
33 (9.9%). It implies that youth made a great composition of the working staff. The
respondents’ educational backgrounds, including certificate, were 13 (3.6%);
Diploma were 33 (9.9%); First degree was 136 (40.9%); Master’s degree were
146 (43.8%), and Doctoral were 5 (1.5%). The results imply that the employers of
these agencies educate their employees or employ highly qualified people.

4.2 Assessment Results of a Reﬂective Measurement Model

This chapter used reflective indicators for its measurement model. Consequently,
reliability and validity evaluating criteria were used, to ascertain the quality of the
measurement model. Indicator reliability is attained when standardized outer loading
is 0.708 or higher (Hair et al. 2017). But outer loadings of indicators ranging from
0.4 to 0.7 were regarded for deletion from the scale only when eliminating them
results in the rise of composite reliability or average variance extracted (AVE) above
the recommended threshold value, 0.5 (Hair et al. 2017).
Hair et al. (2019) propose that indicators exhibiting low loading of 0.4 and below
should permanently be removed from the reflective scales. The model was originally
composed with 24 indicators; however, after analysis, only 22 indicators passed the
required minimum threshold values. Consequently, two indicators were eradicated
due to an absence of significant input to AVE and composite reliability. Hence, the
revised measurement model results indicated that all variables have achieved the
quality indicator reliability.
The chapter assessed internal consistency reliability by using composite reliabil-
ity (ρc). The ρc values ranging from 0.6 to 0.7 are relevant in exploratory research.
However, in more advanced research stage, the ρc values between 0.7 and 0.9 are
regarded as satisfactory, whereas values above 0.95 are redundant (Hair et al. 2017).
Entirely, the latent variables exceeded the minimum required threshold of ρc, proof
that there was internal consistency reliability on all measures (see Table 3 below).
Convergent validity was measured by using an average variance extracted (AVE), an
AVE of 0.5 or higher indicates that, on average, the construct explains more than
half of the variance of its indicators (Hair et al. 2017). The AVE values for all
324 R. Massae et al.

Table 3 Measurement model evaluation results

Latent variable Code Loading Composite reliability AVE
Turnover intention 0.944 0.772
TOI1 0.891
TOI2 0.897
TOI3 0.818
TOI5 0.913
TOI6 0.872
Coworkers’ relationship 0.876 0.588
CR1 0.768
CR2 0.855
CR3 0.725
CR4 0.826
CR6 0.643
Work–family conﬂict 0.897 0.595
WFC1 0.765
WFC2 0.819
WFC3 0.841
WFC4 0.831
WFC5 0.667
WFC6 0.692
Transformational leadership 0.894 0.585
TL1 0.796
TL2 0.883
TL3 0.716
TL4 0.731
TL5 0.731
TL6 0.721

reflective latent variables were 0.5 or higher, confirming latent variables convergent
validity.
The traditional Fornel-Lacker criterion and cross-loading calculation have an
undesirable low sensitivity, as such they are unable to sense an absence of discrim-
inant validity. Therefore, Henseler et al. (2015) have suggested a novel measure for
evaluating constructs’ discriminant validity using heterotrait–monotrait ratio
(HTMT). This discriminant validity assessment overcomes the weaknesses.
HTMT 0.85 and HTMT 0.90 are the criteria of assessing discriminant validity; of
the two criteria, HTMT 0.85 is the most conservative one. On the other hand, HTMT
inference is the statistical test of assessing discriminant validity. HTMT inference is
the most liberal approach of all due to its much higher specific values.
Hence, with HTMT inference, it is possible to compute bootstrap confidence
interval (C.I), whereby the C.I containing Fig. 1 indicates lack of discriminant
validity. However, if the value of 1 is not within the C.I range, this recommends
that constructs are empirically distinct (Henseler et al. 2015).
The Interplay Between Push Factors and Transformational Leadership. . . 325

Push Factors

Co-workers’
Relationship H1
Inter-organiza-
tional Labor Mo-
bility (Turnover
Intention)
Work-family
H2
Conflict H3 H3

Transformational
Leadership

Fig. 1 Conceptual model

Table 4 Discriminant validity assessment using HTMT criterion

HTMT criterion
Coworkers’ Transformational Turnover Work–family
Construct relationship leadership intention conflict
Coworkers’
relationship
Transformational 0.083
leadership [0.087; 0.154]
Turnover intention 0.154 0.095
[0.092; 0.250] [0.072; 0.187]
Work–family 0.183 0.069 0.495
conflict [0.136; 0.263] [0.079; 0.163] [0.411;
0.572]
Notes: HTMT represents the heterotrait–monotrait ratio; Whereas the 95% bias corrected and
accelerated C.I. are specified by the numbers inside the brackets (Henseler et al. 2015) the whole
numbers imply the HTMT values

The maximum HTMT value of 0.495 from Table 4 below, is below the most
conservative critical HTMT, i.e., 0.85. Additionally, the derived bootstrapping C.I
results show that Fig. 1 falls outside the C.I range. As such, the model is considered
of high quality due to meeting the threshold criteria for measurement model assess-
ment of reliability, convergent and discriminant validity.
326 R. Massae et al.

4.3 Assessment of the Structural Model Relationships

Before assessing structural model relationships, collinearity was examined to make

sure that it does not bias the regression results. VIF is used to assess the severity of
multicollinearity. Multicollinearity is the correlation of two or more variables
(Saunders et al. 2016). The existence of multicollinearity can cause severe problems
in estimating the PLS-SEM path coefficient of the individual predictor variable and
interpretation of results (Joshi et al. 2012). VIF values above 5 indicate collinearity
problems among predictor variables (Becker et al. 2015); nonetheless, it can also
happen at a minor value of 3–5. The accepted VIF values should be close to 3 or
below, of which the analysis results from Table 2 indicate that multicollinearity is
not a problem.
Then coefficient of determination (R2) which representing model explanatory
power as presented in Figs. 2, 3, and 4. The rule of thumb for R2 is as follows: 0.19
implies weak; 0.33 implies moderate; and 0.67 implies substantial (Chin 1998).
However, according to Hair et al. (2017), R2 of 0.10 can be considered satisfactory.
Figure 2 above indicates that the R2 for the direct effect model has attained a value of
0.210, which is considered satisfactory, confirming the direct model explanatory
power. Precisely, this implies that coworkers’ relationship and work–family conflict
explain 21.0% of the variation in ILM, whereas the outstanding 79% will be
accounted for by other factors. Additionally, Fig. 3 above indicates that the R2 for
the main effect model was 0.219, which is also considered satisfactory. Explicitly,
this implies that coworkers’ relationship, work–family conflict, and TL describe

CR2

CR3 0.855
0.725
CR4
0.826 TOI1
0.644
CR6
0.768
Co-workers 0.891 TOI2
CR1 relationship –0.082
0.897
TOI3
WFC1 0.822
0.210
0.910
TOI5
WFC2 0.765 0.871
0.437 Tumover intention
0.819
WFC3 TOI6
0.841
0.831
WFC4 0.662

WFC5 0.692 Work-family

conflict

WFC6

Fig. 2 Measurement model for the direct relationship

The Interplay Between Push Factors and Transformational Leadership. . . 327

Fig. 3 Measurement model for the moderated relationship/direct effect PLS path model

Fig. 4 The measurement model for the interaction effect pls path model

21.9% of the variation in ILM, whereas other factors account the remaining 78.1%.
Moreover, Fig. 4 indicates that R2 for the interaction effect model was 0.233.
According to Ramayah et al. (2018), R2 change becomes an essential issue in
moderation analysis. R2 for the main effect model was 0.219, and R2 for the
328 R. Massae et al.

Table 5 PLSpredict results

PLS-SEM LM PLS-SEM-LM
RMSE Q2predict RMSE RMSE
TOI1 1.095 0.171 1.125 -0.030
TOI2 1.165 0.157 1.190 -0.025
TOI3 1.057 0.131 1.089 -0.032
TOI5 1.167 0.176 1.199 -0.032
TOI6 1.090 0.111 1.116 -0.026
Notes: RMSE represents Root Mean Square Error; LM represents Linear Model; Q2predict represents
Out-of-sample prediction power; whereas TOI represents Turnover intention

interaction effect model was 0.233. The R2 change of 0.014 indicates that R2 has
changed about 1.4% with the addition of two interaction terms.
The next step was to calculate the effect size of the interaction term using the
formula f 2 = R2 included - R2 excluded/1 - R2 included. R2 included representing the
interaction effect model R2 and R2 excluded representing the main effect model R2.
The f 2 obtained was medium which is 0.02. This interpretation was based on the
following suggested guidelines 0.005 small, 0.01 medium, and 0.025 large (Kenny
2016).
The blindfolding procedure indicates that the influence of push factors on ILM
has attained a small Q2 value of 0.15 (15%), confirming the direct effect model
predictive relevance. This is according to the suggested Q2 threshold value of
0, 0.25, and 0.50, indicating the PLS path model’s small, medium, and large
predictive relevance (Hair et al. 2017). Additionally, Q2 for the main effect model
value was 0.155 (15.5%), consequently, confirms the main effect model predictive
relevance.
Then, the model’s predictive power was calculated by running the PLSpredict
procedures with 10 folds and ten reiterations. Out-of-sample predictions are helpful
for evaluation when the focus is on the model’s ability to predict the outcome of
heretofore unseen cases (Shmueli et al. 2016). The focus was on the model’s key
target variable, TOI, and its five scale items. Table 4 shows that all five indicators
achieve Q2predict greater than zero, demonstrating that the model outdoes the naïve
standard. Furthermore, prediction errors analysis indicates non highly unsymmetric
distribution. Henceforth, the appropriate examination concentrates on the root mean
square error (RMSE) statistics. The analyses show that the RMSE values produced
by the PLS path model were reliably lower than those of the linear model
(LM) standard, as indicated in Table 5. According to the rule of thumb for running
PLSpredict, PLS-SEM < LM for all TOI scale items: Hence if all indicators in the
PLS-SEM analysis have lower RMSE values than the naïve LM standard, the model
has high predictive power, consequently the results confirming the model’s large
out-of-sample predictive power (Hair et al. 2019).
The Interplay Between Push Factors and Transformational Leadership. . . 329

4.4 Signiﬁcance Testing Results of the Structural Model

Table 6 shows that coworkers’ relationship exerted a signiﬁcant inﬂuence on ILM,

similar to the influence that work–family conflict (WFC) exerted on ILM. Then,
bootstrapping examination confirms the significant, negative influence of
coworkers’ relationship on ILM [β = -0.087, p < 0.05] (Fig. 5). Further, the
bootstrapping analysis indicated that the C.I [-0.146; 0.018] does not contain
zero, leading to the acceptance of H1: coworkers’ relationship negatively influences
ILM. Equally, the positive relationship between WFC and ILM is significant
[β = 0.426, p < 0.001] (Fig. 5).
Further, the bootstrapping analysis indicated that the C.I [0.350; 0.493] does not
contain zero, which leads to the acceptance of H2: WFC has a positive influence
on ILM.

4.4.1 Hypotheses for a Moderation Effect

Table 6 above the bootstrapping analysis shows that only one hypothesis out of two
was significant [β = 0.118, p < 0.05] (Fig. 6). Further, the bootstrapping analysis
indicated that the C.I [0.036; 0.187] does not contain zero, justifying the moderating
effect. The result implies that TL has a moderating role on the relationship between
WFC and ILM (WFC*TL→ILM); therefore, compliance with H3b: Hence, the
finding implies that the higher the TL, the weaker the influence of WFC on ILM
and vice versa for the lower.

Table 6 Signiﬁcance testing results

Path p- 95% Confidence
Relationships coefficients t-values values interval f2
CR→TOI/ILM -0.082 1.832* 0.034 [-0.135; 0.017] 0.008
WFC→TOI/ILM 0.437 10.466*** 0.000 [0.364; 0.501] 0.238
Moderating effect 0.019 0.326 NS 0.372 [-0.077; 0.115] N/A
1→TOI/ILM
Moderating effect 0.118 2.331** 0.009 [0.036; 0.187] N/A
2→TOI/ILM
R2 direct effect model 0.210
R2 main effect model 0.219
R2 simple effect model 0.233
Q2 direct effect model 0.150
Q2 main effect model 0.155
f 2 of interaction effect 0.020
Notes: NS Not significant, N/A Not applicable COMP Compensation, TD Training and develop-
ment, TOI/ILM Turnover intention/Interorganizational labor mobility ***p < 0.001, **p < 0.01,
*p < 0.05
330 R. Massae et al.

CR2

CR3
0.855 (0.000)

0.725 (0.000)
CR4
0.826 (0.000) TOI1
0.644 (0.000)
CR6
0.768 (0.000)
Co-workers 0.891 (0.000) TOI2
CR1 relationship –0.082 (0.037)
0.897 (0.000)
TOI3
0.822 (0.000)
WFC1
0.910 (0.000)
TOI5
WFC2 0.765 (0.000) 0.871 (0.000)
0.437 (0.000) Tumover intention
0.819 (0.000)
WFC3 TOI6
0.841 (0.000)
0.831 (0.000)
WFC4 0.662 (0.000)

WFC5 0.692 (0.000) Work-family

conflict

WFC6

Fig. 5 Structural model results for the direct effect model

Fig. 6 Structural model results for the indirect effect model

Furthermore, the interpretation of the moderating effect can be explained more by

looking at the interaction plot. Standardized β coefﬁcients are 0.437 from WFC to
TOI, 0.120 from TL to TOI, and the interaction effect of 0.118, with R2 of 0.233, as
shown in Fig. 4. Hence, by looking at the interaction plot from Fig. 7, we can
The Interplay Between Push Factors and Transformational Leadership. . . 331

5
4.5 Low
4 TL
3.5 High
TOI/ILM

TL
3
2.5
2
1.5
1
Low WFC High WFC

Fig. 7 Simple slope plot for moderating effect of transformational leadership

interpret it as follows; the relationship between WFC and TOI is 0.426 for an average
level of TL. For a higher level of TL, the relationship between WFC and TOI
increases by the size of the interaction term from 0.437 to 0.555 (i.e.,
0.437 + 0.118 = 0.555) because the slope is steep. Similarly, for the lower level
of TL, the relationship between WFC and TOI decreases by the size of the interac-
tion term from 0.437 to 0.319 (i.e., 0.437–0.118 = 0.319) because the slope is not
steeper. Conversely, the results indicated that TL fails to moderate the association
between coworkers’ relationship and ILM [β = 0.019, p = 0.371, C. I (-0.077;
0.115)] (Fig. 6); thus, Hypothesis 3a was insigniﬁcant.

4.5 Importance-Performance Map Analysis (IPMA)

of the Path Modeling Results

IPMA is generally helpful in generating further ﬁndings and conclusions in

PLS-SEM results. It identifies predecessor variables with relatively high importance
to the target variable TOI/ILM, but which achieve a relatively low performance for
explaining ILM and, consequently, for maintaining employee retention. Since all
independent variables of this study were measured reflectively, IPMA was limited to
the structural model. Figure 8 shows each latent variable’s performance and impact
on ILM. The IPMA is divided into four areas, with importance and performance
values below and above average (Ringle and Sarstedt 2016). In each IPMA, con-
centration is on the lower right corner to enhance improvement because items plotted
in this area have high importance with low performance.
Figure 8 shows that WFC is highly relevant to ILM due to its significant
influence. Therefore, government agencies need to maintain the good performance
of WFC in organizations by ensuring that effective and efficient WFC programs and
policies are implemented. Moreover, government agencies ought to take initiatives
to possible overkill or restrain coworkers’ relationship because though it is
332 R. Massae et al.

Fig. 8 Importance-performance map analysis

negatively signiﬁcant, it is not important in explaining employee ILM and their

ultimate retention in working organizations.
The overall results indicate that both CR and WFC are main predecessors of
employees’ ILM, and TL is a signiﬁcant pull-to-stay factor, government agencies
must have appropriate training and development programs to empower their leaders
at all levels to acquire TL behavior, which is very important in employee retention.
This is so because previous studies reveal that organizational issues such as trau-
matic work environments and insufﬁcient human resource management practices are
the key determinants of WFC and employee TOI (Boamah and Laschinger 2016;
Batt and Valcour 2003). Consequently, if the impact of WFC is being minimized by
the presence of TL as a moderator variable, other push factors will be indirectly
moderated.

5 Discussion

The primary goal of this chapter was twofold, to examine the direct impact of push
factors, coworkers’ relationship (CR), and WFC on ILM and to analyze the extent to
which TL moderates the influence of push factors on ILM in the public sector. The
chapter was guided by SET and Herzberg’s Two-Factor Theory.
Concerning the first objective (CR→TOI/ILM), the statistical results revealed
that “CR has a significant negative influence on ILM.” The results were similar to
other studies which confirm that employees who experience healthy interpersonal
relations in the form of cooperative work groups, task interdependence, good
The Interplay Between Push Factors and Transformational Leadership. . . 333

employee relationship, and supportive leaders have conveyed lower TOI (Azam
et al. 2017; Jabutay and Rungruang 2020). The results are consistent with the
previous studies (e.g., Haldorai et al. 2019) which opine that the CR is significantly
negatively related to TOI.
The findings draw attention to the management of CR to reduce employees’ ILM.
Regarding indicators, empirical findings conform to Emerson (1976) that, when an
individual receives favorable treatment from another, he/she feels indebted to repay
in kind.
Statistical results relating to H2 (WFC→TOI/ILM) indicated that “WFC is sig-
nificant and positively related to ILM.” The findings are consistent with previous
studies by Haldorai et al. (2019) and Yildiz et al. (2021), who showed that WFC
significantly positively related to ILM.
The findings are in accord with Adisa et al. (2016) and Putra and Suwandana
(2020), who assert that employees experiencing work intrusion into their family
lives experience broken marriage/family, unhappy workforce, and young misbehav-
iors, stimulating to seek employment elsewhere.
The statistical results for H3a (Moderating Effect 1→TOI/ILM) showed that TL
does not significantly moderate the relationship between CR and ILM. The results
are consistent with Waldman et al. (2015) who asserted that, despite the direct
loyalty supporting effect of TL, followers are still exposed to some push forces for
leaving, in this case, CR. However, H3b (Moderating Effect 2→TOI/ILM) was found
to be significant: empirical findings established that “the higher the TL, the weaker
the influence of WFC on ILM.” The results are in agreement with Tse et al. (2013),
indicating that SET helps to explain how and why TL should be considered an
essential “pull-to-stay” factor. Thus, integration of WFC with supportive, caring, and
empathetic leadership such as TL is likely to enhance employee organizational
commitment and lessen withdrawal behaviors (Bass et al. 2003; Avolio et al. 1991).

6 Conclusion

This chapter sought to shed light on how leadership affects followers’ responses to
push factors, impacting ILM in the public sector. Specifically, the researchers
developed and tested hypotheses about how TL shapes followers’ reactions to
push factors and how these responses relate to ILM behaviors. This chapter repre-
sents empirical efforts validating methodical investigation of the causal relationship
between push factors (CR and WFC) and ILM, with moderating effect of TL
(Fig. 9).
After testing the research model using survey data, it was found that CR is
significantly negatively related to ILM, and WFC is significantly positively related
to ILM consistently with previous studies. However, the study did not find signif-
icant results on the moderating effect of TL, CR, and ILM. The study findings are
consistent with (Waldman et al. 2015). Moreover, the study findings reveal that
TL*WFC’s interactive effect will reduce ILM in the public sector.
334 R. Massae et al.

Push Factors

Co-workers’
Relationship H
Inter-organiza-
tional Labor Mo-
bility (Turnover
Intention)
Work-family
H2
Conflict H3 H3

Transformational
Leadership

Fig. 9 Modiﬁed push forces for interorganizational labor mobility in Tanzania public sector

7 Implications for Research

This chapter will advance knowledge in the following ways: Firstly, this chapter is
most appropriate due to limited research in TL (Tse et al. 2013; Sun and Wang
2017). Scholars have developed a number of theoretical frameworks to forecast
employee ILM in the working context (de Luis Carnicer et al. 2004; Haldorai et al.
2019). However, none of the existing studies have integrated TL, push-to-leave
factors, and ILM. Hence, the chapter findings reveal that TL can moderate the
association between WFC and ILM in the Tanzania public sector. Hence, it is
imperative for management to employ TL as an efficient tool to address the conse-
quence of WFC.
Secondly, the researchers tested the research model in the context of public
agency employees in Tanzania to expand the applicability of TL as a moderator
variable. The results were similar to previous scholars who confirmed that TL can act
as an essential pull-to-stay factor, lessening the impact of both push and pull to leave
factor, increase the generalizability (Oh and Chhinzer 2021; Waldman et al. 2015).
Consequently, the researchers gain deeper insight into the interaction effect of TL
and push factors on ILM in public sector, hence extending the existing labor mobility
model.
The Interplay Between Push Factors and Transformational Leadership. . . 335

8 Research Limitations and Future Direction

The findings of this chapter indicated that the CR, WFC, and TL interactively
explain 23.3% of ILM variation. This means that other factors (76.7%) influence
ILM; however, they have been not accounted for in this study. Hence, the findings
call for further studies to investigate the influence of both push and pull factors on
leaving.
Furthermore, future studies may consider doing qualitative research design to
explore other factors influencing ILM in the public sector. Lastly, it is also worth for
future study to do a comparative study between the public and private sectors to
ascertain similarities and differences.

References

Adisa TA, Osabutey EL, Gbadamosi G (2016) Understanding the causes and consequences of
work-family conflict: an exploratory study of Nigerian employees. Empl Relat 38(5):770–778
Agarwal R, Bidwell M, Cirillo B, Tzabbar D (2020) Two perspectives on employee mobility: a
conversation between Rajshree Agarwal and Matthew Bidwell. In: Tzabbar D, Cirillo B (eds)
Employee inter-and intra-firm mobility (Advances in strategic management), vol 41. Emerald
Publishing Limited, Bingley, pp 37–54
Akgündüz YE, Aldan A, Bağır YK, Torun H (2019) Job mobility in Turkey. Central Bank Rev
19(3):83–91
Alshmemri M, Shahwan-Akl L, Maude P (2017) Herzberg’s two-factor theory. Life Sci J 14(5):
12–16
Avolio BJ, Waldman DA, Yammarino FJ (1991) Leading in the 1990s: the four I′ s of transfor-
mational leadership. J Eur Ind Train 15(4):9–16
Azam S, Rafique A, Butt P (2017) Investigating the role of coworkers’ relationship on job turnover
intention with mediating effects of job satisfaction for city traffic police. Sci Int 29(1):303–306
Bass BM (1985) Leadership and performance beyond expectations. Free Press, New York, NY
Bass BM, Avolio BJ, Jung DI, Berson Y (2003) Predicting unit performance by assessing
transformational and transactional leadership. J Appl Psychol 88(2):207–218
Batt R, Valcour PM (2003) Human resources practices as predictors of work-family outcomes and
employee turnover. Ind Relat: J Econ Soc 42(2):189–220
Becker J-M, Ringle CM, Sarstedt M, Völckner F (2015) How collinearity affects mixture regression
results. Mark Lett 26(4):643–659
Blau P (1964) Exchange and power in social life. John Wiley and Sons, New York
Boamah SA, Laschinger H (2016) The influence of areas of work-life fit and work-life interference
on burnout and turnover intentions among new graduate nurses. J Nurs Manag 24(2):164–174
Bryant PC, Allen DG (2013) Compensation, benefits and employee turnover: HR strategies for
retaining top talent. Compens Benefits Rev 45(3):171–175
Carlson DS, Kacmar KM, Williams LJ (2000) Construction and initial validation of a
multidimensional measure of work–family conflict. J Vocat Behav 56(2):249–276
Chiat LC, Panatik SA (2019) Perceptions of employee turnover intention by Herzberg’s motivation-
hygiene theory: A systematic literature review. J Res Psychol 1(2):10–15
Chin WW (1998) The partial least squares approach to structural equation modeling. Mod Methods
Bus Res 295(2):295–336
336 R. Massae et al.

Chowdhury M (2016) A quantitative study on push, pull and personal factors affecting employees’
turnover intentions: A case of nationalized commercial banks (NCBs) in Bangladesh. Inf Knowl
Manag 6(9):1–14
Cohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum
Associates, Hillsdale
Creswell JW (2012) Educational research: planning, conducting and evaluating quantitative and
qualitative research, 4th edn. Pearson Education, Inc, Boston
Cropanzano R, Mitchell MS (2005) Social exchange theory: an interdisciplinary review. J Manag
31(6):874–900
Cropanzano R, Anthony EL, Daniels SR, Hall AV (2017) Social exchange theory: A critical review
with theoretical remedies. Acad Manag Ann 11(1):479–516
de Luis Carnicer MP, Sanchez AM, Pérez MP, Jiménez MJV (2004) Analysis of internal and
external labor mobility: A model of job-related and non-related factors. Pers Rev 33(2):222–240
Emerson RM (1976) Social exchange theory. Annu Rev Sociol 2(1):335–362
Goodluck C (2009) Competitive advantage in small and medium scale enterprises: A resource-
based analysis of Tanzanian family and non-family firms. PhD Thesis, University of Dar es
Salaam, Dar es Salaam, Tanzania
Greenhaus JH, Parasuraman S, Granrose CS, Rabinowitz S, Beutell NJ (1989) Sources of work-
family conflict among two-career couples. J Vocat Behav 34(2):133–153
Griffeth RW, Hom PW, Gaertner S (2000) A meta-analysis of antecedents and correlates of
employee turnover: update, moderator tests, and research implications for the next millennium.
J Manag 26(3):463–488
Hain CA, Francis L (2004) Development and validation of a coworker relationship scale. In: 65th
Annual convention, Canadian Psychological Association, Newfoundland, Canada
Hair J, Black WC, Babin BJ, Anderson RE (2010) Multivariate data analysis, 7th edn. Prentice Hall,
Upper Saddle River, NJ
Hair J, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract 19(2):
139–152
Hair J, Hult TM, Ringle CM, Sarstedt M (2017) A primer on partial least squares structural equation
modeling (PLS – SEM), 2nd edn. SAGE Publications, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Haldorai K, Kim WG, Pillai SG, Park TE, Balasubramanian K (2019) Factors affecting hotel
employees’ attrition and turnover: application of pull-push-mooring framework. Int J Hosp
Manag 83(1):46–55
Hall DT (1996) Protean careers of the 21st century. Acad Manag Perspect 10(4):8–16
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Herzberg F, Mausner B, Snydermann B (1959) The motivation to work. Wiley, New York
Jabutay FA, Rungruang P (2020) Turnover intent of new workers: social exchange perspectives.
Asia-Pac J Bus Admin 13(1):60–79
Jackson-Smith D, Flint CG, Dolan M, Trentelman CK, Holyoak G, Thomas B, Ma G (2016)
Effectiveness of the drop-off/pick-up survey methodology in different neighborhood types. J
Rural Soc Sci 31(3):35–67
Jakobsen M, Jensen R (2015) Common method bias in public management studies. Int Public
Manag J 18(1):3–30
Jensen UT, Andersen LB, Bro LL, Bøllingtoft A, Eriksen TLM, Holten A-L et al (2019) Concep-
tualizing and measuring transformational and transactional leadership. Adm Soc 51(1):3–33
Joshi H, Kulkarni H, Deshpande S (2012) Multicollinearity diagnostics in statistical modeling and
remedies to deal with it using SAS. Pharmaceutical Users Software Exchange 1:1–34
Jung HS, Yoon HH (2013) The effects of organizational service orientation on person–organization
fit and turnover intent. Serv Ind J 33(1):7–29
Kenny DA (2016) Moderation. http://davidakennynet/cm/moderationhtm. Accessed 3 Jul 2021
The Interplay Between Push Factors and Transformational Leadership. . . 337

Kim TG, Lee JK, Lee JH (2013) Do interpersonal relationships still matter for turnover intention? A
comparison of South Korea and China. Int J Hum Resour Manag 24(5):966–984
Kock N (2015) Common method bias in PLS-SEM: A full collinearity assessment approach. Int J
e-Collaboration 11(4):1–10
Liljegren M, Ekberg K (2009) Job mobility as predictor of health and burnout. J Occup Organ
Psychol 82(2):317–329
Linhartová L, Urbancová H (2012) Results of analysis of employee mobility: factors affecting
knowledge continuity. Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis
60(4):235–244
Mbah SE, Ikemefuna CO (2012) Job satisfaction and employees’ turnover intentions in total
Nigeria PLC in Lagos state. Int J Humanit Soc Sci 2(14):275–287
Michael FK (2009) Human resources management practices and employee job satisfaction in
private work organizations: A Tanzanian context. PhD Thesis, University of Dar es Salaam,
Dar es Salaam, Tanzania
Moon K-K, Park J (2019) Leadership styles and turnover behavior in the US Federal Government:
does span of control matter? Int Public Manag J 22(3):417–443
Oh J, Chhinzer N (2021) Is turnover contagious? The impact of transformational leadership and
collective turnover on employee turnover decisions. Leadersh Organ Dev J 42(7):1089–1103
Podsakoff PM, MacKenzie SB, Lee J-Y, Podsakoff NP (2003) Common method biases in behav-
ioral research: a critical review of the literature and recommended remedies. J Appl Psychol
88(5):879–903
Podsakoff PM, MacKenzie SB, Podsakoff NP (2012) Sources of method bias in social science
research and recommendations on how to control it. Annu Rev Psychol 63:539–569
Putra IPKSD, Suwandana IGM (2020) Effect of organizational commitment, job stress and work-
family conflict to turnover intention. Int Res J Manag, IT Social Sci 7(2):30–37
Ramayah T, Cheah J, Chuah F, Ting H, Memon MA (2018) Partial Least Squares Structural
Equation Modeling (PLS-SEM) using smartPLS 3.0: an updated and practical guide to statistical
analysis, 2nd edn. Pearson, Kuala Lumpur Malaysia
Ringle CM, Sarstedt M (2016) Gain more insight from your PLS-SEM results: the importance-
performance map analysis. Ind Manag Data Syst 116(9):1865–1886
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. SmartPLS GmbH, Boenningstedt
Rubenstein AL, Eberly MB, Lee TW, Mitchell TR (2018) Surveying the forest: A meta-analysis,
moderator investigation, and future-oriented discussion of the antecedents of voluntary
employee turnover. Pers Psychol 71(1):23–65
Sarstedt M, Ringle CM, Hair JF (2017) Partial least squares structural equation modeling. Hand-
book Mark Res 26(1):1–40
Saunders M, Lewis P, Thornhill A (2016) Research methods for business students, 7th edn. FT
Prentice Hall, Harlow, England
Shmueli G, Ray S, Estrada JMV, Chatla SB (2016) The elephant in the room: predictive perfor-
mance of PLS models. J Bus Res 69(10):4552–4564
Sousa-Poza A, Henneberger F (2004) Analyzing job mobility with job turnover intentions: an
international comparative study. J Econ Issues 38(1):113–137
Steenackers K, Guerry M-A (2016) Determinants of job-hopping: an empirical study in Belgium.
Int J Manpow 37(3):494–510
Sun R, Wang W (2017) Transformational leadership, employee turnover intention, and actual
voluntary turnover in public organizations. Public Manag Rev 19(8):1124–1141
Tang HK (1998) An inventory of organizational innovativeness. Technovation 19(1):41–51
Tehseen S, Ramayah T, Sajilan S (2017) Testing and controlling for common method variance: A
review of available methods. J Manag Sci 4(2):142–168
Tett RP, Meyer JP (1993) Job satisfaction, organizational commitment, turnover intention, and
turnover: path analyses based on meta-analytic findings. Pers Psychol 46(2):259–293
Trottier T, Van Wart M, Wang X (2008) Examining the nature and significance of leadership in
government organizations. Public Adm Rev 68(2):319–333
338 R. Massae et al.

Tse HHM, Huang X, Lam W (2013) Why does transformational leadership matter for employee
turnover? A multi-foci social exchange perspective. Leadersh Q 24(5):763–776
URT (1999) Public service management and employment policy. President Office-Public Service
Management, Dar es Salaam
Waldman DA, Carter MZ, Hom PW (2015) A multilevel investigation of leadership and turnover
behavior. J Manag 41(6):1724–1744
Walsh JP, Ashford SJ, Hill TE (1985) Feedback obstruction: the influence of the information
environment on employee turnover intentions. Hum Relat 38(1):23–46
Wong IA, Wan YKP, Gao JH (2017) How to attract and retain generation Y employees? An
exploration of career choice and the meaning of work. Tour Manag Perspect 23:140–150
Wynen J, Op de Beeck S, Hondeghem A (2013) Interorganizational mobility within the US federal
government: examining the effect of individual and organizational factors. Public Adm Rev
73(6):869–881
Yamane T (1973) Statistics; an introductory analysis, 2nd edn. Harper & Row, New York
Yildiz B, Yildiz H, Ayaz Arda O (2021) Relationship between work–family conflict and turnover
intention in nurses: A meta-analytic review. J Adv Nurs 77(8):3317–3330
www.estabs.go.tz
Integrated Leadership: Assessing
an Integrated Principal Leadership
Practices Construct

Ahmed Mohamed, Ahmad Zabidi Abdul Razak, and Zuraidah Abdullah

1 Introduction

Principals practice various leadership roles based on situations and policy enact-
ment. Thus, principal leadership continues to evolve promptly in response to new
demands of the school environment (Hallinger and Heck 1998). The terms “principal
leadership” or “principal leadership practice” have become prevalent terms in the
context of educational leadership. Eventually, principal leadership (Hallinger and
Heck 1998; Sebastian and Allensworth 2012) and principal leadership practices
(Perera 2015) were conceptualized as constructs. Hence, recent literature on leader-
ship in educational context move toward integrated leadership since single leader-
ship style is insufficient to address complexities of present-day school settings. In
many instances, integrated leadership comprises transformational leadership and
instructional leadership in an educational context.
Integration of leadership styles establishes higher-order models that are complex
in nature (Thien et al. 2019). However, developing a hierarchical construct could
reduce the model complexity and issues of multicollinearity (Hair et al. 2017).
Hence, there is a need to develop an accurate measurement model of “integrated
principal leadership practices” to minimize the issue of measurement model
misspecification in leadership research. Thus, the purpose of the chapter is to
validate a reflective–formative model of integrated principal leadership practices

A. Mohamed (✉)
Faculty of Education, The Maldives National University, Malé, Maldives
e-mail: [email protected]
A. Z. A. Razak · Z. Abdullah
Faculty of Education, University of Malaya, Kuala Lumpur, Malaysia
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 339

(IPLP) to assist educational leadership researchers who employ PLS-SEM in their

research.

2 Background

At the beginning of the new century, researchers have called for an integrated model
of leadership that combines transformational leadership and shared instructional
leadership (Hallinger 2003; Marks and Printy 2003). Over the years, transforma-
tional leadership and instructional leadership are the two predominant leadership
approaches in educational context (Hallinger 2003). Thus, principals use both the
leadership styles collectively for capacity development and school improvement.
Principal leadership practices can be defined as “actions taken by principals to
influence people, processes, and organizational structures” (Camburn et al. 2010,
p. 714). However, conceptualization of principal leadership practices varies with
leadership practices taken into consideration (Lai 2015; Sebastian and Allensworth
2012).
Transformational leadership and instructional leadership are two distinct first-
order constructs based on the conceptualization and measures of the constructs.
Subsequently, it could be proposed that two first-order constructs of transformational
leadership and instructional leadership generate the second-order reflective–forma-
tive IPLP. Proposed conceptual model of second order IPLP is shown in Fig. 1.

3 Methodology

This chapter employed a cross-sectional survey method. The targeted population of

the study was teachers working in government schools of Maldives. After data
cleaning, a sample of 375 respondents were used. Responses were received from

Nine Transforma-
Indicators tional Lead-
ership
Integrated
Principal
Leadership
Practices

Five Instructional
Indicators Leadership

Fig. 1 The proposed second order integrated principal leadership practices construct
Integrated Leadership: Assessing an Integrated Principal. . . 341

32 government schools. Both the teachers and schools were selected using random
sampling.
The instrument used for the study was transformational school leadership survey
based on core leadership practices (Leithwood 2012). A survey questionnaire was
sent to the respondents with the approval from all the concerned institutions.
Participation in the survey was voluntary and anonymous. Prior to the actual data
collection, a pilot test was conducted to do the exploratory factor analysis.
Partial least square-structural equation modelling (PLS-SEM) was employed to
assess and validate the IPLP. PLS-SEM is suitable for assessing complex models,
including reflective and formative constructs (Hair et al. 2022; Hair et al. 2019;
Sarstedt et al. 2019). The study used SmartPLS 3 software to perform the data
analysis. The analysis involved assessment of first-order reflective measurement
model and assessment of second-order formative measurement model. The assess-
ment of reflective measurement model involves internal consistency, convergent
validity, and discriminant validity. All the three were examined. In the next step, the
measurement model was evaluated after generating second-order construct, Inte-
grated Principal Leadership Practices (IPLP). To create a second order construct, a
two-stage approach recommended by Becker et al. (2012) was used, as the construct
is a higher order formative construct. In the assessment of formative measurement
model, these three steps were tested: (1) assessing convergent validity of formative
measurement model; (2) addressing collinearity issues; and (3) assessing the (statis-
tical) significance and relevance of the formative indicators. A global item was used
as a reflective measure in the redundancy analysis.

4 Results

The ﬁrst-order reﬂective constructs, namely, (1) transformational leadership and

(2) instructional leadership were involved to establish integrated principal leadership
practices (IPLP). The two first-order reflective constructs were evaluated by consid-
ering the criteria for the assessment of the reflective measurement model.
Results showed that all the item loadings for the first-order reflective constructs,
Instructional Leadership and Transformational Leadership, have factor loadings
above 0.708, as shown in Table 1. As a rule of thumb, item loadings should be
0.708 and above (Hair et al. 2017). Results also indicated that the composite
reliability for the reflective constructs was within acceptable range after deleting
certain items from the model (i.e., 0.948 and 0.950 for instructional leadership and
transformational leadership, respectively). The deleted items were ISLIL4, ISLTL4,
ISLTL6, and ISLTL12. Thus, composite reliability for the constructs achieved
acceptable internal consistency reliability. In addition, Table 1 shows that AVE for
both of these constructs is higher than the threshold/cut-off value of 0.5. Hence, the
results revealed that convergent validity was ensured for the first-order constructs.
Once internal consistency and convergent validity were established, discriminant
validity was assessed. Discriminant validity is the extent to which each construct is
342 A. Mohamed et al.

Table 1 Internal consistency reliability and convergent validity of reﬂective constructs

First order construct Items Loadings AVE CR
Instructional leadership (IPLPIL) IPLPIL1 0.885 0.784 0.948
IPLPIL2 0.884
IPLPIL3 0.902
IPLPIL5 0.878
IPLPIL6 0.879
Transformational leadership (IPLPTL) IPLPTL1 0.786 0.680 0.950
IPLPTL2 0.830
IPLPTL3 0.794
IPLPTL5 0.858
IPLPTL7 0.865
IPLPTL8 0.802
IPLPTL9 0.860
IPLPTL10 0.819
IPLPTL11 0.801

Table 2 HTMT criterion of ﬁrst order constructs

IPLPIL IPLPTL
IPLPIL
IPLPTL 0.882
C.I.95 (0.840, 0.916)

distinct from other constructs in a model (Chin 2010). To evaluate discriminant

validity, the heterotrait–monotrait (HTMT) criterion can be used (Henseler et al.
2015; Ramayah et al. 2018). In terms of HTMT, Table 2 shows the HTMT ratio
(0.882) fulfills the criterion of HTMT 0.90 (Gold et al. 2001). Thus, the result
revealed that discriminant validity is established. In addition to HTMT ratio, the
result of HTMT inference in Table 2 also shows that the confidence interval does not
show a value of 1 on any of the constructs which confirms discriminant validity.
As evident in Table 3, the formative construct of integrated principal leadership
practices (IPLP) yielded path coefficient of 0.73, more than threshold value of 0.70
(Hair et al. 2017), thus the formative construct (IPLP) has achieved sufficient degree
of convergent validity. According to Table 3, VIF values of all the indicators
(IPLPIL and IPLPTL) for the formative construct (IPLP) were below the threshold
value of 5 (Hair et al. 2017), thus integrated principal leadership construct satisfies
the VIF values. This concludes that collinearity is not an issue for the estimation of
the PLS path model as collinearity does not reach a critical level in any of the
formative indicators. In the third step, the significance and relevance of the outer
weights of the formative constructs were assessed. The results in Table 3 showed
that weights of both formative indicators (IPLPIL and IPLPTL) were significant.
Therefore, the formative measurement model of integrated principal leadership
practices with two formative indicators was achieved.
Integrated Leadership: Assessing an Integrated Principal. . . 343

Table 3 Measurement properties for formative constructs

Higher order Lower order Convergent t-value
construct construct validity Weights VIF weights Sig.
IPLP IPLPIL 0.73 0.366 3.155 3.085** 0.002
IPLPTL 0.676 3.155 5.951* 0.000
Notes: > 1.96**

5 Conclusion and Implications

In relation to this chapter, a higher order model of integrated principal leadership

practices was conceptualized and validated in school context using measurement
model assessments. Results presented in the chapter have learned that integrated
leadership must be studied as a reﬂective–formative hierarchical component model
in future leadership research. For theoretical implications, results supported the
conceptualization of IPLP that was established by transformational leadership and
instructional leadership. This chapter contributes to the body of knowledge on
school leadership. In addition, the ﬁndings informed on added knowledge to inte-
grated leadership that principals’ leadership practices can be perceived as a hierar-
chical model. One of the important practical implications of the chapter is to use
integrated principal leadership practices construct in empirical studies to examine the
effect of this construct on other factors in educational settings.
The current chapter is limited to a single instrument that captures both instruc-
tional and transformational leadership. This limitation could be addressed in future
research using a separate set of measures, including the most widely used instruc-
tional leadership questionnaires. Additionally, various other dimensions of both
transformational leadership and instructional leadership could be included in future
research to generate a third-order integrated principal leadership practices construct.

Acknowledgments I acknowledge the ﬁnancial support provided by the Maldives National

University (MNU) to present this paper at the 2022 International Conference on Partial Least
Squares Structural Equation Modeling (PLS-SEM).

References

Becker JM, Klein K, Wetzels M (2012) Hierarchical latent variable models in PLS-SEM: guidelines
for using reﬂective-formative type models. Long Range Plan 45(5–6):359–394
Camburn EM, Spillane JP, Sebastian J (2010) Assessing the utility of a daily log for measuring
principal leadership practice. Educ Adm Q 46(5):707–737
Chin WW (2010) Bootstrap cross-validation indices for PLS path model assessment. In: Esposito
Vinzi V, Chin W, Henseler J, Wang H (eds) Handbook of partial least squares. Springer
handbooks of computational statistics. Springer, Berlin, Heidelberg, pp 83–97
Gold AH, Malhotra A, Segars AH (2001) Knowledge management: an organizational capabilities
perspective. J Manag Inf Syst 18(1):185–214
344 A. Mohamed et al.

Hair JF, Hult GTM, Ringle C, Sarstedt M (2017) A primer on partial least squares structural
equation modeling (PLS-SEM), 2nd edn. Sage, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Hult GTM, Ringle C, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks, CA
Hallinger P (2003) Leading educational change: reﬂections on the practice of instructional and
transformational leadership. Camb J Educ 33(3):329–352
Hallinger P, Heck RH (1998) Exploring the Principal’s contribution to school effectiveness:
1980-1995. Sch Eff Sch Improv 9(2):157–191
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Lai E (2015) Enacting principal leadership: exploiting situated possibilities to build school capacity
for change. Res Pap Educ 30(1):70–94
Leithwood K (2012) Core practices: the four essential components of the leader’s repertoire. In:
Leithwood K, Louis KS (eds) Linking leadership to student learning. Jossey-Bass, San
Francisco, CA
Marks HM, Printy SM (2003) Principal leadership and school performance: an integration of
transformational and instructional leadership. Educ Adm Q 39(3):370–397
Perera CJ (2015) Principal leadership practices and teacher collegiality in Malaysian high
performing schools. PhD Thesis, University of Malaya, Kuala Lampur, Malaysia. http://
studentsrepo.um.edu.my/5993/. Accessed 10 Oct 2021
Ramayah T, Cheah J, Chuah F, Ting H, Memon MA (2018) Partial least squares structural equation
modeling (PLS-SEM) using SmartPLS 3.0 - an updated and practical guide to statistical
analysis, 2nd edn. Pearson, Kuala Lumpur
Sarstedt M, Hair JF, Cheah J-H, Becker J-M, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J AMJ 27(3):197–211
Sebastian J, Allensworth E (2012) The inﬂuence of principal leadership on classroom instruction
and student learning: a study of mediated pathways to learning. Educ Adm Q 48(4):626–663
Thien LM, Rasoolimanesh SM, Ng AYM, Ramayah T (2019) Conceptualizing and assessing an
integrated hierarchical leadership construct in education context. J Appl Struct Equ Model 3(1):
15–30
Exploring the Interrelationship Among
Management Accounting Systems,
Decentralization, and Organizational
Performance

Elsa Pedroso and Carlos F. Gomes

1 Introduction

Organizations competing in the global market face major challenges. These chal-
lenges relate to several factors, including the efﬁcient use of resources and an
uncertain external environment. This uncertainty becomes much more challenging
for small and medium enterprises (SMEs). Companies sometimes decentralize
decision-making power to address this uncertainty and seek more market informa-
tion. These approaches aim to bring relevant information to managers at all organi-
zational levels so they can respond quickly and effectively to changes in the external
environment. They can thus increase their departments’ performance and conse-
quently improve organizational performance. In this process, management account-
ing systems (MAS) play a key role in managing and sharing this information
throughout the organization. As such, decentralization of decision-making may
inﬂuence the relationship between the use of information provided by MAS and
the organizational performance of SMEs.

2 Background

The results presented by literature relating the relationship between MAS and
organizational performance are not consistent (Cadez and Guilding 2008;
Krumwiede et al. 2008; Harrison 2009; Tontiset and Ussahawanitchakit 2010;
Hoque 2011; Tuanmat and Smith 2011; Ismail et al. 2018; Jariya and Velnampy

E. Pedroso (✉) · C. F. Gomes

University of Coimbra, CeBER, Faculty of Economics, Coimbra, Portugal
e-mail: [email protected]; [email protected]

© Springer Nature Switzerland AG 2023 345

Decentralization
of decisions

Integration

Aggregation

Management ac- Organizational

counting system* performance
Scope

*Second-order reflective construct

Timeliness

Fig. 1 Conceptual model

2022). These results could suggest the introduction of a variable that moderates this
relationship (Baron and Kenny 1986).
In a highly competitive market, it is necessary to respond quickly to changes that
are taking place dynamically. These responses are mainly operational. Therefore,
information from the market is required for decisions made at all hierarchical levels.
Thus, increasing decentralization will allow a faster response to the market.
Decentralization of the decision process can inﬂuence the relationship between
MAS and organizational performance, improving access to information for man-
agers at intermediate levels that will lead to a more effective response from these
managers and thus increase organizational performance.
In this context, this chapter aims to study the relationship between the use of MAS
information and the organizational performance of SMEs. In addition, we also
analyze the inﬂuence of the decentralization of decision-making on this relationship.
The conceptual model is presented in Fig. 1.

3 Methodology

The data was collected through an online survey of Portuguese SMEs for this
research. The names and addresses of 1500 SMEs were obtained from Informa
DB, which belongs to the Dun and Bradstreet Worldwide Network. All these
enterprises were contacted by phone to explain the objective of the research study.
Twelve enterprises, despite several attempts, never answered our calls. Ninety-three
enterprises declined to collaborate in this research study for several reasons, such as
no longer operating and a lack of autonomy to respond to the questionnaire.
Exploring the Interrelationship Among Management Accounting Systems,. . . 347

Table 1 Operationalization of the constructs

Constructs Scale Sources
Management The extent of use of MAS informa- Pedroso and Gomes (2020)
accounting tion on a 7-point scale could range
systems from “never” (1) to “always” (7).
Organizational Evaluation of organizational perfor- Govindarajan (1984), Cadez and
performance mance for each of the performance Guilding (2008), Hoque and James
measures included in the construct on (2000), Harrison (2009), Hoque
a 7-point scale, which could range (2011)
from “unacceptable” (1) to “excel-
lent” (7).
Decentralization Rate the extent of the company’s Gordon and Narayanan (1984),
of decisions decentralization on a 7-point scale Abdel-Kader and Luther (2008),
that could range from “very low” Soobaroyen and Poorundersing
(1) to “very high”(7). (2008)

Therefore, an email explaining the purpose of the research and containing the link to
the online survey was sent to 1407 Portuguese SMEs. A total of 255 usable
responses were obtained from Chief Financial Officers (CFOs), representing a
response rate of 18.12%. This response rate cannot be considered high, but is in
line with other studies in the same research area.
The research instrument used in this chapter was designed based on an extensive
literature review. During the first phase of the questionnaire design, it was translated
and adapted to the Portuguese business environment. In the second phase, the
instrument was submitted to a panel of experts from several organizations. During
this phase, particular attention was given to the use of terminology consistent with
the background of the survey participants.
The final version of the constructs used in this research included 31 items
representing the observed variables. All these observed variables were measured
on a Likert scale with a range of 1 to 7 (Table 1).
Partial least squares-structural equation modeling (PLS-SEM) was used to ana-
lyze the data.
All the analyses were performed using the IBM-SPSS Statistics version 25 and
SmartPLS 3.3.9 (Ringle et al. 2015) and followed procedures suggested in the
literature (Hair et al. 2013; Hair et al. 2019).

4 Results and Discussion

4.1 Measurement Model

The ﬁrst step was assessing the constructs’ reliability. Almost all outer loadings were
above 0.708, the threshold recommended by the literature (Hair et al. 2019). Only
two measurement items were slightly lower than this threshold (Table 2).
348 E. Pedroso and C. F. Gomes

Table 2 Validity and reliability constructs

Construct items Loading CR AVE Cronbach’s alpha
Scope 0.905 0.657 0.870
SCO01 0.838
SCO02 0.816
SCO03 0.805
SCO04 0.796
SCO05 0.797
Timeliness 0.914 0.727 0.875
TIM01 0.833
TIM02 0.866
TIM03 0.859
TIM04 0.853
Aggregation 0.916 0.608 0.892
AGG01 0.794
AGG02 0.772
AGG03 0.770
AGG04 0.822
AGG05 0.787
AGG06 0.744
AGG07 0.767
Integration 0.936 0.829 0.897
INT01 0.891
INT02 0.929
INT03 0.911
Organizational performance 0.894 0.548 0.862
ORP01 0.789
ORP02 0.746
ORP03 0.766
ORP04 0.793
ORP05 0.742
ORP06 0.656
ORP07 0.679
Decentralization of decisions 0.912 0.674 0.880
DEC01 0.781
DEC02 0.835
DEC03 0.859
DEC04 0.843
DEC05 0.783

Additionally, the Cronbach’s alpha values and the Composite reliability

(CR) values obtained for each construct exceeding 0.7 indicated sufﬁcient construct
reliability. All the average variance extracted (AVE) values were higher than the
recommended threshold of 0.5. Based on these results, the items with outer loadings
Exploring the Interrelationship Among Management Accounting Systems,. . . 349

Table 3 Discriminant validity

Agg Dec Int OP Sco Tim
Aggregation –
Decentralization 0.436
Integration 0.879 0.444
Organizational performance 0.372 0.325 0.286
Scope 0.771 0.413 0.628 0.302
Timeliness 0.727 0.243 0.677 0.312 0.637 –
Notes: None of the correspondent bootstrap conﬁdence intervals includes the value 1

Table 4 Structural model results

95% conﬁdence
Path relationship Path coefﬁcient interval f2
Decentralization - > Organizational performance 0.180** [0.061; 0.325] 0.032
MAS - > Organizational performance 0.256*** [0.144; 0.385] 0.065
Moderating effect of decentralization 0.121** [0.008; 0.226] 0.019
R2 Q2
Organizational performance 0.160 0.079
Notes: ***p < 0.001, **p < 0.05, *p < 0.10

slightly below 0.708 were maintained, as their exclusion did not improve the AVE
and CR values (Hair et al. 2013).
Regarding discriminant validity, the heterotrait–monotrait (HTMT) ratio of cor-
relations was used (Hair et al. 2017; Usakli and Kucukergin 2018). All values for
HTMT are less than 0.90 (Table 3), which means that the discriminant validity of the
constructs is assured. Regarding the MAS construct, all the loadings of the first-order
constructs on the second-order constructs are significant ( p < 0.001) and above
0.765. They indicate that MAS can be measured as the second-order construct
proposed by Pedroso and Gomes (2020), reflecting the four dimensions of informa-
tion characteristics: scope, timeless, aggregation, and integration.
The common method variance was verified with Harman’s single-factor test by
conducting an exploratory factor analysis (Podsakoff et al. 2003). The results of this
test show that the first factor only accounts for 26.28% of the total variance, which
means that the data has no common method variance issues.

4.2 Structural Model

The results of the structural model using a bootstrapping procedure with a

resampling of 5000 are presented in Table 4. All the paths of relationships are
positive and significant. The values of f 2 are also positive and follow a similar
rank order of the path coefficients, which means that large significant path values
350 E. Pedroso and C. F. Gomes

Table 5 PLSpredict results PLS-SEM LM PLS-SEM - LM

Item RMSE Q2predict RMSE RMSE
ORP01 0.947 0.103 0.994 -0.047
ORP02 0.979 0.071 1.021 -0.043
ORP03 0.912 0.081 0.952 -0.040
ORP04 0.823 0.083 0.857 -0.034
ORP05 0.816 0.064 0.854 -0.037
ORP06 1.088 0.081 1.139 -0.051
ORP07 1.030 0.031 1.089 -0.060

correspond to large effect sizes. In addition, Q2 is positive, which means that the
model has predictive relevance (Hair et al. 2013).
After analyzing the graphs regarding PLS-SEM prediction errors, we found that
their distribution does not deviate much from symmetry. For this reason, we based
our assessment of the predictive power of our model on the RMSE. The results of
PLSpredict, based on ten samples and ten repetitions, are presented in Table 5.
Comparing the RMSE values from the PLS-SEM with the linear regression
(LM) benchmark, we found that the PLS-SEM produces lower prediction errors
for all the indicators, which means that our model has high predictive power for
organizational performance (Shmueli et al. 2019).
The results of this chapter validate a multidimensional approach to measuring the
effectiveness of MAS. They also confirm the four dimensions of MAS and show that
these dimensions are distinct and interrelated.
According to the results, using information with the characteristics represented by
the multidimensional approach to MAS positively influences the organizational
performance of SMEs. In addition, the moderation effect of decentralizing
decision-making on this relationship was significant. This means that increasing
the level of decentralization will increase the influence of the MAS on organizational
performance.

5 Conclusions and Implications

This chapter analyzes the moderating effect of decentralization on the relationship

between management accounting systems and organizational performance.
Based on the results, it seems that decentralizing decision processes can enhance
the impact of MAS effectiveness on organizational performance. We highlight the
innovative nature of the multidimensional MAS and the synergies resulting from its
four dimensions of information, which can be created through their balanced devel-
opment. This multidimensional approach should enrich management accounting
knowledge and provide researchers with a valuable tool for measuring the effective-
ness of MAS and its inﬂuence on organizational performance. In addition, it should
Exploring the Interrelationship Among Management Accounting Systems,. . . 351

facilitate the comparison between different studies in the ﬁeld of management

accounting.
Overall, it appears that executives of Portuguese SMEs are effectively using
accounting management information to anticipate market threats. As such, regard-
less of company size and the industry to which they belong, executives use man-
agement accounting information to predict threats and maintain the competitiveness
of their companies in the global market.
This chapter contributes to a better knowledge of the factors that can enhance
MAS effectiveness and its inﬂuence on the performance of SMEs. It also contributes
to executive decision support by providing an instrument for assessing the quality of
information they use to improve their company’s competitiveness.

Acknowledgments: This research has been funded by national funds through FCT – Fundação
para a Ciência e a Tecnologia, I.P., Project UIDB/05037/2020.

References

Abdel-Kader M, Luther R (2008) The impact of ﬁrm characteristics on management accounting

practices: a UK-based empirical analysis. Br Account Rev 40(1):2–27
Baron RM, Kenny DA (1986) The moderator–mediator variable distinction in social psychological
research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 51(6):
1173–1182
Cadez S, Guilding C (2008) An exploratory investigation of an integrated contingency model of
strategic management accounting. Acc Organ Soc 33(7–8):836–863
Gordon LA, Narayanan VK (1984) Management accounting systems, perceived environmental
uncertainty and organization structure: an empirical investigation. Acc Organ Soc 9(1):33–47
Govindarajan V (1984) Appropriateness of accounting data in performance evaluation: an empirical
examination of environmental uncertainty as an intervening variable. Acc Organ Soc 9(2):
125–135
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2013) A primer on partial least squares structural
equation modeling (PLS-SEM). SAGE Publications, Los Angeles
Hair JF, Hollingsworth CL, Randolph AB, Chong AYL (2017) An updated and expanded assess-
ment of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Harrison JL (2009) Untangling the value of information scope: an investigation in retail pharmacies.
J Manag Organ 15(4):470–485
Hoque Z (2011) The relations among competition, delegation, management accounting systems
change and performance: a path model. Adv Account 27(2):266–277
Hoque Z, James W (2000) Linking balanced scorecard measures to size and market factors: impact
on organizational performance. J Manag Account Res 12:1–17
Ismail K, Isa CR, Mia L (2018) Evidence on the usefulness of management accounting systems in
integrated manufacturing environment. Pac Account Rev 30(1):2–19
Jariya AMI, Velnampy T (2022) The effect of management accounting practices on organizational
performance of listed manufacturing sectors in Sri Lanka, with the moderating effect of
complexity of production process. SMART J Bus Manag Stud 18(1):30–37
352 E. Pedroso and C. F. Gomes

Krumwiede KR, Suessmair A, MacDonald J (2008) An exploratory study of the factors affecting
the implementation success of German cost accounting methods. AAA 2008 MAS Meeting
Paper
Pedroso E, Gomes CF (2020) The effectiveness of management accounting systems in SMEs: a
multidimensional measurement approach. J Appl Acc Res 21(3):497–515
Podsakoff PM, MacKenzie SB, Lee J-Y, Podsakoff NP (2003) Common method biases in behav-
ioral research: a critical review of the literature and recommended remedies. J Appl Psychol
88(5):879–903
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. SmartPLS computer program
Shmueli G, Sarstedt M, Hair JF, Cheah JH, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Soobaroyen T, Poorundersing B (2008) The effectiveness of management accounting systems:
evidence from functional managers in a developing country. Manag Audit J 23(2):187–219
Tontiset N, Ussahawanitchakit P (2010) Building successful cost accounting implementation of
electronics manufacturing businesses in Thailand: how do its antecedents and consequences
play a signiﬁcant role? J Acad Bus Econ 10(3):1–24
Tuanmat TZ, Smith M (2011) Changes in management accounting practices in Malaysia. Asian Rev
Account 19(3):221–242
Usakli A, Kucukergin KG (2018) Using partial least squares structural equation modeling in
hospitality and tourism: do researchers follow practical guidelines? Int J Contemp Hosp
Manag 30(11):3462–3512
The Impact of Ethical Leadership
on Employee Intrapreneurship, Work–Life
Balance, and Psychological Empowerment:
A PLS-SEM Analysis

Huma Bashir, Mumtaz Ali Memon, Naukhez Sarwar, Asﬁa Obaid,

and Muhammad Zeeshan Mirza

1 Introduction

Ethical principles inﬂuence people’s behaviors and allow them to restrict themselves
in accordance with socially desirable behaviors. As a result, the ethical leadership
style is gaining widespread recognition from researchers and practitioners alike, who
are interested in the emergence of moral crises in various organizational contexts
(Freire and Bettencourt 2020). Ethical leadership style not only incorporates the
leader’s moral personality traits like trustworthiness, fairness, honesty, and integrity
but also alternative ways through which they manage their employees (Joplin et al.
2021). An ethical leader develops and communicates ethical standards that are
acceptable in the workplace and holds their employees accountable if they exhibit
any unethical behavior (Li and Bao 2020; Treviño et al. 2000).
A recent survey indicates that 94% of service professionals in the USA work
more than 50 hours per week, with many of them also working on weekends (Apollo
Technical 2022). This shows that American workers are struggling to maintain a
work–life balance. This trend is also prevalent in Asia as professionals were less
optimistic about their work–life balance in 2021 than they were in 2020, with 46%
saying it was either good or very good, down from 50% (Hays 2021). According to a
Gallup survey, only 24% of employees believe their employers care about their well-
being (Harter 2022). According to a survey conducted by the World Economic
Forum (WEF 2021), 86% of respondents call for moral leadership in businesses,
and only about 8% of top executives consistently show moral leadership behavior at

H. Bashir · M. A. Memon (✉) · N. Sarwar · A. Obaid · M. Z. Mirza

NUST Business School, National University of Sciences and Technology, Islamabad, Pakistan
e-mail: [email protected]; [email protected];
asﬁ[email protected]; [email protected]

© Springer Nature Switzerland AG 2023 353

a high level. Considering these statistics, moral/ethical leadership is a need of time in

organizations.
An insensitive leader gives the impression that the entire organization is insensi-
tive, which makes employees tense, stressed, and burned out (Cowart et al. 2014).
On the other hand, a leader’s qualities of support, trust, understanding, and commu-
nication are vital for employees to solve problems of work–life balance (Cowart
et al. 2014). These qualities are part of ethical leadership, which possesses its own
unique form of leadership, albeit it can be practiced with other leadership styles
(Schwepker and Dimitriou 2021). According to previous research, ethical leaders are
viewed as trustworthy, disciplined, and caring in the decision-making process
(Li and Bao 2020). Moreover, ethical leadership can be a source of employees’
positive behaviors in the workplace, such as employee engagement (Joplin et al.
2021), organizational citizenship behavior, and better organizational performance
(Sharif and Scandura 2014). Employees typically consider ethical leaders as trust-
worthy and credible role models so that they start to adopt their leader’s character-
istics and also reciprocate such quality treatment by showing more desirable
behavior and ethical conduct in the workplace (Li and Bao 2020). One of those
behaviors represents intrapreneurial behavior, which is exhibited at a personal level
(Farrukh et al. 2021; Revuelto-Taboada et al. 2020).
Intrapreneurial behaviors refer to “all actions taken by ﬁrm members that relate to
the discovery and exploitation of entrepreneurial ideas and opportunities” (Sieger
et al. 2013, p. 362). Intrapreneurial behavior is not a part of an employee’s job
description, and such behavior does not emerge as a result of an organization’s
policies and strategies (Valsania et al. 2016; Rigtering and Weitzel 2013). Therefore,
it is crucial to investigate the factors that encourage intrapreneurial behavior in
employees within an organization.
Previous research recommended probing the impact of ethical leadership on
positive work-related outcomes (Suifan et al. 2020). It was suggested that ethical
leadership produces an immense impact on work–life balance along with job
satisfaction (Freire and Bettencourt 2020). However, the relation between ethical
leadership and work–life balance needs to be addressed with mediating effects of
different variables (Haar and Brougham 2022). This study has addressed this gap
and contributes to the literature by studying employee intrapreneurial behavior and
work–life balance. It was intriguing to ﬁgure out the effect of ethical leadership on
positive outcomes of an individual’s behavior. Past research has explained the
impact of ethical leadership on employees as ethical leaders empower them with
increased decision-making authority regarding their work and increase their work
autonomy (McKenna and Jeske 2021).
Psychological empowerment comprises a motivational factor that enables
employees to exhibit a sense of intrinsic task motivation. In addition, previous
studies suggested psychological empowerment as a variable to consider for future
research (Ilyas et al. 2020; Wang et al. 2016). The mediating effect of psychological
empowerment has rarely been studied with ethical leadership, intrapreneurial behav-
ior, and work–life balance to date. The causal relationship between moral leadership,
entrepreneurial activity, and work–life balance has been examined in this study, with
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 355

the mediating role of psychological empowerment. This study made several contri-
butions to the literature. First, by examining the novel relationship between ethical
leadership, intrapreneurial behavior, and work–life balance with the mediating effect
of psychological empowerment. Second, the research was conducted in the IT and
telecom sector of Pakistan, where it intends to broaden the theoretical foundation on
leadership literature in the South Asian context.

2 Theoretical Foundation and Hypothesis Development

2.1 Social Exchange Theory

Social Exchange Theory (SET, Blau 1964) has been extensively utilized in previous
literature (Memon et al. 2016; Walumbwa et al. 2020). The theory (Blau 1964) states
that actions of an individual depend upon the reactions of another individual. The
action of one is a reward for the other and vice versa. Blau (1964) defines social
exchanges as actions taken voluntarily by an organization with the intention that
employees will reciprocate with their actions in return (Memon et al. 2016). To put
this theory concisely, exchange relation happens when two parties enter into a
relationship with the understanding that performing for the benefit of the other will
be rewarding for them.
SET provides a theoretical base for this chapter. An exchange relationship in the
workplace is established when an employer contributes to the empowerment and
well-being of its employee. In return, employees begin to reciprocate by acting
positively and performing well at the job since they feel valued by the organization
(Memon et al. 2016). Previous research has found that when leaders encourage their
employees in managing work–life domains, acknowledge their effort in improving
their job performance, it will help leaders achieve their stated goals (Freire and
Bettencourt 2020). Since employees are treated ethically and have the leader’s trust,
the relationships are reciprocated by them, and as a result, employees give their best
to pay back to the organization. Furthermore, prior research indicates that individ-
uals would engage in intrapreneurial actions when they feel confidence in their
abilities/skills and believe that their efforts will be appreciated by their leaders or
organizations (Chouchane et al. 2021; Eisenberger et al. 1986). Ethical leaders also
play a huge role in empowering employees as they communicate the significance and
value of employee’s work, which leads to the sense of fulfillment among employees
(Suifan et al. 2020; Brown et al. 2005). Therefore, SET provides a firm theoretical
foundation for this chapter.
356 H. Bashir et al.

2.2 Ethical Leadership and Work–Life Balance

Ethical leadership is deﬁned as “the demonstration of normatively appropriate

conduct through personal actions and interpersonal relationships, and the promotion
of such conduct to followers through two-way communication, reinforcement and
decision-making” (Brown et al. 2005, p. 120). Brown et al. (2005) ﬁgured out two
major aspects of ethical leadership: moral person and moral manager. The former
depicts the personality related attributes and characteristics whereas the latter depicts
the transformation of follower’s actions in ethical context under his/her supervision.
Additionally, ethical leaders’ actions have the power to lessen employee’s work-
related issues since they are viewed as more dependable and thoughtful, which helps
employees feel less stressed about their jobs (Demirtas 2013). According to a prior
study, leaders are considered as ethical leaders if they build a trust-based relationship
with the employee and respect employee’s work–life balance than those who do not
(Cowart et al. 2014). Work–life balance refers to “the support by organization in an
employee’s personal life aspects, such as ﬂexibility in work hours, personal leaves,
leisure time etc.” (Balven et al. 2018, p. 29).
According to SET, an ethical leader’s behavior demonstrates fairness, integrity,
and trust to their employees, and it triggers employees to reciprocate such behavior
(Garba et al. 2018). Such leaders are also respectful toward work–life balance of
employees. Previous research indicated that when leaders regard an employee’s
work–life balance, employees will reciprocate with extra-role behavior such as
Organizational Citizenship Behavior (Poohongthong et al. 2014). With solid foun-
dation of this theory, it is hypothesized:
H1: Ethical leadership is positively related to work–life balance.

2.3 Ethical Leadership and Intrapreneurial Behavior

Intrapreneurial behavior in an organizational context is viewed as higher order

construct defined as “voluntary employee behavior aimed at generation of ideas,
perception of opportunities, creation of new products or the development of new
business lines” (Valsania et al. 2016, p. 133). Specifically, Innovative, proactive and
risk-taking behaviors of employees in an organization are considered as intrapre-
neurial behavior (Farrukh et al. 2019). Innovativeness is the ability of an individual
to develop new ideas in their work (Farrukh et al. 2019). Proactiveness is the
anticipation of an individual to take actions on future needs and challenges instead
of waiting to respond to demand (Farrukh et al. 2019). Risk-taking is defined as the
ability of an individual to take risks with the intention of providing benefit to the
organization, along with the accountability in case of a failure (Farrukh et al. 2019).
Such behaviors are mostly exhibited on an individual level by the employees and the
managers in an organization with the bottom-top approach in form of extra-role
behavior (Mahmoud et al. 2020). In previous studies, it was also concluded that
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 357

leadership stimulates employee extra-role work behaviors (Qiu et al. 2019; Lee and
Kim 2017). According to SET, when leaders reduce fear among employees and have
trust on them, employee will be engaged in intrapreneurial behavior in return
(Farrukh et al. 2021). Hence, it is hypothesized:
H2: Ethical leadership is positively related to intrapreneurial behavior.

2.4 Ethical Leadership and Psychological Empowerment

Empowerment generally includes the concept of delegating responsibility and

decision-making authority to employees, allowing them to act on their own along
with making sure that they have the resources to take decisions (Javed et al. 2017).
Psychological empowerment is defined as a “sense of control that employees
perceive in the organization, and that sense of control is reflected in four different
dimensions: impact, meaning, competence, and self-determination” (Zhang et al.
2021; Mishra and Spreitzer 1998, p. 557). Impact refers to the extent an individual
believes that they can influence an outcome as opposed to the difference their work
makes in achieving that outcome (Zhang et al. 2021). Meaning refers to an individ-
ual’s perception about value of their work according to own thoughts and standards
(Zhang et al. 2021). Competence refers to feelings of an individual’s self-efficacy
that they can perform a particular task successfully (Zhang et al. 2021). Self-
determination refers to an individual’s feelings of autonomy and responsibilities in
making decisions about work and their work-related actions (Zhang et al. 2021). The
performance of the employees usually affected by those in the organization who
regularly experience work–life imbalance. People want to work for an organization
which supports employee’s well-being, show confidence in their abilities, and intend
to be a part of that organization for a longer period (Tirta and Enrika 2020).
Employee’s potential success widely depends on their active involvement in the
work facilitated by psychological empowerment (Dust et al. 2018). Psychologically
empowered employees feel confident as they can shape their workplace roles and
duties. In such confidence building and supporting employee well-being, ethical
leaders can play a huge role due to their trust on employees and supporting nature.
By allowing employees to participate in decision-making and work design, ethical
leaders are more likely to help them understand the impact that they have on the
organization (Zhu et al. 2004). Such kind of supportive behavior of an ethical leader
creates a feeling of empowerment among employees. So, it is hypothesized that:
H3: Ethical leadership is positively related to psychological empowerment.
358 H. Bashir et al.

2.5 Psychological Empowerment and Intrapreneurial

Behaviors

Individual level intrapreneurial behavior is considered as an important factor for the

success of an organization (Mahmoud et al. 2018). Employees who feel empowered
are more likely to be creative because they believe their work has meaning and
autonomy. Additionally, empowered workers are more likely to use creativity to
complete their tasks successfully because of the sense of self-efficacy that is asso-
ciated with them (Mahmoud et al. 2018; Redmond et al. 1993). Employees who feel
empowered tend to be more innovative, proactive, and risk-takers, which ultimately
improves performance (Afsar et al. 2017). It has been tested empirically and proven
that enhancing employee empowerment and autonomy is a pertinent way of encour-
aging extra-role behavior (Farrukh et al. 2019). Employees who are more
empowered in their jobs have more opportunities to go above and beyond their
roles and find more efficient ways to complete their tasks to benefit the organization
(Asgari et al. 2012). As a result of being empowered, employees have an innate
motivation to improve organizational performance. So, with the firm support from
existing literature it is hypothesized that:
H4: Psychological empowerment is positively related to intrapreneurial behavior.

2.6 Psychological Empowerment and Work–Life Balance

As per past study, psychological empowerment may have a favorable impact on the
employee’s psychological attitude toward their job, behavior at work, and how they
perceive their roles in the organization (Panda and Sahoo 2021). Work–life balance
and psychological empowerment share the same parameters that characterize indi-
vidual’s level of satisfaction with their lives, and level of involvement in the
workplace (Panda and Sahoo 2021; Ickes et al. 2018; Grealish et al. 2017). When
the basic assumptions about an employee’s workplace and job are suitably met, only
then employees experience pleasure from professional lives. Unfulﬁlled perceptions
of control limit the productivity and commitment of professionals who experience
stress from work–life conﬂict and both personal and professional demand (Panda
and Sahoo 2021). Psychological empowerment increases employee’s sense of per-
sonal control which motivates them to work, resulting in favorable organizational
and individual level outcomes (Quinn and Spreitzer 1997). When employees have
control over issues like intense work pressure and personal issues; psychological
empowerment may increase the degree of work–life balance among employees. On
the basis of above literature, it is hypothesized that:
H5: Psychological empowerment is positively related to work–life balance.
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 359

2.7 Mediating Role of Psychological Empowerment

Past research has shown that leadership has a huge impact on the development of
intrapreneurial behavior by empowering employees (Valsania et al. 2016). The
ethical leader exhibits genuine concern and care for employees and behaves ethically
both in their work–life dealings as well as in their personal lives (Chughtai et al.
2015). Employee’s potential gets channelized, and competencies are enhanced when
an ethical leader enables them to accomplish their tasks and provides them with
developmental opportunities (Ahmad and Gao 2018). Such reciprocated behaviors
of both employees and leaders are supported by SET and following hypotheses have
been proposed:
H6: Psychological empowerment mediates the relationship between ethical leader-
ship and intrapreneurial behavior.
H7: Psychological empowerment mediates the relationship between ethical leader-
ship and work–life balance (Fig. 1).

3 Methodology

3.1 Context

The investigation was conducted in the telecommunication and IT sectors in

Pakistan. Despite the Covid-19 pandemic, Pakistan’s telecom sector contributed
129% more to the national treasury in 2020 than it did in 2019, indicating its
prominence as the main driver of the economy (Samaa web desk 2021). Previous
research suggested that the new working modes affected work–life balance of
employees in Latvia, which included employees of both genders in the survey and

Fig. 1 Conceptual framework

360 H. Bashir et al.

concluded that both men and women felt that their ability to focus on the work
responsibilities and task is affected by family life, especially the respondents aged
25–44 and those having children under the age of 18 in their homes (Lonska et al.
2021). Hofstede (2001) shed light on clusters of cultures existing across nations like
Asian Countries which are characterized with high power distance and high collec-
tivism (Ilyas et al. 2020). There was a need to test how the employees of telecom-
munication and the IT sector manage their work–life balance under ethical
leadership in Pakistan, as the employees in these sectors mostly work in teams and
groups under the supervision of their leaders. Also, whether they are empowered
enough to perform certain tasks which might result in their intrapreneurial behavior
as it is perceived that innovations and creativity is a job demand of employees
working in these sectors. It was noteworthy that this chapter has helped in ﬁguring
out the ways employees manage their work–life balance and exhibit intrapreneurial
behavior under ethical leadership in telecom and the IT sector.

3.2 Instruments

Ethical leadership was measured using a “six-item scale” adapted from Bormann
(2017), originally developed by Brown et al. (2005). Ethical leadership is defined as
“the demonstration of normatively appropriate conduct through personal actions and
interpersonal relationships, and the promotion of such conduct to followers through
two-way communication, reinforcement and decision-making” (Brown et al. 2005,
p. 120). This chapter adapted the scale and made small amendments such as the
beginning of questions replaced “today” with “my supervisor” because the present
study didn’t measure ethical leadership on a daily basis. A sample item includes “My
Supervisor/Boss listens to what employees have to say.”
A nine-item scale was used to measure intrapreneurial behavior which had three
dimensions of innovativeness, risk-taking, and pro-activeness adapted from Farrukh
et al. (2021), originally developed by Stull and Singh (2005) with reliability of
0.913. Intrapreneurial behavior is defined as “voluntary employee behavior aimed at
the generation of ideas, perception of opportunities, creation of new products or the
development of new business lines” (Valsania et al. 2016, p. 133). The statements
were amended to better fit the context of the chapter, for example, “this employee” in
the adapted scale was changed to “I” because data was gathered from employees. A
sample item includes “I contribute to the implementation of new ideas at work.”
A three-item scale was used to measure work–life balance adapted from (Rashmi
and Kataria 2021) originally developed by (Haar 2013) with the reliability of 0.80. In
the current study, work–life balance is defined as “support by organization in
employee’s personal life aspects, such as flexibility in work hours, personal leaves,
leisure time, etc.” (Balven et al. 2018, p. 29). A sample item includes “Nowadays, I
seem to enjoy every part of my life equally well”.
Psychological empowerment was measured by a “five-item scale” adapted from
(Rafique et al. 2021) with the reliability of 0.801 originally developed by (Spreitzer
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 361

1995). Psychological empowerment is deﬁned as a “sense of control that employees

perceive in the organization, and that sense of control is reflected in four different
dimensions: impact, meaning, competence, and self-determination” (Mishra and
Spreitzer 1998, p. 557). The adapted scale has six items, but this chapter eliminated
an item due to similarity with another item. The eliminated item is “I can decide on
my own how to go about doing my own work” which is similar to the item “I have
significant autonomy in determining how I do my job”. A sample item includes “I
am confident about my ability to do my job”.
All 23 items were measured using a 5-point, Likert-type scale (1 = strongly
disagree to 5 = strongly agree). Additionally, the questionnaire was pretested before
moving to data collection in order to identify any ambiguity related to the question-
naire statements or demographic information that respondents would not feel com-
fortable to answer (Memon et al. 2020a). There was no serious issue found in
pretesting. All items of survey questionnaire are given in the appendix of the chapter.

3.3 Sampling and Data Collection

For data collection, a quantitative approach with snowball sampling technique was
used. The snowball sampling technique, also known as sampling by reference, is
used when respondents are difficult to locate and are approached through references
or contacts (Memon et al. 2020a; Cooper and Schindler 2011). The investigation was
conducted in Covid-19 times, where physical data collection was not permitted in
software houses and telecom companies in Pakistan due to the prevailing policies of
reduced face-to-face interaction. For the smooth process of data collection, referrals
in different software houses and telecom companies were used. Data collection
period comprised ten days in which a total of 219 samples were collected. Initially,
almost 150 samples were collected without reminders, while the remaining 69 sam-
ples were collected by sending reminders. It was challenging to collect data due to
the time difference among people working in both sectors. Employees working in
software houses mostly worked in evening shifts and had busy schedules so they
requested to send them reminders so they can spare time as they volunteered to
participate.
Out of 219, a few samples were eliminated at initial screening process as they did
not meet the criteria of the research work. These included the respondents who filled
out the survey inappropriately like straight line patterns or suspicious demographic
information (n = 23), and a few didn’t belong to sectors being investigated (i.e., IT
and Telecom) (n = 12), leaving the final sample size of 184 for data analysis. Most
participants were men (66%), reflecting the predominance of men in Pakistan’s
telecom and software sectors. As there were two sectors under investigation, respon-
dents from the telecom industry made up 44% of the sample, while those in the IT
sector made up 55%.
The majority of the respondents have completed graduation, i.e., 96%. Most of
the respondents were Engineers (18%), followed by Assistant Developers (14%),
362 H. Bashir et al.

Table 1 Demographic information

Variable Category Frequency Percentage
Gender Male 120 65.2
Female 64 34.8
Age 25–30 14 7.6
31–35 70 38
36–40 52 28.3
41–45 33 17.9
46–50 9 4.9
51–55 4 2.2
56–60 2 1.1
Qualiﬁcation Bachelors 81 43.8
MBA 62 33.5
MSc 34 18.9
PhD 4 2.2
Other 3 1.6
Position Engineers 33 17.9
Assistants 23 12.5
Field specialist 13 7.1
Admin staff 11 6.0
Executives 14 7.6
Developers 26 14.1
Operational staff 24 13
Others 40 21.7

Operational staff (13%), Assistants (12%), Field Specialists/Executive Employees

(7%), and Admin staff (6%). The experience level of the employees ranges from
1–4 years to 14–19 years. Most of the respondents work in Islamabad (42%),
followed by Lahore (23%), Rawalpindi (8%), Karachi (6%), and Faisalabad/Multan
(4%). The majority of the respondents were married (54%). Final sample size of
184 samples was supported by sample size requirement for non-complex and
medium size quantitative study, i.e., 100 to 200 samples (Memon et al. 2020b;
Kline 2005). The demographic information of respondents is also summarized in
Table 1.

4 Data Analysis and Results

The research hypotheses were tested using partial least squares structural equation
modelling (PLS-SEM). A two-stage analytical procedure was used in this study, as
proposed by Anderson and Gerbing (1988). Internal consistency reliability, conver-
gent and discriminant validity of the measurement model were tested in the ﬁrst
stage, and the structural model was examined in the second stage, a stage for
hypotheses testing.
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 363

4.1 Measurement Model Assessment

The measurement model underwent testing to determine the discriminant validity

(DV), convergent validity (CV), and internal consistency reliability of the constructs
used in this chapter. Internal consistency reliability evaluates how well the items
reflect the latent constructs (Hair et al. 2014; Ramayah et al. 2018). Internal
consistency was tested using composite reliability (Hair et al. 2017). For each
construct, a measurement model is deemed satisfactory if its composite reliability
is greater than the threshold value of 0.7 (Nunnally 1978; Nunnally and Bernstein
1994; Richter et al. 2016). The findings revealed that the composite reliability for
each construct – Ethical Leadership (0.881), Intrapreneurial Work Behavior (0.891),
Psychological Empowerment (0.874), and Work–Life Balance (0.956) exceeded the
cut-off value (0.7), demonstrating the high internal consistency of the measures.
Another measure to consider is CV, which evaluates “the extent to which a
measure correlates positively with alternative measures of the same construct”
(Hair et al. 2017, p. 112). CV was estimated by examining the item’s outer loading
and the average variance extracted (AVE). As a general rule, outer loadings should
be 0.708 or greater, while an AVE score of 0.5 is acceptable (Avkiran 2017).
Additionally, items with a value of 0.6 outer loading are also considered as accepted
(Chin et al. 1997). According to Hair et al. (2017), if factors with higher loadings can
account for at least 50% of the variance (AVE = 0.50), indicators with lower factor
loadings should be retained. The results show that, apart from IWB6, all of the items
had satisfactory outer loadings. As a result, IWB6 “I would be willing to give up
some salary in exchange for the chance to try out my business idea if the rewards for
success were adequate” was omitted due to a weak factor loading value. All the
constructs, including ethical leadership (0.554), intrapreneurial work behavior
(0.508), psychological empowerment (0.582), and work–life balance (0.878),
attained acceptable AVE after the removal of IWB6. This validated the CV of the
constructs. Results of internal consistency reliability and CV tests are summarized in
Table 2.
This chapter also confirmed the DV by using another method of the heterotrait–
monotrait ratio (HTMT, Henseler et al. 2015). An HTMT value greater than 0.90
depicts the lack of DV (Hair et al. 2017). For HTMT, 0.85 is a more conservative
cut-off value (Henseler et al. 2015). The HTMT criterion results showed that the
present study did not deviate from the assumptions of discriminate validity, as
shown in Table 3, with a value of 0.85 HTMT. The overall measurement model’s
results show acceptable internal consistency reliability, CV, and DV.

4.2 Structural Model Assessment

To calculate the statistical signiﬁcance, the bootstrapping method (5000 subsamples

was used). The results of this chapter indicated that ethical leadership was
364 H. Bashir et al.

Table 2 Convergent validity and internal consistency reliability

Constructs Items Loadings AVE CR
Ethical leadership EL1 0.789 0.554 0.881
EL2 0.767
EL3 0.790
EL4 0.687
EL5 0.769
EL6 0.652
Intrapreneurial work behavior IWB1 0.662 0.508 0.891
IWB2 0.785
IWB3 0.754
IWB4 0.701
IWB5 0.796
IWB7 0.707
IWB8 0.600
IWB9 0.675
Psychological empowerment PE1 0.747 0.582 0.874
PE2 0.782
PE3 0.663
PE4 0.831
PE5 0.779
Work–life balance WLB1 0.919 0.878 0.956
WLB2 0.948
WLB3 0.944
Notes: IWB6 was deleted due to low loading

Table 3 Discriminant validity (HTMT criterion)

Ethical leadership IWB PE WLB
Ethical leadership
IWB 0.332
PS 0.440 0.648
WLB 0.394 0.329 0.330
Notes: PE Psychological empowerment, IWB Intrapreneurial work behavior, WLB Work–life
balance

signiﬁcantly positively related to work-life balance (H1: β = 0.285, p =0.000, LL=

0.144, UL=0.400). In contradiction to the initial assumption, the findings of second
hypothesis showed that ethical leadership does not have a direct effect on
employees’ intrapreneurial behavior (H2: β = 0.085, p =0.101, LL= -0.027,
UL=0.195). Further, ethical leadership positively influence psychological empow-
erment (H3: β = 0.374, p =0.000, LL= 0.262, UL=0.469 and psychological
empowerment (H4: β = 0.530, p =0.000, LL= 0.395, UL=0.631) was found to
have a positive and significant effect on employees’ intrapreneurial behavior. Fur-
thermore, psychological empowerment was also found to positively and
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 365

Table 4 Hypotheses testing (Direct effect)

Hypotheses Beta STDEV p CI [LL] CI [UL]
Ethical leadership→IWB 0.085 0.067 0.101 -0.027 0.195
Ethical leadership→PE 0.374 0.062 0.000 0.262 0.469
Ethical leadership→WLB 0.285 0.078 0.000 0.144 0.400
PE→IWB 0.530 0.070 0.000 0.395 0.631
PE→WLB 0.183 0.076 0.008 0.057 0.310
Notes: PE Psychological empowerment, IWB Intrapreneurial work behavior, WLB Work–life
balance

Table 5 Hypotheses testing (Indirect effect)

Hypotheses Beta STDEV CI [LL] CI [UL]
Ethical_Leadership→Psy_Emp→Intrap_Beh 0.198 0.043 0.117 0.283
Ethical_Leadership→Psy_Emp→WLB 0.069 0.033 0.010 0.140
Notes: PE Psychological empowerment, IWB Intrapreneurial work behavior, WLB Work–life
balance

signiﬁcantly inﬂuence work-life balance of employees (H5: β = 0.183, p =0.008,

LL= 0.057, UL=0.310). Sixth (H6) and seventh (H7) hypotheses (H3) tested the
mediating effect of psychological empowerment between the relationship of ethical
leadership and outcomes (intrapreneurial behavior and work-life balance). This
chapter followed Preacher et al.’s (2007) definition of mediation effect according
to which “an indirect effect or mediation, is said to occur when the causal effect of an
independent variable (X) on a dependent variable (Y) is transmitted by a mediator
(M)” (Wu 2017). If the indirect effect of independent variable on dependent variable
is significant and the confidence intervals do not contain zero, it means mediation is
established (Qian et al. 2017). The results for indirect effect showed that psycho-
logical empowerment (H6: β =0.198, LL: 0.117, UL: 0.283) mediates the relation-
ship between ethical leadership and intrapreneurial behavior. The fourth hypothesis
H4 tested the mediating nature of psychological empowerment between the rela-
tionship of ethical leadership and work–life balance. The results indicated that
psychological empowerment (H7: β = 0.069, LL: 0.010, UL: 0.140) mediates
the relationship between ethical leadership and work–life balance. All four
hypotheses were supported, and the structural model analysis results are reported
in Tables 4 and 5.
According to Hair et al. (2017), researchers should also report the coefficient of
determination (R2), effect size ( f2), and predictive relevance (Q2) in addition to
describing the significance of the relationships. R2 measures how well the indepen-
dent variable or variables explain the corresponding dependent variable or variables.
The R2 value indicates that ethical leadership explains 14.5% of employee’s psy-
chological empowerment (R2 = 0.145), 31.7% of employee’s intrapreneurial behav-
ior (R2 = 0.317), and 8.5% of their work–life balance (R2 = 0.085). Following that,
f 2 denotes effect size, or how much an independent variable contributes to the R2 of
the dependent variable. f 2 could be calculated from the following equation: f 2 = (R2
366 H. Bashir et al.

included - R2 excluded) ÷ (1 - R2 included). Cohen (1988) suggests f 2 values of

0.02 as small, 0.15 as medium, and 0.35 as large size effect. The results of f 2 showed
that ethical leadership has no direct effect on intrapreneurial behavior of employees
( f2 = 0.009), a medium size effect on psychological empowerment ( f2 = 0.163), a
small to medium size effect on work–life balance. Psychological empowerment has a
large direct effect on intrapreneurial behavior ( f2 = 0.356) and a small size effect on
work–life balance ( f2 = 0.034).
The Q2 measures the research model’s predictive relevance or out-of-sample
predictive power for the speciﬁc construct (Chin et al. 2008; Hair et al. 2014). The
cross-validated redundancy values of the constructs were calculated via a
blindfolding procedure which is a measure of Q2. “Q2 values greater than zero for
a certain reﬂective endogenous latent variable indicate the path model’s predictive
relevance for the particular construct” (Hair et al. 2014, p. 178). The results showed a
satisfactory level of predictive ability for ethical leadership on psychological
empowerment (Q2 = 0.079), intrapreneurial behavior (Q2 = 0.147), and work–life
balance (Q2 = 0.068). Values of (R2) and (Q2) were calculated.

4.3 Common Method Bias

As the data were collected by self-reporting survey, so there might be a chance for
common method bias (Memon et al. 2020a; Podsakoff et al. 2003). To minimize the
impact of common method bias, both procedural remedies and statistical methods
were used. At procedural level, i.e., during data collection, participants were assured
with their anonymity and their responses were kept conﬁdential and only used for
current study purpose (Memon et al. 2016). Additionally, pretesting was conducted
in order to make sure that respondents understand statements easily by avoiding
difﬁcult terminologies (Memon et al. 2020a). Additionally, a Harman one-factor test
(Harman 1976) was used to identify common method bias after data collection. The
results showed the maximum variance could be 30%, which indicated there was no
issue of common method bias in the current study (Memon et al. 2020a; Babin et al.
2016).

5 Discussion

The chapter investigated the relationship between ethical leadership, intrapreneurial

behavior, and work–life balance with the mediating effect of psychological empow-
erment. The ﬁrst hypothesis (H1) which states that ethical leadership is positively
related to work–life balance is supported with a signiﬁcant direct relationship
between ethical leadership and work–life balance. It shows that ethical leadership
is effective in improving work–life balance of employees. It implies that ethical
leaders have the ability to create ethical environment in the organization which
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 367

nurtures work–life balance of employees. By having a work–life balance, employees

will be happier and healthier both physically and psychologically and they will feel
less stressed, which is a common problem in the organizations (Bhende et al. 2020;
Chiang et al. 2010).
The second hypothesis (H2) hypothesized that ethical leadership is positively
related to intrapreneurial behavior. Surprisingly, the results of H2 indicates that
ethical leadership does not influence employee’s intrapreneurial behavior. It means
only ethical aspects of leader behavior are not sufficient to motivate employees to
exhibit intrapreneurial behavior at workplace. The results of H2 are consistent with
existing studies which state that leaders may use different mediating strategies to
encourage employee intrapreneurial behavior, such as an entrepreneurial mindset,
job autonomy, organizational identification, and empowerment along with organi-
zational support (Chang et al. 2017; Amankwaa et al. 2019; Lei et al. 2020).
The H3 of research work is accepted which states that ethical leadership posi-
tively affects psychological empowerment of employees. The results of H3 also
indicates that ethical leadership is a source of psychological empowerment among
employees. This result is not unexpected because ethical leaders are better at
imparting the work values that give the employees a sense of purpose. Additionally,
ethical leaders highlight skills and capabilities required for performing the task and
offer constructive criticism to instill a sense of competence in their followers. These
results also provide a support for a previous study (Suifan et al. 2020) which states
that ethical leaders positively affect psychological empowerment and reduce turn-
over intention among employees working in the banking sector.
The findings of H4 indicate that psychological empowerment results in intrapre-
neurial behavior among employees. Employees are more likely to engage in more
creative work to improve job performance when they feel they have control on
making job-related decisions, work flexibility and freedom, the power to influence
others, and a sense of the work’s meaning. So, the findings also provide support to a
previous study by Mahmoud et al. (2018) which states that psychological empow-
erment is an important factor for nurturing intrapreneurial behavior among middle
managers in medium enterprises.
The findings of H5 indicate that psychological empowerment is an important
aspect in establishing work–life balance. The idea of work–life balance is extremely
individualized and subjective, making it specific to each employee. In light of this,
an employee’s level of satisfaction depends on maintaining a work–life balance for
their overall well-being, which is a psychological state everyone strives to achieve.
Unless employees have confidence in their capabilities, they will not feel
empowered (Spreitzer 1995). Therefore, results are also aligned with past study
which states that the feeling of having capabilities to deal with job requirement
results in work–life balance (Ambad and Bahron 2012).
The H6 is also accepted which states that mediating effect of psychological
empowerment strengthens the relationship between ethical leadership and intrapre-
neurial behavior of employees. Ethical leaders consider the growth needs of
employees and psychologically empower them in the decision-making of their
work design. Also, the target audience of investigation were employees from the
368 H. Bashir et al.

IT and telecom sectors who need great support from their leaders for bringing
innovation in their products, ideas, and services. Ethical leader’s sincere and sup-
portive relationship with employees helps them in getting involved in intrapreneurial
behavior in the organization. Such intrapreneurial behaviors of employees make
them motivated and committed to their jobs and become a source of organizational
prosperity. Findings are also consistent, to some extent, with previous studies which
state that ethical leadership positively affects employee creativity (Javed et al. 2017).
With such a supportive behavior of ethical leaders, employees perceive control over
their task which makes them self-confident and enables them to think out of the box
and bring novelty in job tasks through innovations. Findings also support previous
research which concludes that leaders can stimulate intrapreneurial behavior among
the employees by empowering them (Valsania et al. 2016).
The H7 of this chapter is also accepted, which states that psychological empow-
erment mediates the relationship between ethical leadership and low work–life
balance. This implies that due to ethical leadership and psychological empowerment
in job task, employees will be satisfied with their work and personal life. Employees
will have more positive attitude toward work-related tasks under ethical leaders who
provide them autonomy in integrating work and non-work life. Employees will be
more motivated and engaged in their job tasks and try to maintain balance between
professional and personal life. In a previous study, it is concluded that psychological
empowerment positively affects work–life balance of employees which also
strengthens current study findings (Panda and Sahoo 2021).
In line with SET, the results of this chapter revealed that when ethical leaders
psychologically empower employees, it results in reciprocated behavior of intrapre-
neurial behavior and creates a work–life balance along with improved work perfor-
mance. This shows that employees consider ethical leaders more credible,
trustworthy, and supportive and exhibit intrapreneurial behavior and maintain
work–life balance well by perceiving psychological empowerment. Employees
show positive behavior (i.e., more intrapreneurial behavior) and commitment
under ethical leaders who provide opportunities to employees in decision-making
related to job tasks.

6 Implications

The ﬁndings of this study show that ethical leadership is not only imperative in
resolving issues of low work–life balance among employees but also carries a key
role in shaping positive work behavior among employees, such as intrapreneurial
behavior. This fact is valuable not only for scholars but for practitioners as well.
Results established the importance of ethical leadership for practitioners of the IT
and Telecom sectors in Pakistan. Findings of the chapter demonstrate that ethical
leadership inﬂuences employee intrapreneurial behavior and work–life balance.
Managers in both sectors should demonstrate ethical leadership by creating an
ethical atmosphere, safeguarding employees’ rights and dignity, encouraging and
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 369

empowering employees to produce ideas that can be implemented in the organiza-

tion, without intruding in employees’ personal life.
Behaviors modeled by ethical leaders, such as honesty, integrity, and respect,
create an empowering environment in a company which leads to the growth of
innovative and talented employees, an asset to an organization’s management.
Previous research has also recommended that managers should focus on building
strategies and policies that adhere to moral and ethical standards for psychological
empowerment of employees (Javed et al. 2017). For example, managers can make
employees aware of the importance of their work in the organization, provide them
autonomy and control over their job, and motivate them intrinsically for work
performance rather than through external rewards such as bonuses, raises, and so
on. These practices can assist managers in encouraging intrapreneurial behavior
among staff members.
Organizations should take the initiative to enhance their employees’ quality of life
because it has a direct impact on employee well-being, but also has some indirect
effects on organizational performance. According to the results, ethical leadership
has been discovered to play a role in increasing employees’ ability to balance their
personal and professional lives through task empowerment. IT and telecom compa-
nies should design policies to incorporate employee needs into their processes,
which can only be accomplished by instilling ethical leadership qualities in man-
agers. Managers should create trust with their employees by giving them some
autonomy in doing their job responsibilities rather than micromanaging them, as
well as some flexibility in working hours to complete their job obligations. With this
practice, employees will be more punctual by arriving and leaving the office on time,
allowing them to maintain a balance between work and non-work life. Previous
research has shown that when firms emphasize work–life balance policies and
provide people job control, employees experience fewer negative behaviors at
work, such as stress (Bhende et al. 2020; Chiang et al. 2010).
By examining a novel relationship between ethical leadership and intrapreneurial
behavior, this chapter theoretically makes a significant contribution to the literature
on leadership and human resource management. The chapter expands the knowledge
on the importance of ethical leadership in promoting intrapreneurial behavior and
helps employees in maintaining work–life balance through psychological empow-
erment. Although the direct influence of ethical leadership on producing intrapre-
neurial conduct in employees is not significant, the psychological empowerment
strengthens the relationship in both sectors considerably. The concept of behavior
reciprocity in the SET is strengthened by the fact that employees who work for
ethical leaders have a high level of association with the firm and perform better in
roles and tasks unrelated to their normal tasks.
370 H. Bashir et al.

7 Conclusion

The primary goal of this chapter was to investigate the impact of ethical leadership
on intrapreneurial behavior and work–life balance. Additionally, this chapter inves-
tigated the mediating role of psychological empowerment between the relationship
of intrapreneurial behavior and work–life balance with ethical leadership. The
findings revealed that ethical leadership has a significant direct effect on work–life
balance of employees. The direct effect of ethical leadership did not have a signif-
icant effect on intrapreneurial behavior of employees, but through psychological
empowerment, the effect was more significant.
This chapter includes a few limitations. Firstly, as the data is only collected from
software houses and the telecom sector of Pakistan, it may not be generalizable to
other sectors in Pakistan. Future researchers can replicate this study in other sectors
and geographical regions which will further validate the current study’s findings.
Moreover, data was only collected from employees in the present study; future
studies should incorporate manager’s perspective as well for holistic understanding
of subject matter. Additionally, research design was cross-sectional in nature due to
time constraints. In future, longitudinal or qualitative research design is also
suggested for better clarity and in-depth understanding of subject matter. Lastly, it
was detected in findings of investigation that an ethical leader has no direct impact
on intrapreneurial behavior, future studies can examine intrapreneurial behavior with
other leadership styles.

References

Afsar B, Badir YF, Saeed BB, Hafeez S (2017) Transformational and transactional leadership and
employee’s entrepreneurial behavior in knowledge–intensive industries. Int J Hum Resour
Manag 28(2):307–332
Ahmad I, Gao Y (2018) Ethical leadership and work engagement: the roles of psychological
empowerment and power distance orientation. Manag Decis 56(9):1991–2005
Amankwaa A, Gyensare MA, Susomrith P (2019) Transformational leadership with innovative
behaviour: examining multiple mediating paths with PLS-SEM. Leadersh Organ Dev J 40(4):
402–420
Ambad SNA, Bahron A (2012) Psychological empowerment: the inﬂuence on organizational
commitment among employees in the construction sector. J Glob Bus Manag 8(2):73–81
Anderson JC, Gerbing DW (1988) Structural equation modeling in practice: a review and
recommended two-step approach. Psychol Bull 103(3):411–423
Apollo Technical (2022, Mar 20) 10 statistics on work-life balance that May surprise you. Apollo
Technical. https://www.apollotechnical.com/statistics-on-work-life-balance/. Accessed
18 Jun 2022
Asgari MH, Taleghani M, Mohalleh SGA (2012) The relationship between psychological empow-
erment of employees with corporate entrepreneurship of Guilan Payamnour university. J Basic
Appl Sci Res 2(4):3181–3187
Avkiran NK (2017) An in-depth discussion and illustration of partial least squares structural
equation modeling in health care. Health Care Manag Sci 21(3):401–408
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 371

Babin BJ, Griffin M, Hair JF (2016) Heresies and sacred cows in scholarly marketing publications. J
Bus Res 69(8):3133–3138
Balven R, Fenters V, Siegel DS, Waldman D (2018) Academic entrepreneurship: the roles of
identity, motivation, championing, education, work-life balance, and organizational justice.
Acad Manag Perspect 32(1):21–42
Bhende P, Mekoth N, Ingalhalli V, Reddy YV (2020) Quality of work life and work–life balance. J
Hum Values 26(3):256–265
Blau P (1964) Exchange and power in social life. Wiley, New York, NY
Bormann KC (2017) Linking daily ethical leadership to followers’ daily behaviour: the roles of
daily work engagement and previous abusive supervision. Eur J Work Organ Psy 26(4):
590–600
Brown ME, Treviño LK, Harrison DA (2005) Ethical leadership: a social learning perspective for
construct development and testing. Organ Behav Hum Decis Process 97(2):117–134
Chang YY, Chang CY, Chen CW (2017) Transformational leadership and corporate entrepreneur-
ship: cross-level mediation moderation evidence. Leadersh Organ Dev J 38(6):812–833
Chiang FF, Birtch TA, Kwan HK (2010) The moderating roles of job control and work-life balance
practices on employee stress in the hotel and catering industry. Int J Hosp Manag 29(1):25–32
Chin WW, Gopal A, Salisbury WD (1997) Advancing the theory of adaptive structuration: the
development of a scale to measure faithfulness of appropriation. Inf Syst Res 8(4):342–367
Chin WW, Peterson RA, Brown SP (2008) Structural equation modeling in marketing: some
practical reminders. J Mark Theory Pract 16(4):287–298
Chouchane R, Fernet C, Austin S, Zouaoui SK (2021) Organizational support and intrapreneurial
behavior: on the role of employees’ intrapreneurial intention and self-efficacy. J Manag Organ
29(2):1–17
Chughtai A, Byrne M, Flood B (2015) Linking ethical leadership to employee Well-being: the role
of trust in supervisor. J Bus Ethics 128(3):653–663
Cohen J (1988) Statistical power analysis for the behavioral sciences, 2nd edn. Lawrence Erlbaum,
New Jersey
Cooper DR, Schindler PS (2011) Business research methods, 11th edn. McGraw-Hill,
New York, NY
Cowart T, Gilley A, Avery S, Barber A, Gilley JW (2014) Ethical leaders: trust, work-life balance,
and treating individuals as unique. J Leadersh Account Ethics 11(3):70–81
Demirtas O (2013) Ethical leadership influence at organizations: evidence from the field. J Bus
Ethics 126(2):273–284
Dust SB, Resick CJ, Margolis JA, Mawritz MB, Greenbaum RL (2018) Ethical leadership and
employee success: examining the roles of psychological empowerment and emotional exhaus-
tion. Leadersh Q 29(5):570–583
Eisenberger R, Huntington R, Hutchison S, Sowa D (1986) Perceived organizational support. J
Appl Psychol 71(3):500–507
Farrukh M, Lee JWC, Shahzad IA (2019) Intrapreneurial behavior in higher education institutes of
Pakistan: the role of leadership styles and psychological empowerment. J Appl Res Higher Educ
11(2):273–294
Farrukh M, Meng F, Raza A (2021) Believe they can succeed, and they will: intrapreneurial
behavior and leadership. Eur J Innov Manag 25(3):661–679
Freire C, Bettencourt C (2020) Impact of ethical leadership on job satisfaction: the mediating effect
of work–family conflict. Leadersh Organ Dev J 41(2):319–330
Garba OA, Babalola MT, Guo L (2018) A social exchange perspective on why and when ethical
leadership foster customer-oriented citizenship behavior. Int J Hosp Manag 10:1–8
Grealish A, Tai S, Hunter A, Emsley R, Murrells T, Morrison AP (2017) Does empowerment
mediate the effects of psychological factors on mental health, Well-being, and recovery in
young people? Psychol Psychother Theory Res Pract 90(3):314–335
Haar JM (2013) Testing a new measure of work–life balance: a study of parent and non-parent
employees from New Zealand. Int J Hum Resour Manag 24(17):3305–3324
372 H. Bashir et al.

Haar J, Brougham D (2022) Work antecedents and consequences of work-life balance: a two
sample study within New Zealand. Int J Hum Resour Manag 33(4):784–807
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2014) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Los Angeles
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2017) A primer on partial least squares structural
equation modeling (PLS-SEM), 2nd edn. Sage, Thousand Oaks
Harman HH (1976) Modern factor analysis. The University of Chicago Press, Chicago
Harter J (2022, Mar 18) Percent who feel employer cares about their wellbeing plummets. https://
www.gallup.com/workplace/390776/percent-feel-employer-cares-wellbeing-plummets.aspx.
Accessed 26 Jul 2022
Hays (2021) Asia’s employees face salary and work-life balance difficulties, and must upskill to
thrive. Hays. https://www.hays.co.jp/en/press-release/content/asia-s-employees-face-salary-
and-work-life-balance-difficulties-and-must-upskill-to-thrive. Accessed 25 Jul 2022
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Hofstede G (2001) Culture’s recent consequences: using dimension scores in theory and research.
Int J Cross-cult Manag 1(1):11–17
Ickes SB, Wu M, Mandel MP, Roberts AC (2018) Associations between social support, psycho-
logical Well-being, decision making, empowerment, infant and young child feeding, and
nutritional status in Ugandan children ages 0 to 24 months. Matern Child Nutr 14(1):e12483
Ilyas S, Abid G, Ashfaq F (2020) Ethical leadership in sustainable organizations: the moderating
role of general self-efficacy and the mediating role of organizational trust. Sustain Prod Consum
22:195–204
Javed B, Khan AA, Bashir S, Arjoon S (2017) Impact of ethical leadership on creativity: the role of
psychological empowerment. Curr Issue Tour 20(8):839–851
Joplin T, Greenbaum RL, Wallace JC, Edwards BD (2021) Employee entitlement, engagement, and
performance: the moderating effect of ethical leadership. J Bus Ethics 168(4):813–826
Kline RB (2005) Principles and practice of structural equation modeling. Guilford Publications, NY
Lee SA, Kim SH (2017) Role of restaurant employees’ intrinsic motivations on knowledge
management: an application of need theory. Int J Contemp Hosp Manag 29(11):2751–2766
Lei H, Leaungkhamma L, Le PB (2020) How transformational leadership facilitates innovation
capability: the mediating role of employees’ psychological capital. Leadersh Organ Dev J 41(4):
481–499
Li C, Bao Y (2020) Ethical leadership and positive work behaviors: a conditional process model. J
Manag Psychol 35(3):155–168
Lonska J, Mietule I, Litavniece L, Arbidane I, Vanadzins I, Matisane L, Paegle L (2021) Work–life
balance of the employed population during the emergency situation of COVID-19 in Latvia.
Front Psychol 12:1–15
Mahmoud MA, Ahmad S, Poespowidjojo DAL (2018) The role of psychological safety and
psychological empowerment on intrapreneurial behavior towards successful individual perfor-
mance: a conceptual framework. Sains Humanika 10(3):65–72
Mahmoud MA, Ahmad S, Poespowidjojo DAL (2020) Intrapreneurial behavior, big five personal-
ity and individual performance. Manag Res Rev 43(12):1499–1516
McKenna J, Jeske D (2021) Ethical leadership and decision authority effects on nurses’ engage-
ment, exhaustion, and turnover intention. J Adv Nurs 77(1):198–206
Memon MA, Salleh R, Baharom MNR (2016) The link between training satisfaction, work
engagement and turnover intention. Eur J Train Dev 40(6):407–429
Memon MA, Salleh R, Mirza MZ, Cheah JH, Ting H, Ahmad MS, Tariq A (2020a) Satisfaction
matters: the relationships between HRM practices, work engagement and turnover intention. Int
J Manpow 42(1):21–50
Memon MA, Ting H, Cheah JH, Thurasamy R, Chuah F, Cham TH (2020b) Sample size for survey
research: review and recommendations. J Appl Struct Equ Model 4(2):1–20
The Impact of Ethical Leadership on Employee Intrapreneurship, Work–Life. . . 373

Mishra AK, Spreitzer GM (1998) Explaining how survivors respond to downsizing: the roles of
trust, empowerment, justice, and work redesign. Acad Manag Rev 23(3):567–588
Nunnally JC (1978) Psychometric theory. McGraw-Hill, New York, NY
Nunnally JC, Bernstein IH (1994) Psychometric theory, 3rd edn. McGraw-Hill, New York, NY
Panda A, Sahoo CK (2021) Work–life balance, retention of professionals and psychological
empowerment: an empirical validation. Eur J Manag Stud 26(2):103–123
Podsakoff PM, MacKenzie S, Lee J, Podsakoff NP (2003) Common method biases in behavioural
research. MIS Q 30(1):115–141
Poohongthong C, Surat P, Sutipan P (2014) A study on the relationships between ethical leadership,
work-life balance, organizational socialization, and organizational citizenship behavior of
teachers in northern Thailand. J Behav Sci 9(2):17–28
Preacher KJ, Rucker DD, Hayes AF (2007) Addressing moderated mediation hypotheses: theory,
methods, and prescriptions. Multivar Behav Res 42(1):185–227
Qian J, Wang B, Han Z, Song B (2017) Ethical leadership, leader-member exchange and feedback
seeking: a double-moderated mediation model of emotional intelligence and work-unit struc-
ture. Front Psychol 8:1174
Qiu S, Alizadeh A, Dooley LM, Zhang R (2019) The effects of authentic leadership on trust in
leaders, organizational citizenship behavior, and service quality in the Chinese hospitality
industry. J Hosp Tour Manag 40:77–87
Quinn RE, Spreitzer GM (1997) The road to empowerment: seven questions every leader should
consider. Organ Dyn 26(2):37–49
Raﬁque MA, Hou Y, Chudhery MAZ, Gull N, Ahmed SJ (2021) The dimensional linkage between
public service motivation and innovative behavior in public sector institutions: the mediating
role of psychological empowerment. Eur J Innov Manag 26(1):207–229
Ramayah TJFH, Cheah J, Chuah F, Ting H, Memon MA (2018) Partial least squares structural
equation modeling (PLS-SEM) using smartPLS 3.0. An updated guide and practical guide to
statistical analysis, 2nd edn. Pearson, Kuala Lumpur
Rashmi K, Kataria A (2021) The mediating role of work-life balance on the relationship between
job resources and job satisfaction: perspectives from Indian nursing professionals. Int J Organ
Anal (vol ahead of print)
Redmond MR, Mumford MD, Teach R (1993) Putting creativity to work: effects of leader behavior
on subordinate creativity. Organ Behav Hum Decis Process 55(1):120–151
Revuelto-Taboada L, Canet-Giner MT, Balbastre-Benavent F (2020) Fostering intrapreneurial
behavior through human resource management system. Balt J Manag 15(3):353–373
Richter NF, Sinkovics RR, Ringle CM, Schlägel C (2016) A critical look at the use of SEM in
international business research. Int Mark Rev 33(3):376–404
Rigtering JC, Weitzel U (2013) Work context and employee behaviour as antecedents for
intrapreneurship. Int Entrep Manag J 9(3):337–360
Samaa Web desk (2021, Jan 15) Telecom sector contributed Rs278b to Pakistan economy in 2020:
PTA. Samaa News. https://www.samaa.tv/news/pakistan/2021/01/telecom-sector-contributed-
rs278b-to-pakistan-economy-in-2020-pta/. Accessed 27 Nov 2021
Schwepker CH Jr, Dimitriou CK (2021) Using ethical leadership to reduce job stress and improve
performance quality in the hospitality industry. Int J Hosp Manag 94:102860
Sharif MM, Scandura TA (2014) Do perceptions of ethical conduct matter during organizational
change? Ethical leadership and employee involvement. J Bus Ethics 124(2):185–196
Sieger P, Zellweger T, Aquino K (2013) Turning agents into psychological principals: aligning
interests of non-owners through psychological ownership. J Manag Stud 50(3):361–388
Spreitzer GM (1995) Psychological empowerment in the workplace: dimensions, measurement, and
validation. Acad Manag J 38(5):1442–1465
Stull M, Singh J (2005) Intrapreneurship in nonproﬁt organizations examining the factors that
facilitate entrepreneurial behaviour among employees. Case Western Reserve University, pp
192–210
374 H. Bashir et al.

Suifan TS, Diab H, Alhyari S, Sweis RJ (2020) Does ethical leadership reduce turnover intention?
The mediating effects of psychological empowerment and organizational identification. J Hum
Behav Soc Environ 30(4):410–428
Tirta AH, Enrika A (2020) Understanding the impact of reward and recognition, work-life balance,
on employee retention with job satisfaction as mediating variable on millennials in Indonesia. J
Bus Retail Manag Res 14(03):88–99
Treviño LK, Hartman LP, Brown M (2000) Moral person and moral manager: how executives
develop a reputation for ethical leadership. Calif Manag Rev 42(4):128–142
Valsania SE, Moriano JA, Molero F (2016) Authentic leadership and intrapreneurial behavior:
cross-level analysis of the mediator effect of organizational identification and empowerment. Int
Entrep Manag J 12(1):131–152
Walumbwa FO, Christensen-Salem A, Perrmann-Graham J, Kasimu P (2020) An identification-
based framework examining how and when salient social exchange resources facilitate and
shape thriving at work. Hum Resour Dev Rev 19(4):339–361
Wang D, Gan C, Wu C (2016) LMX and employee voice: a moderated mediation model of
psychological empowerment and role clarity. Pers Rev 45(3):605–615
WEF (2021, Feb 19) Why moral leadership matters now more than ever, Weforum. https://www.
weforum.org/agenda/2021/02/why-moral-leadership-matters-now-more-than-ever/. Accessed
18 Jul 2022
Wu YC (2017) Mechanisms linking ethical leadership to ethical sales behavior. Psychol Rep
120(3):537–560
Zhang S, Bowers AJ, Mao Y (2021) Authentic leadership and teachers’ voice behaviour: the
mediating role of psychological empowerment and moderating role of interpersonal trust.
Educ Manag Adm Leadersh 49(5):768–785
Zhu W, May DR, Avolio BJ (2004) The impact of ethical leadership behavior on employee
outcomes: the roles of psychological empowerment and authenticity. J Leadersh Organ Stud
11(1):16–26
Part VI
Education
PLS and Educational Research:
Epistemological and Methodological
Interpretations in Philosophy

Hongfeng Zhang, Haoqun Yan, and Johnny F. I. Lam

1 Introduction

In the sphere of social sciences and educational research, empirical research has been
in the mainstream and has resulted in a complete set of paradigms and theories.
Throughout the development of the methodological system, the epistemology of
modern philosophy has a different epistemological orientation. This chapter aims to
identify the intellectual implications in social science research, clarify the philo-
sophical foundations of the two research paradigms, and the role played by partial
least squares structural equation modeling (PLS-SEM) as an important analytical
tool in empirical research in educational research and the future direction of
development.

2 The Epistemological Implications of Social Science

Research Methods

Knowledge in the social sciences is acquired and accumulated mainly with “issue
awareness” as a precursor. Questions are sometimes manifested as beliefs, or as
determination, and require the guidance of non-rationalistic ways of thinking on the
way to facts and truth, namely, “we recognize matters inaccessible to reason, the
sacred, and recognize non-rationalistic means to apprehend them” (Grafstein and
Dobbs 1988). However, the problem is often rationally presented to us. Empiricists
tend to approach questions based on sensory experience, and some “correct” results

H. Zhang · H. Yan (✉) · J. F. I. Lam

Faculty of Humanities and Social Sciences, Macao Polytechnic University, Macao, China
e-mail: [email protected]; [email protected]; ﬁ[email protected]

© Springer Nature Switzerland AG 2023 377

can be obtained under the inductive synthesis of a certain research paradigm. But
rationalists believe that human beings have the ability to grasp self-evident rules,
axioms, and logic, and that truth can be reduced to the universal necessary propo-
sitions and knowledge grasped by this ability, as well as to the system of knowledge
and logic rationally deduced from these knowledge propositions. In essence, empir-
icism and rationalism have very different emphases in constituting a system of
knowledge, with the former emphasizing sense experience and the latter valuing
rational thought. However, an insistent emphasis on the irreplaceability of sense
experience inevitably fragmented facts and ideas and prevented the formation of a
universally necessary system of knowledge. Likewise, self-evident truths and prin-
ciples exist only in a priori concepts and categories, and everything logically
deduced from them is isolated from the empirical world. Therefore, “Congenital
Comprehensive Judgment” by Kant organically uniﬁes the two epistemologies, that
is, in the process of knowing, the a priori categories of universal necessity and the
real-world sense experience can be uniﬁed by the a priori unity. While freeing the
rules of induction and deduction from their limitations, thus enabling the research
methodology of the social sciences to gain new insights.

3 The Divide between Empirical and Normative Research

In terms of general methodological cognition, research methods in social sciences

can be broadly divided into two categories: one category belongs to empirical
research. It collects data and information through observation, interviews, question-
naires, experiments, physical objects, and so forth to obtain experience; and then
generalizes or describes the experience in-depth, to verify, predict, develop theories,
and interpret the meaning. The other category belongs to normative research, which
starts from certain identiﬁed principles, rationales, or value judgments, and forms
universal value judgments or new theoretical discoveries through logical deduction.
These two types of studies take empiricism and rationalism as their epistemological
foundations, respectively. Scholars who represent these different domains are said to
be guided by different theories, assumptions, and norms that often result in misun-
derstanding or a lack of appreciation for each other’s endeavors (Rosenthal and
Buchholz 2000). Research in the social sciences has been developed to date that
there is no longer a clear division between the two types of research methods. Many
normative studies also require data collection and the use of empirical observation,
while the rules of empirical studies themselves contain theoretical presuppositions
and value tendencies, and the components of rational analysis have long been
nurtured in them.
Further, empirical research can be divided into quantitative research and qualita-
tive research, each with different philosophical foundations and ways of data
collection and analysis. Quantitative research is undoubtedly more characteristic of
natural science, which believes that the use of ever-changing statistical analysis tools
and procedural empirical means can make research results reach a level of
PLS and Educational Research: Epistemological and. . . 379

reﬁnement and accuracy. While qualitative research takes phenomenology and

hermeneutics as its philosophical foundation, emphasizing in-depth and holistic
inquiry into social phenomena in a natural context, forming conclusions and theories
from primary data, gaining an interpretive understanding of their behavior and
meaning construction through interaction with the research subjects. In addition to
empirical research and normative research, there exists a category of conceptual
research, which cannot be placed in the same category as the above two types of
research at the philosophical level. On the one hand, conceptual research seeks to
explain the concept of the research object, the theory formed based on deductive
reasoning, so conceptual research can be classified under normative research. On the
other hand, it also draws on empirical content, including quantitative and qualitative
research, as Kant said, “ pure intuition without pure category is blind,” and these
empirical data are combined with the researcher’s observation and reflection based
on concepts to obtain normative interpretations. With the rapid development of
scientific research and the emergence of a large number of statistical analysis
tools, “we may therefore be more certain of the empirical character of any given
bit of knowledge than we are of the principle from which that knowledge could be
deduced” (Benjamin 1941). It is conceivable that empirical research is gradually
occupying an essential place in social science research.

4 Conclusion: Toward a PLS-SEM Empirical Approach

to Educational Research

The purpose of this chapter is to summarize the application and development of

PLS-SEM research methods, an analytical tool for empirical research in educational
research. PLS-SEM has been widely used in social science disciplines, and a review
of previous literature on PLS-SEM suggests that it should be the preferred approach
in the following situations. First, when the research objective is to better understand
increasing complexity by exploring extensions of established theories, and the other
case is when the path model includes one or more formatively measured constructs
(Hair et al. 2019a). Secondly, compared with covariance-based structural equation
modeling (CB-SEM), PLS-SEM is a causal-predictive approach that focuses on the
explanation of variance in the dependent variables (Hair et al. 2019a). Moreover,
PLS-SEM is not only suitable for exploratory research but also confirmatory
research (Hair et al. 2022). Strictly speaking, educational research does not belong
to the field of social science research. The basic task of education is to cultivate
human beings, and the cultivation mainly involves the cognitive, behavioral, and
moral development of students, which is closely related to the epistemological,
methodological, and ethical aspects at the philosophical level. Therefore, the field
of education should not only explore the questions of “what” and “how,” more
importantly, the question of “why,” and a deeper understanding of essential resto-
ration is inevitable. The deeper understanding or essence restoration must be closer
380 H. Zhang et al.

to normative research and qualitative interpretation in empirical research. From

another perspective, education is inseparable from social science, because many
questions in the educational field are economic, political, social, legal, and so forth.
Consequently, educational research can be cross-fertilized with other social science
fields and even with natural sciences, and the quantitative research methods advo-
cated by empiricism are similarly applied. As a result, researchers can use more
scientific and sophisticated technical means to deal with various connections
between educational phenomena as objective existences. From the development of
the last decade, PLS-SEM, as a new multivariate statistical data analysis tool,
realizes the simultaneous completion of regression modeling, data structure simpli-
fication, and correlation analysis between two sets of variables under one algorithm.
The research of Hair et al. (2019b) noted that the small sample size capability of
PLS-SEM was confirmed to be real, models with a formal specified structure should
be analyzed using PLS-SEM and it was superior to regression analysis when
evaluating mediation. A study by Boubker et al. (2021) analyzed the impact of
entrepreneurship education on the entrepreneurial intentions of Moroccan university
students by using PLS-SEM to test the proposed model and found that the entrepre-
neurial intentions of 98 management students depend mainly on two variables,
entrepreneurship education, and entrepreneurial attitudes. Gora et al. (2019) ana-
lyzed the data by surveying students from two Romanian public universities using
PLS-SEM, thus proposing a model that links the quality management characteristics
of higher education institutions to students’ competencies and labor market employ-
ment opportunities and emphasizing the direct and indirect relationships between
them. In the face of complex and changing educational phenomena, the scientific
empirical model of PLS-SEM undoubtedly has advantages over other mathematical
and statistical methods and has led to the rapid development of theories, methods,
and applications in the field of educational research.

Acknowledgments This work was supported by Macao Polytechnic University (RP/FCHS-02/

2022).

References

Benjamin AC (1941) Is empiricism self-refuting? J Philos 38(21):568–573

Boubker O, Arroud M, Ouajdouni A (2021) Entrepreneurship education versus management
students’ entrepreneurial intentions. A PLS-SEM approach. Int J Manag Educ 19(1):100450
Gora AA, Ștefan SC, Popa ȘC, Albu CF (2019) Students’ perspective on quality assurance in higher
education in the context of sustainability: a PLS-SEM approach. Sustainability 11(17):4793
Grafstein R, Dobbs D (1988) Rationalism or revelation? Am Polit Sci Rev 82(2):579–587
PLS and Educational Research: Epistemological and. . . 381

Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019a) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Sarstedt M, Ringle CM (2019b) Rethinking some of the rethinking of partial least squares.
Eur J Mark 53(4):556–584
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks, CA
Rosenthal SB, Buchholz RA (2000) The empirical-normative split in business ethics: a pragmatic
alternative. Bus Ethics Q 10(2):399–408
Public Higher Education Organizational
Climate’s Structural Model

Joel Bonales-Valencia

1 Introduction

The investigation considered two scopes, a descriptive one because it sought to

specify the characteristics of the variables under study, resulting from the review of
the facts and data compiled in the Institution, which allowed interpreting its reality in
relation to the latent variables, and the second was the structural model in order to
measure the degree of relationship between the endogenous variables and the
exogenous variables, as well as the degree of association of the variables that
determine the organizational climate.
To carry out the study, six latent variables were selected: salary, promotions,
leadership, motivation, and gender equity and the organizational climate. Based on
these variables, dimensions, and indicators (Table 1), a survey was developed with
the purpose of being applied to teachers who work in an HEPI in Mexico.
The objective of this research is to create a Structural Model with the latent
variables of salary, promotions, leadership, motivation, and gender equity of an
HEPI, using the PLS-SEM technique.
The hypothesis is, the organizational climate of HEPI depends on salary, the
granting of promotions to teachers, decisions based on gender equity, labor moti-
vation, and leadership.
The HEPI that served as a case study develops teaching activities at a higher level
since 1965. The courses it offers are: ten Engineering, two bachelor’s degrees, ﬁve
master’s degrees and one Doctorate.
The object of study is composed of 335 teachers of the Institution, of which
65 have Doctorate, 101 masters, 19 Specialization, and 150 bachelor. For the

J. Bonales-Valencia (✉)
Michoacan University of Saint Nicolas of Hidalgo, Morelia, Michoacán, México
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 383

Table 1 Operationalization of latent variable

Latent variable Indicator Key
Salary Complexity SC1
Performance SP2
SP5
Economic characteristics SE7
Labor market SL3
SL4
Government provisions SG9
Satisfactory SS6
SS8
Promotions Employee’s merit PM11
PM13
Antiquity PA10
Time PT14
PT15
Knowledge PK17
Experience PE18
Skills PS19
Promotions’ information PP12
PP16
Gender equality Inequality GI30
GI34
Discrimination GD31
GD33
GD32
Motivation Power MP35
Achievement MA36
MA37
Membership MM38
Compensations MC39
Promotions MT40
Incentives MI41
Leadership Task execution LT42
Team activities LT43
Worker’s participation LA47
Functions’ delegation LW44
Decision making LF45
Employees’ orientation LD46
Tasks’ orientation LE48
Goals LO49
LO50
LG51
(continued)
Public Higher Education Organizational Climate’s Structural Model 385

Table 1 (continued)
Latent variable Indicator Key
Organizational climate Rules CR21
Common interest CC25
Goals CG24
Physical proximity CP27
Role CL22
CL23
Integration CI19
C120
Cultural similarity CS28
Harmony CH26
CH29

diagnosis, a survey was constructed with 51 items under the Likert type format.
According to the sample size of 335 teachers, with a maximum error of 5% and a
reliability of 95%, the consultation (Mitofsky 2009) was used, giving a result of
93 teachers to survey.

2 Methodology

2.1 Organizational Climate

The organizational climate can be described as “the subjective effects received from
the system that form the informal style of the administrators and other important
environmental factors about the activities, beliefs, values and motivation of the
people who work in a given organization.”
Based on the above, the organizational climate is composed of several compo-
nents and this multidimensional nature is important to create a questionnaire and
proceed to the evaluation in an HEPI (Table 1).

2.2 Partial Least Square

PLS-SEM is a method that bears some relation to principal components regression;

instead of ﬁnding hyperplanes of maximum variance between the response and
independent variables, it ﬁnds a linear regression model by projecting the predicted
variables and the observable variables to a new space. Because both the X and Y data
are projected to new spaces, the PLS-SEM family of methods are known as bilinear
factor models.
386 J. Bonales-Valencia

PLS-SEM is used to find the fundamental relations between two matrices (X and
Y), a latent variable approach to modeling the covariance structures in these two
spaces.
The basic PLS-SEM algorithm follows a two-stage approach. In the first stage,
the latent constructs’ scores are iteratively estimated via a four-step process (Roldán
and Sánchez-Franco 2012). The second stage calculates the final estimates of
coefficients (outer weights, loadings, and path coefficients) using the ordinary least
squares method for each partial regression in the model. Applying both processes,
the PLS-SEM algorithm aims to minimize the residual variances of dependent
variables. Accordingly, PLS-SEM rests on a main assumption, that is, the predictor
specification, which forms the basis for PLS-SEM modeling (Hair et al. 2011).
Hence, the cause-and-effect directions between all the variables need to be specif-
ically defined. Besides, PLS-SEM follows a segmentation process, so, its estimates
for a particular construct are limited to the immediate blocks to which it is
structurally tied.

2.2.1 Model of Structural Equations

The model of structural equations with PLS-SEM is a statistical technique that

allows the calculation of simultaneous estimation of a set of equations in which
the concepts (measurement model) and the relationships between them are measured
(structural model) and has the capacity to deal with concepts that are not directly
observable (Cepeda-Carrión et al. 2022).
Based on Barclay et al. (1995), PLS-SEM is an iterative combination of major
component analysis linking measurements to constructs and path analysis that
allows the development of a construction system. The relationships between indi-
cators and constructs are guided by the theory, and the parameter estimation of
relationships between constructs is performed with Ordinary Least Squares (OLS).
So PLS-SEM seeks the prediction of dependent variables, both latent and manifest,
so it seeks to maximize the explained variance R2 of dependent variables. Unlike
traditional methods based on covariances (MBC), PLS-SEM is better suited in
predictive applications and theory development or exploratory analysis; however,
it can also be used for theory confirmation (Lamberti et al. 2017).
Chin et al. (2003) mention that PLS-SEM is a powerful method of analysis
because its requirements for the measurement of variables, sample size, and distri-
butions are minimal. PLS-SEM avoids two serious problems: improper or inadmis-
sible solutions and indeterminate factors.
PLS-SEM offers a significant advantage over covariance-based (MBC) models,
PLS-SEM models allow operating with both reflective (e.g., personality traits,
attitudes) and formative indicators (measures that give rise to a latent theoretical
construct, for example, the social status construct that includes the indicators occu-
pation, income, place of residence), while the MBC model only operates with
reflective indicators (Cepeda-Carrión and Roldán 2004).
Public Higher Education Organizational Climate’s Structural Model 387

As for the relationships presented in the variables, they can be recursive, when
they only go in a single direction, or not recursive, when they are bidirectional.
MBCs accept both relationships, while PLS-SEM only accepts one-way
relationships.
Finally, Falk and Miller (1992) mention that MBC seeks to ﬁnd structural or
functional parameters that explain how the world works, they seek a causality
statement, a description of causal mechanisms (rigid modeling). The main problem
with this technique is restrictive assumptions, data distribution, measurement levels
and large sample size.

3 Results and Discussion

The Likert scale was used for processing data; this scale is a very useful tool as it is
designed to measure attitudes. From the arithmetical point of view, it is a summation
scale and the score or measure of each person in the attitude in question is given by
the sum of their responses to questions that were implemented. For this study, we
work with 29 items to apply on 93 HEPI professors, which were reviewed in detail
and were developed according to each study variable. It can obtain in different kind
of score for each variable between the maximum and minimum values as shown in
the Table 1.
After the estimation of the model, SmartPLS opens the results report per default.
The results presentations in the Modeling window give us a first overview of the
outcomes. As shown in Fig. 1, we see the standardized outer weights for the
formative measurement models. We see the path coefficients as well as the R2 values
of the endogenous constructs (shown in the circles) and the discriminant validity.
Then each of the measurements that are presented in the construction of this model
will be explained.
Table 2 shows the Coefficient of Determination (R2). The most used to evaluate
the structural model. This coefficient is a measure of the model’s predictive power
and is calculated as the squared correlations between a specific endogenous con-
struct’s actual and predicted values. The coefficient represents the exogenous latent
variables’ combined effects on the endogenous latent variable.
The Leadership (0.387) and Motivation (0.228) can be considered moderate,
whereas the R2 value of Climate (0.554) is rather strong. In addition, we can observe
the same manner with the R2Adj.
Table 3 shows the results report for the path coefficients in matrix format. The
table reads from the row to the column. The value of 0.622 is among the latent
variables Motivation and Leadership. Climate column is the standardized path
coefficient of the relationship from the Genders. Moreover, the three constructs
show the internal consistency reliability of each construct (Table 4).
Table 5 shows that the highest correlations are those between organisational
climate and a) Gender (0.703), b) Promotions (0.569), and c) Leadership (0.554)
(see Henseler et al. 2015).
388 J. Bonales-Valencia

Fig. 1 Structural model of the latent variables PLS-SEM algorithm

Table 2 Coefﬁcient of determination (R2)

Variable R2 R2adj
Climate 0.554 0.544
Leadership 0.387 0.381
Motivation 0.228 0.210

Table 3 Path coefﬁcients

Variable I III
I. Climate
II. Gender 0.569
III. Leadership 0.277
IV. Motivation 0.622
V. Promotions
VI. Salary

4 Conclusions and Implications for Theory and Practice

Finally, some important conclusions can be drawn on the quality variables of the
exporting companies using the Partial Least Square model.
Public Higher Education Organizational Climate’s Structural Model 389

Table 4 Construct reliability and validity

Cronbach’s Composite Average variance extracted
Variable alpha rho_A reliability (AVE)
Climate 0.920 0.923 0.936 0.677
Gender 0.541 0.618 0.739 0.497
Leadership 0.831 0.858 0.878 0.552
Motivation 0.607 0.792 0.748 0.439
Promotions 0.705 0.860 0.812 0.457
Salary 0.772 0.795 0.813 0.360

Table 5 Latent variable correlations

Variables I II III IV V
I. Climate
II. Gender 0.703
III. Leadership 0.554 0.486
IV. Motivation 0.419 0.454 0.622
V. Promotions 0.569 0.497 0.547 0.441
VI. Salary 0.370 0.316 0.335 0.386 0.523

The teachers of the HEPI are interested in the fact that this type of organizational
climate measurements is carried out mainly with the purpose of having some
situations corrected by the managers of the same.
We studied what related to the Theory of Organizational Climate, Human Rela-
tions and Systems Theory, as well as the theoretical part of each of the six latent
variables, from which their dimensions and indicators were extracted to carry out the
field study. Therefore, the result obtained from the general hypothesis is validated in
its entirety.
The Leadership (0.387) and Motivation (0.228) can be considered moderate,
whereas the R2 value of Climate (0.554) is rather strong.
The field work responded to the general objective that was proposed in the sense
of providing concrete results endorsed with opinions and points of view of the
teachers who collaborated in this investigation.
Although the salary does not have a strong impact on the organizational climate,
there are disagreements among teachers due to the delay in the payment of benefits.
After the performance evaluation, teachers would be motivated if the institution
grants them financial compensation for their work, to reward those who achieve
good results.
The correlation of latently variable variables, which had the most impact was
Gender with 0.703, Promotions with 0.569 and Leadership with 0.554.
Integrating a work team that is responsible for detecting the problem of Organi-
zational Climate existing in the HEPI.
390 J. Bonales-Valencia

References

Barclay D, Higgins C, Thompson R (1995) The partial least squares (PLS-SEM) approach to casual
modeling: personal computer adoption and use as an Illustration. Technol Stud 2:285–309
Cepeda-Carrión G, Roldán JL (2004) Aplicando en la práctica la técnica PLS-SEM en la
administración de empresas. In: Conocimiento y Competitividad, XIV Congreso Nacional
ACEDE, Murcia. https://idus.us.es/handle/11441/76333. Accessed 13 Mar 2022
Cepeda-Carrión G, Hair JF, Ringle CM, Roldán JL, García-Fernández J (2022) Guest editorial:
Sports management research using partial least squares structural equation modeling
(PLS-SEM). Int J Sports Market Sponsorship 23(2):229–240
Chin WW, Marcolin BL, Newsted PR (2003) A partial least squares latent variable modeling
approach for measuring interaction effects: Results from a Monte Carlo simulation study and an
electronic-mail emotion/adoption study. Inf Syst Res 14(2):189–217
Falk RF, Miller NB (1992) A primer for soft modeling. University of Akron Press, Akron
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: Indeed, a silver bullet. J Market Theory Pract
19(2):139–152
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Market Sci 43(1):115–135
Lamberti G, Banet Aluja T, Sanchez G (2017) The Pathmox approach for PLS-SEM path modeling:
Discovering which constructs differentiate segments. Appl Stochastic Models Bus Ind 33(6):
674–689
Mitofsky WJ (2009) http://72.52.156.225/Tamaño-muestra.aspx. Accessed 23 May 2010
Roldán JL, Sánchez-Franco MJ (2012) Variance-based structural equation modeling: Guidelines for
using partial least squares in information systems research. In: Mora M, Gelman O,
Steenkamp A, Raisinghan M (eds) Research methodologies, innovations and philosophies in
software systems engineering and information systems. IGI Global, Hershey, PA, pp 193–221
An Investigation of Predictive Relationships
Between University Students’ Online
Learning Power and Learning Outcomes
in a Blended Course

Yue Zhu, Ming Hua Li, Lu Li, Rong Wei Huang, and Jia Hua Zhang

1 Introduction

With the advance of the Internet technologies, people are provided with a ﬂexible
access to education through an online or blended learning platform. However, they
are facing challenges to successfully accomplish their learning goals, as learning in
such an environment requires them to be well-scheduled and remain active, such as
making the ﬁrst move to learning, determining their needs, expressing their learning
goals clearly, managing resources and time for learning, applying proper learning
strategies, and achieving satisfying learning outcomes (Gedik et al. 2012). In this
sense, to be an effective learner, knowing how to learn would be critical for people’s
success in a complex and challenging online or blended learning environment.
The concept of learning power was developed in the late 1990s and the assess-
ment tool (i.e., the Effective Lifelong Learning Inventory ELLI) has tested the
constructs in relation to motivated and effective learners’ dispositions, attitudes,
and values (Deakin Crick et al. 2015). Recently, the ELLI has been applied to assess
university students’ learning capability in China (Li et al. 2018; Ren 2012; Ren and

Y. Zhu (✉)
Key Laboratory of Intelligent Education Technology and Application of Zhejiang Province,
Zhejiang Normal University, Jinhua, Zhejiang, China
M. H. Li · J. H. Zhang
Zhejiang Normal University, Jinhua, Zhejiang, China
e-mail: [email protected]
L. Li
Yantai Zhifu Wanhua Primary School, Zhifu District, Yantai, Shandong, China
R. W. Huang
Linyi Hedong District Tangtou Neighborhood Gegou Central Primary School, Hedong District,
Linyi, Shandong Province, China

© Springer Nature Switzerland AG 2023 391

Small 2013). Ren and Small (2013) interpreted the ELLI and provided suggestions
for teachers’ pedagogical design in online courses. Based on the ELLI, Li et al.
(2018) developed a set of instruments measuring Chinese university students’ online
learning power and approved the validity and reliability of these instruments.
However, few studies have been conducted to test how the factors of online learning
power are related to each other and how these factors will affect university students’
course outcomes in a blended learning environment. Therefore, we further tested the
validity and reliability of the questionnaire instruments developed by Li et al. (2018)
among the population of Chinese university students. Moreover, we examined the
inner relationships between the factors of online learning power and the predictive
relationships between online learning power and university students’ course
outcomes.

2 Background

In the era of information and the Internet technologies, we have witnessed a dramatic
shift in the way that people learn. The collection, storage, and manipulation of data
or information can be digitized and virtually automated by emerging technologies
and ﬁnally be instantly and widely available through networked access for all
learners (Deakin Crick 2007). People can receive education from various online
learning platforms, such as Moodle, edX, and Sakai, because the Internet technol-
ogies have broken down physical and geographical barriers set by traditional face-to-
face learning mode. Educational sectors at different levels (e.g., universities, voca-
tional educational organizations, and private training sectors) are offering courses in
a blended or online learning environment. As a result, people can undertake educa-
tion anywhere and at any time to suit their needs, which provides a chance for their
lifelong learning for the purpose of self-improvement or professional development.
In the context of lifelong learning supported by a ﬂexible access to online or
blended courses, it is critical for learners to “know-how” and “know-why,” rather
than “know-what” (Deakin Crick 2007). Therefore, a person’s capability of self-
awareness and taking responsibility as a learner is a key to success in such a
challenging learning environment. Learning power is regarded as peoples’ need to
want to learn, awareness of themselves as learners, capacity of being responsible for
their own learning trajectories, and maintaining their cognitive and affective self-
regulation of learning action, whether in or out of school and over a lifespan (Deakin
Crick 2007; Hautamäki et al. 2002).
Claxton proposed the ELLI and considered learning as a consequence of the
interaction between learning willingness (attitude, value, emotion, intention, and
motivation) and learning outcomes (acquired knowledge, skills, and understanding)
(Deakin Crick et al. 2004). Learning power is explained as “a form of consciousness
characterized by particular dispositions, values and attitudes, expressed through the
story of our lives and through the relationships and connections we make with other
people and our world” (Deakin Crick 2006, p. 5). In the ELLI project, seven
An Investigation of Predictive Relationships Between University Students’. . . 393

dimensions of learning power were identiﬁed as changing and learning, critical

curiosity, meaning making, creativity, learning relationships, and strategic aware-
ness (Deakin Crick 2006). Alternatively, Kirby (2005) found that learning power
contained learning motivation, learning attitude, learning ability, learning efficiency,
creative thinking, and creative ability. Based on the concept of learning power
theorized by Deakin Crick (2007), Claxton (2007), and Kirby (2005), Li et al.
(2018) developed 56 items after two-round questionnaire surveys among 1080 and
1986 university students undertaking online courses in China. The result of factor
analysis showed that there were five dimensions of online learning power—resil-
ience, reflection, resourcefulness, relationships, and learning motivation. Li et al.
(2018)’s research approved the validity and reliability of their instruments to inves-
tigate Chinese students’ learning power in an online learning environment.
Some existing research explored how to assist students to build learning power.
Stoten (2013) introduced Building Learning Power (BLP) as a new philosophy and
framework for learning and teaching in a college. Stoten concluded that BLP
challenged students and teachers’ beliefs about learning and there were mixed
responses from teachers that BLP is yet to be fully embedded. Thus, changing
entrenched learning behaviors seemed to take longer than initially expected. Based
on the concept of the ELLI, Ren and Small (2013) provided suggestions for students
to enhance their learning capacity and described a framework for curriculum and
pedagogy reformation in schools of China. Deakin Crick (2007) emphasized a few
pedagogical themes to promote the development of students’ learning power,
including (a) teacher commitment to learner-centered values and willingness to
make professional judgments, (b) positive interpersonal relationships,
(c) developing a language of learning, (d) modeling and imitation, (e) learning
dialogue, (f) reflection, (g) learner self-awareness and ownership, (h) students’
responsibility for making choices, (i) sequencing of learning materials, and (j) a
toolkit of skills and strategies for learning how to learn.
Although the concept and constructs of learning power in both face-to-face and
online learning environments have been well interpreted in the reviewed research, it
still remains unclear about how the factors of students’ online learning power will
relate to each other. Moreover, most of the existing studies focused on the imple-
mentation of learning intervention or training sessions to build students’ learning
power. Few studies have been done on students’ learning power in a blended
learning environment and the relationships between their learning power and course
outcomes in such a learning context. Thus, we explored the inner relationships
between the factors of Chinese students’ online learning power and the effect of
online learning power on their learning outcomes in a blended course.

3 Methodology

3.1 Research Design

We administered a questionnaire survey among 62 Chinese university students who

were enrolled in a blended course. The questionnaire contained self-referring
394 Y. Zhu et al.

statements assessing the participants’ online learning power. Meanwhile, the infor-
mation of the participants’ completion of online learning tasks every week was
recorded as their online course engagement. The participants’ course results and
marks for all learning tasks were obtained at the end of the course as their learning
outcomes. We tested the construct validity of online learning power and explore the
relationships between online learning power, online course engagement, and learn-
ing outcomes in a blended learning environment.

3.2 Participants

Sixty-two junior teacher education students, who were enrolled in an 18-week

blended course offered by a Chinese university, participated in our questionnaire
survey. The course aimed to provide the students with theoretical foundation and
practice of using educational technologies in teaching. The students attended a
two-hour face-to-face tutorial, including practice on Interactive White Board
(IWB) every week in a computer lab. Additionally, all course materials and
resources were provided online and the students were able to keep communication
with their teachers and other students through email, online forum, and mobile at
their convenience.
In regard to course assessment tasks, the full mark for the course is 100. The
students were required to complete (a) weekly online learning tasks from Week 1 to
Week 17 (i.e., watching lecture video, participating in online discussion, taking
online notes, and completing online questions), (b) PowerPoint and micro-lecture
designing task, and (c) a test on IWB operation held in the computer lab. The results
of these tasks were summed up as the students’ overall result in the course (Table 1).

Table 1 Detailed information of the course assessments

Full
Course assessments Requirement mark
Overall online learning Online lecture video Duration of 15.71 h 25
tasks Online discussion Ten posts in total 5
Online notes Ten notes in total 5
Online questions Multiple-choice items 15
Designing tasks PPT designing task From 10 to 15 slides 15
Micro-lecture designing Duration of 6–8 min 15
task
Test on IWB operation in a computer lab Completing a required task on 20
IWB
An Investigation of Predictive Relationships Between University Students’. . . 395

3.3 Instruments

The scales measuring the participants’ online learning power in our study were based
on the tools developed by Li et al. (2018). In their research, the scales were approved
to be suitable for the population of university students in China with tested validity
and reliability (Cronbach’s alpha =0.99). These scales were self-referring state-
ments, which assessed the participants’ individual differences in resilience, reﬂec-
tion, resourcefulness, relationships, and learning motivation, such as “I can follow
my schedule or plan in an orderly way when I learn online,” “I will ask for help and
learn from others regarding learning strategies or methods,” and “I can do my best in
a team work.” For each item, the participants were asked to respond to the statement
on a ﬁve-point Likert scale from “1 = not like me at all” to “5 = very much like me.”
Concerning the participants’ online course engagement, the data of their comple-
tion of online learning tasks were downloaded at the end of each week. Students’
participation in online discussion and learning activities are commonly accepted as
indicators of their online course engagement (Cheng et al. 2011; Hara et al. 2000;
Picciano 2002). As explained in Table 1, the participants were required to complete
four online learning tasks. Except the task of watching online lecture video that was
due on the Sunday of Week 16, the due dates for all other online learning tasks were
set on the Sunday of the 17th week. Thus, the participants needed to arrange their
time to complete these tasks. To achieve a full score for each online learning task, the
participants should watch all lecture videos, post at least ten messages to the online
forum, submit ten online notes, and answer all online questions correctly.
The learning management system (LMS) of the course automatically recorded the
information about online learning tasks accomplished by the participants every
week, including the number of hours of online video being watched, number of
online forum messages and online notes submitted, and scores of online questions.
Meanwhile, the LMS calculated and transformed the data of the completed online
learning tasks into a total mark for every participant each week, which represented
how much work the participant had completed online and how well he or she did the
online learning tasks per week. The researchers downloaded these data by the end of
each week from Week 1 to Week 17, except Week 9, Week 11, and Week 12, as
there were public holidays in these 3 weeks. Furthermore, the researchers ranked all
participants in order of the marks they received for their weekly online learning
tasks. Therefore, the participants’ online course engagement per week was indexed
as the following variables:
• Number of hours spent watching online lecture video per week
• Number of posts sent to the online forum per week
• Number of online notes submitted per week
• Marks received for all online learning tasks accomplished per week
• Ranking for the participants according to the marks they achieved per week
As indicated in Table 1, the participants’ overall course results at the end of the
course were calculated based on the marks they achieved for all online learning
396 Y. Zhu et al.

tasks, PPT and micro-lecture designing tasks, and test on IWB operation. Therefore,
the participants’ overall course results and marks for each course assessment task
were indexed as their learning outcomes.

3.4 Procedure

After obtaining the approval from the course coordinator and the Ethics Committee
of the university, the participants were informed of the aims of the study, the
information to be gauged in the questionnaire survey, and how the data would be
collected, stored, and accessed. Finally, the participants were assured that their
personal information would not be identiﬁable in any form of the report.
The participants were asked to complete a questionnaire survey about their
capability of online learning at the beginning of the course. During the course, the
researchers accessed the online learning platform and downloaded the data of
completed online learning tasks every Sunday from Week 1 to Week 17. The
participants’ marks and rankings of weekly online learning tasks from Week 1 to
Week 17 were regarded as the variables of their online course engagement. By the
end of the course, the participants’ overall course results and marks for each online
task were obtained and indexed as their learning outcomes in the blended course.

3.5 Data Analysis

The analysis on the data from the questionnaire survey, weekly online learning tasks
that were completed, and learning outcomes were performed on the Statistical
Package for Social Science (SPSS) and SmartPLS. First of all, in order to ensure
the construct validity and reliability, factor analysis was conducted on the 62 partic-
ipants’ responses to all items of online learning power scales. Pearson correlation
was used to analyze the signiﬁcance of correlations between online learning power,
completion of weekly online learning tasks, and learning outcomes.
The predictive relationships between the above-mentioned factors were investi-
gated through using the partial least squares approach (PLS) for path modeling
(Ringle et al. 2005). Structural Equation Modeling through Partial Least Square
(PLS-SEM) enabled the researchers to trace variance through a series of regression
analyses which balanced an outer model (the items which deﬁned the constructs)
with an inner model (relationships articulated between the constructs). The outer
model employed traditional factor analysis procedures and variance extraction
methods to ensure integrity of the constructs and also computed factor scores that
were used to calculate the inner model expressed in terms of regression effects. PLS
is capable of making it more appropriate than other techniques such as multiple
regression and LISREL, in the case that the sample size is small (e.g., N = 40)
(Goodhue et al. 2006).
An Investigation of Predictive Relationships Between University Students’. . . 397

4 Results

4.1 Factor Analysis

The participants’ responses to the 59 items measuring their online learning power
were subjected to factor analysis. Generally, the items with strong factor loadings
higher than 0.5 were used for analysis purposes. As shown in Table 2, a four-factor
solution was identiﬁed and these factors were labeled as: (a) goal orientation,
(b) resilience, (c) problem solving, and (d) metacognitive awareness (KMO = 0.76).

4.2 Descriptive Results

Among the 62 participants, there were 14 male and 48 female students. Regarding
online learning power, the percentage of participants, who were above the natural
mid-point of goal orientation, problem solving, resilience, and metacognitive aware-
ness, was, respectively, 95%, 76%, 92%, 57%, and 94%. At the end of the course,
58 participants have gained the marks between 90 and 97 for their overall course
results and the results for the other participants were ranged from 73 to 86. Table 3
presents the descriptive results for the factors of online learning power and learning
outcomes.
Regarding the participants’ weekly online course engagement from Week 1 to
Week 17, Fig. 1 reveals that the participants remained active in video watching most
of the time during the semester, except in the eighth week and at the beginning and
end of the course. In regards to the task of online questions, most participants started
to work on the questions from Week 13 and answered most questions in Week 15.
Similarly, the participants submitted online notes mainly in Week 14 and Week 15.
Concerning online forum participation, it appeared that participants were more
involved in responding to others’ posts, rather than sending original posts to the
online forum. Generally, except the task of watching online lecture video, the
participants did not show strong initiation at the beginning of the course, most of
the tasks of online forum, online notes, online questions were completed later in the
semester.

4.3 Correlations Between Online Learning Power Factors,

Online Course Engagement, and Learning Outcomes

It was found that the participants’ overall course results were signiﬁcantly correlated
with most of the marks and rankings of weekly online learning tasks from Week 5 to
Week 17. The marks and rankings of weekly online learning tasks completed by the
participants later in the semester (e.g., from Week 10 to Week 16) were signiﬁcantly
398 Y. Zhu et al.

Table 2 Factor analysis: loadings for the items measuring online learning power
Loading
Factor 1: Goal orientation (AVE = 0.57, Cronbach’s alpha = 0.74)
I believe that undertaking online learning would improve my professional skills and 0.76
benefit my professional development
I believe I can expand my horizon and further develop my ability 0.78
I am more interested in development of my social ability, especially getting to know 0.44
more colleagues and friends
I need to obtain a higher educational qualification to find a better job 0.39
Factor 2: Resilience (AVE = 0.51, Cronbach’s alpha = 0.88)
I am willing to continue extending my learning while learning online. 0.73
I feel really good when I am absorbed in online learning 0.76
While learning online, I am able to concentrate on my learning tasks with no 0.68
distractions
I deeply understand learning online is challenging and demands continuous effort 0.66
I will make a plan when I learn online 0.63
I will schedule my online learning in a good order 0.79
I will predict my achievement in online learning 0.83
I am flexible to make necessary adjustment during online learning 0.71
As an online learner, I know my strengths and weaknesses 0.49
Factor 3: Problem solving (AVE = 0.66, Cronbach’s alpha = 0.83)
I am determined to find solutions for any problems emerging in online learning, no 0.49
matter how difficult it would be
I will predict what will happen and try to figure out a way to solve any problems which 0.42
might happen during online learning
I will consider whether there is any concept departing from my learning goals in an 0.55
online course
When encountering any problem, I will be aware of details and find out solutions 0.53
Factor 4: Metacognitive awareness (AVE = 0.54, Cronbach’s alpha = 0.88)
I am aware of how the course is structured and presented and what the course consists 0.79
of
I am happy to raise questions in online forum when there is anything I do not 0.43
understand
I like to ask: what and why it would be, if. . .. . . 0.50
I always see the essence through the surface and draw my own conclusion 0.67
I am able to build up a linkage between the known and unknown and acquired 0.42
knowledge and practice
I like to figure out how knowledge and course system is structured 0.92
I like to think about in which way I can learn better online 0.42
I am able to reflect on my online learning process and figure out how I can learn better 0.57
I can use imagination and multiple senses to facilitate my learning online 0.59
Except online course materials, I can use other online resources 0.44
Notes: (a) N = 62, (b) extraction method: principal component analysis, oblimin rotation
An Investigation of Predictive Relationships Between University Students’. . . 399

Table 3 Descriptive information: online learning power factors and learning outcomes
Mean Median SD Skewness Kurtosis Min Max
Goal orientation 3.94 4.00 0.55 –0.45 1.80 2 5.00
Metacognitive awareness 3.56 3.70 0.61 –1.48 3.80 1.30 5.00
Resilience 3.63 3.83 0.59 –1.02 2.30 1.56 4.78
Problem solving 3.47 3.50 0.69 –1.27 2.50 1.00 5.00
Total score of online learn- 97.72 99.10 4.08 –3.18 10.62 0 92
ing tasks
Marks of PPT designing 82.00 82.50 11.18 –6.66 49.26 0 92
task
Marks of micro-lecture 83.29 83.00 5.32 –3.94 24.97 50 90
designing task
Marks of IWB operation 97.10 100.00 5.77 –2.62 8.08 70 100
Overall course result 93.13 94.00 3.70 –3.38 14.84 73 97
Notes: N = 62

related to their overall marks of online learning tasks and designing tasks at the end
of the course. The factors of problem solving and metacognitive awareness were
found to be related to the overall marks of online learning tasks by the end of the
course (Table 4).
Moreover, a coherent construct was identiﬁed based on the following indices: the
rankings of weekly online learning tasks in Week 4 and Week 5; and the marks and
rankings of weekly online learning tasks in Week 6, Week 7, Week 8, Week
10, Week 13, Week 14, Week 15, and Week 16 (see Appendix). All these variables
were added in the construct of the SmartPLS model one by one progressively to
ensure the validity of the construct for each stage could be maintained (Ringle et al.
2005; Esposito Vinzi et al. 2010). According to the result of the construct reliability
and validity, the variables of rankings of online learning tasks in eight weeks (4, 5,
6, 7, 10, 14,15, and 16) contributed to the construct of online course engagement
(AVE = 0.50, Cronbach’s alpha = 0.88). Thus, the participants’ online course
engagement could be indexed by these factors. Pearson correlation test shows that
the factors of online power, online course engagement, results of online tasks, PPT
and micro-lecture design, and overall course results were signiﬁcantly related to each
other (Table 4).

4.4 Path Modeling of the Impact of Online Learning Power

on Learning Outcomes

The impact of the participants’ online learning power factors on their learning
outcomes at the end of the course was further investigated by the PLS-SEM
approach for path modeling (Ringle et al. 2005). The researchers assumed that the
participants’ online learning power might not only have an impact on how they were
engaged in online learning, but also inﬂuence their learning outcomes by the end of
400 Y. Zhu et al.

2.50
2.00
1.50
1.00
.50
.00
W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16
Mean of number of hours spent watching online lectur video from Week 1 to Week 17, n = 62
3

0
W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16 W17
Mean of number of online quesions completed from Week 1 to Week 17, n = 62

W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16 W17

Mean of number of online notes submitted from Week 1 to Week 17, n = 62

0
W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16 W17
Mean of number of posts submitted to online forum from Week 1 to Week 17, n = 62

0
W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16 W17
Mean of number of reponses to peers posts submitted in online forum, n = 62

W1 W2 W3 W4 W5 W6 W7 W8 W10 W13 W14 W15 W16 W17

Mean of weekly marks of online learning tasks from Week 1 to Week 17, n = 62

Fig. 1 Mean of weekly online learning tasks completed from Week 1 to Week 17
Table 4 Correlations between online learning power factors, online course engagement, and marks of course assessment tasks
1 2 3 4 5 6 7 8 9 10
1. Goal orientation – 0.49** 0.31* 0.49** –0.19 0.28* 0.00 –0.03 –0.09 0.01
2. Resilience – 0.79** 0.79** –0.12 0.45** –0.06 –0.12 0.11 0.13
3. Metacognitive awareness – 0.65** –0.02 0.44** –0.08 –0.10 0.06 0.14
4. Problem solving – –0.04 0.69** –0.08 –0.05 0.04 0.19
5. Online course engagement – 0.26* 0.26* 0.28* 0.27* 0.47**
6. Overall marks for online learning tasks – 0.03 0.07 0.06 0.53**
7. Marks for PPT designing task – 0.12 0.64** 0.71**
8. Marks for micro-lecture designing task – 0.07 0.35**
9. Marks for IWB application – 0.71**
10. Overall course results –
Notes: (a) N = 62, (b) *p < 0.05, **p < 0.01
An Investigation of Predictive Relationships Between University Students’. . .
401
402 Y. Zhu et al.

Fig. 2 The initial model of the impact of online learning power on the participants’ learning
outcomes

the course. The initial model tested is shown as Fig. 2, which presents the initially
expected relationships between online learning power, online course engagement,
and learning outcomes. This form of descriptive modeling operates on an explor-
atory basis, as it is distinct from other forms of structural equation modeling which
were based on confirmatory principles rather than regression. As a result, rather than
testing a prior theory, one advantage of the PLS approach is that it allows sequences
to be depicted in terms of descriptive relationships, which presents variance analyses
in terms of both direct and indirect effects (Zhu et al. 2016). The procedures were
reviewed in depth in the handbook by Esposito Vinzi et al. (2010).
The researchers employed a formative measurement model to further test (a) any
inner-relationships between the factors within the construct of online learning
power; and (b) the relationships between the factors of online learning power and
the participants’ learning outcomes by the end of the course. Bootstrapping in
PLS-SEM is a nonparametric procedure and less subject to violation of normality
assumption. This enables the researchers to test the statistical significance of various
PLS-SEM results, such as path coefficients (Davison and Hinkley 1997; Efron and
Tibshirani 1994; Unal and Uzun 2021). Any non-significant relationships were
trimmed off to produce a parsimonious and descriptive model.
As shown in Fig. 3, problem-solving and online course engagement directly
predicted the participants’ overall marks of online learning tasks, which in turn
affected their overall course results. As a driving force, goal orientation predicted the
participants’ overall marks of online learning tasks through the mediation by resil-
ience and problem solving. The factors of overall marks of online learning tasks and
Micro-lecture designing task could predict 38% of the variance in overall course
results. Goal orientation and resilience, respectively, explained 65% and 68% of the
variance in metacognitive awareness and problem solving. There was no mediation
by online course engagement on the relationships between the online learning power
An Investigation of Predictive Relationships Between University Students’. . . 403

Fig. 3 Final model of the impact of online learning power on the participants’ learning outcomes

factors, overall marks of online learning tasks, and marks of Micro-lecture designing
task.

5 Discussion

We employed Li et al.’s (2018) scales to assess university students’ online learning

power and a five-factor solution was identified as goal orientation, resilience,
problem solving, metacognitive awareness, and resource management. These factors
well contribute to the concept of online learning power at a significant level. This
finding adds to the construct validity of the scales of online learning power by Li
et al. (2018).
We also found that, within the construct of online learning power, goal orienta-
tion in online learning performed as a driving force to predict the participants’
resilience that impacted their metacognitive awareness and problem solving. It is
also worthwhile to note that the predictive influence of online learning power on the
participants’ overall course results was mediated through their overall marks of
online learning tasks. This finding highlights the critical role played by online
learning power in students’ academic achievement in a blended learning environ-
ment. The students with higher level of self-regulating their online learning are more
likely to receive better learning outcomes (Saba 2012; Sharma et al. 2007).
Furthermore, keeping up with online course schedule and completing online
learning tasks in a timely manner can be critical for students’ success in a blended
course. The results showed that (a) the participants’ overall marks received for online
learning tasks and Micro-lecture designing task were found to be determinants for
their overall course results; and (b) the participants with more active online course
engagement tended to achieve better course results by the end of the course. Thus, it
would be essential for online learners to well schedule their online learning activities,
stick to their learning plans, and remain motivated and active to ensure their high
level of online course input. This finding is consistent with the research result by
404 Y. Zhu et al.

Table 5 Number of the participants working on the task of watching online lecture video
W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 W13
Number of the partici- 4 28 43 48 52 59 59 59 60 60 62
pants watching online
video
Notes: N = 62, W = Week

Davies and Graff (2005) and Chen and Jang (2010), who found that the students’
online course engagement, such as frequent accessing online course content, was
very important for their success in online learning.
Furthermore, we did not identify the significant mediated relationship between
the factors of online learning power, online course engagement, and learning out-
comes. Instead, we found a direct prediction from problem solving and online course
engagement on the students’ overall marks of their online learning tasks. This
finding is different from previous research by Zhu et al. (2016), who reported that
the impact of university students’ learning goal orientation, self-management and
application of metacognitive strategies on their learning outcomes in a blended
course was mediated through online course participation indexed by online forum
contribution, weekly learning reports, and hours spent in online learning.
The potential reasons for the above non-significant mediated relationships in our
study maybe due to the abnormality of the data about the participants’ weekly online
learning tasks. The descriptive results indicated that the participants did not remain
active in the online forum participation throughout the semester. The majority of
them started to work on online forum tasks from Week 13 and most of them
responded to others’ opinions instead of sending original posts to the course
forum. Regarding the task of watching online lecture video, 34 and 19 participants
did not begin to watch the video, respectively, in Week 2 and Week 3 (Table 5). It is
assumed that the data collected from the course platform may represent certain
aspects of the participants’ online course input. But due to the delayed involvement
in online learning, such data may not indicate their online learning process in a
comprehensive way. This implies that, in addition to delivery instruction, online
course designers and teachers ought to apply different teaching strategies to motivate
students to participate in online learning on their initiative and create a supportive
and welcoming online learning community through the usage of online forum.

6 Conclusion and Implications Discussion

In summary, our research approved the validity and reliability of online learning
power scales developed by Li et al. (2018) and ensured the suitability of the scales
among the population of university students in China. More importantly, our
investigation depicted how the factors of online learning power could affect the
participants’ online learning outcomes. Although only problem solving was found to
An Investigation of Predictive Relationships Between University Students’. . . 405

signiﬁcantly predict the participants’ learning outcomes, such a predictive relation-

ship provides implications for future research using more comprehensive data, that
can well represent their online learning progress, and exploring effective teaching
methods to improve students’ online learning power and leverage their motivation to
participate in online learning activities.
In terms of improving students’ online learning power, assessing their capacity of
online learning is a good beginning. As students answer a range of questions about
themselves (e.g., the assessment tools developed by Deakin Crick et al. 2004, and Li
et al. 2018), their responses to these questions can provide information about how
they perceive themselves in relation to each of the dimensions of online learning
power. Then teachers can create students’ individual learning profile and class
learning profile to gain a good understanding of students’ online learning capability,
respectively, at an individual and group level (Deakin Crick et al. 2004).
Furthermore, teachers can create a learner-centered learning context and design
online learning activities to promote students to take responsibility of their own
learning (Deakin Crick et al. 2004). Daily routines in online learning activities can be
used to scaffold students’ ability to regulate their learning (Tarullo et al. 2009).
Online course content should be well-structured to make it easy for students to
prioritize and schedule their learning. Learning intervention and training on online
learning capacity are necessary to improve students’ effective use of self-regulation
and metacognitive and cognitive learning strategies (Barnard-Brak et al. 2010; Hu
2007; Rowe and Rafferty 2013; Tsai 2009). Teachers can also provide topics to
prompt students to participate in online discussion. Additionally, peer-review or self-
reflective questions may encourage students to review their learning process in a
web-based note-taking tool. Self-evaluation or self-assessment and peer-review can
improve students’ problem-solving skills by using critical reflective thinking skills,
identifying issues, and seeking for solutions for the issues emerging during learning
process (Bixler 2008; Chang 2007). Finally, teachers can facilitate enquiry-based
learning or project-based learning and empower students with freedom and respon-
sibility to personalize their online learning content and tools (McLoughlin and Lee
2010; Saba 2012). With self-control on their learning tasks and pace, students can be
taught with various learning strategies, including researching, applying critical
thinking, managing information and resources, problem solving, and keeping resil-
ient (Zhu et al. 2016). Continuous and constructive feedbacks will benefit students to
keep their online learning on the right track and be aware of the use of cognitive and
metacognitive strategies.
In summary, students’ online learning power is critical for their success in a
blended learning environment. Such ability can be trained and improved, but
teachers need to invest a great amount of effort to design and implement learning
intervention (Stoten 2013). It would be critical for university educators to foster
students’ ability to control their own learning and continue to improve themselves to
be effective learners to derive the benefits provided by blended courses, but also
have confidence to face and deal with challenges emerging during blended learning.
406 Y. Zhu et al.
Appendix
An Investigation of Predictive Relationships Between University Students’. . . 407

References

Barnard-Brak L, Lan WL, Osland Paton V (2010) Profiles in self-regulated learning in the online
learning environment. Int Rev Res Open Dist Learn 11(1):61–80
Bixler BA (2008) The effects of scaffolding student’s problem-solving process via question
prompts on problem solving and intrinsic motivation in an online learning environment. PhD
Thesis, The Pennsylvania State University
Chang M-M (2007) Enhancing web-based language learning through self-monitoring. J Comput
Assist Learn 23(3):187–196
Chen KC, Jang SJ (2010) Motivation in online learning: Testing a model of self-determination
theory. Comput Hum Behav 26(4):741–752
Cheng CK, Paré DE, Collimore L-M, Joordens S (2011) Assessing the effectiveness of a voluntary
online discussion forum on improving students’ course performance. Comput Educ 56:253–261
Claxton G (2007) Building learning power: Helping young people become better learners. TLO,
Bristol
Davies J, Graff M (2005) Performance in e-learning: Online participation and student grades. Br J
Educ Technol 36(4):657–663
Davison AC, Hinkley DV (1997) Bootstrap methods and their application. Cambridge University
Press, Cambridge
Deakin Crick R (2006) Learning power: What it is? In: Deakin Crick R, Millington D, Small T,
Staricoff M (eds) Learning power in practice: A guide for teachers. Paul Chapman, London, pp
1–14
Deakin Crick R (2007) Learning how to learn: The dynamic assessment of learning power. Curric J
18(2):135–153
Deakin Crick R, Broadfoot P, Claxton G (2004) Developing an effective lifelong learning inven-
tory: the ELLI project. Assess Educ Princ Policy Pract 11(3):247–272
Deakin Crick R, Huang S, Shafi AA, Goldspink G (2015) Developing resilient agency in learning:
the internal structure of learning power. Br J Educ Stud 63(2):121–160
Efron B, Tibshirani RJ (1994) An introduction to the Bootstrap. Chapman and Hall/ CRC,
New York
Esposito Vinzi V, Chin WW, Henseler J, Wang H (2010) Handbook of partial least squares.
Springer, Heidelberg
Gedik N, Kiraz K, Özden MY (2012) The optimum blend: Affordances and challenges of blended
learning for students. Turk Online J Qual Inq 3(3):102–117
Goodhue D, Lewis W, Thompson R (2006) PLS, small sample size, and statistical power in MIS
research. In: Proceedings of the 39th Hawaii international conference on system sciences, 4–6
Jan 2006, Kauai, USA
Hara N, Bonk CJ, Angeli C (2000) Content analysis of online discussion in an applied educational
psychology course. Instr Sci 28(2):115–152
Hautamäki J, Arinen P, Niemivirta MJ, Eronen S, Hautamäki A, Kupiainen S et al (2002) Assessing
learning-to-learn: a framework. Opetushallitus, Helsinki
Hu H (2007) Effects of SRL strategy training on learners’ achievement, motivation and strategy use
in a web-enhanced instructional environment. In: Dissertation abstracts international
Kirby WC (2005) The power of learning. Thousand Publishing House, Haikou
Li B-M, Gong L-L, Zhu Z-T (2018) Development and verification of online learning power
assessment tools for online learners. Open Educ Res 24(3):77–84
McLoughlin C, Lee MJW (2010) Personalised and self-regulated learning in the Web 2.0 era:
International exemplars of innovative pedagogy using social software. Australas J Educ Technol
26(1):28–43
Picciano AG (2002) Beyond student perceptions: Issues of interaction, presence, and performance
in an online course. Online Learn (formerly the Journal of Asynchronous Learning Networks)
6(1):21–40
408 Y. Zhu et al.

Ren K (2012) The assessment tool of learning power and its application in reversing underachieve-
ment at secondary schools: Research design. Contemp Teach Educ 5(1):35–40
Ren K, Small T (2013) How to develop students’ learning power. Curric Teach Mater Method
33(2):29–35
Ringle CM, Wende S, Will A (2005) SmartPLS 2.0(M3) Beta. University of Hamburg, Hamburg
Rowe FA, Rafferty JA (2013) Instructional design interventions for supporting self-regulated
learning: Enhancing academic outcomes in postsecondary E-learning environments. MERLOT
J Online Learn Teach 9(4):590–601
Saba T (2012) Implications of E-learning systems and self-efﬁciency on students’ outcomes: A
model approach. Hum-Centric Comput Inf Sci 2(6):1–11
Sharma S, Dick G, Chin W, Land LPW (2007) Self-regulated learning and e-learning. In: Pro-
ceedings of the ﬁfteenth European conference on information systems, 7–9 June 2007, Univer-
sity of St. Gallen, Switzerland
Stoten DW (2013) Transforming learning and teaching: Introducing building learning power into a
sixth form college. Educ Stud 39(2):235–238
Tarullo AR, Obradovic J, Gunnar MR (2009) Self-Control and the developing brain. Zero Three
29(3):31–37
Tsai M-J (2009) The model of strategic e-learning: Understanding and evaluating student elearning
from metacognitive perspectives. Educ Technol Soc 12(1):34–48
Unal E, Uzun AM (2021) Understanding university students’ behavioral intention to use Edmodo
through the lens of an extended technology acceptance model. Br J Educ Technol 52(2):
619–637
Zhu Y, Au W, Yates G (2016) University students’ self-control and self-regulated learning in a
blended course. Internet High Educ 30:54–62
The Effects of Marketing Orientation
on the Performance of Higher Learning
Institutions in Tanzania: Staff and Students’
Perceptions

Francis Muya and Hawa Tundui

1 Introduction

Studies on market orientation in higher learning institutions (HLIs) have received an

increased attention in the current decade (Anabila et al. 2019; Domański 2014;
Chandler et al. 2021; Khuwaja 2016; Khuwaja et al. 2015; Mokoena 2018;
Niculescu et al. 2013; Sefnedi 2017; Rivera-Camino and Molero Ayala 2010; Starck
and Zadeh 2013; Tjahjadi et al. 2022). The attention is due to the challenges in the
environment where HLIs operate (Akonkwa 2009; Beneke and Human 2010; Mok
2001; Mokoena 2018; Nicolescu 2009), and being marketing-oriented is seen as the
best survival mechanism (Hilman and Kaliappen 2014; Moorman and Rust 1999;
Rivera-Camino and Molero Ayala 2010; Tabaku and Mersini 2013).
Marketing orientation offers a number of beneﬁts such as a better understanding
of customers and competitors (Dolnicar and Lazarevski 2009; Newman 2002), the
possibility of increasing market share, proﬁt and innovativeness and of becoming
competitive both domestically and internationally (Hemsley-Brown and Oplatka
2006; Maringe 2004, 2005, 2006; Stachowski 2011; Tajeddini et al. 2006). Other
possibilities include realizing institutional goals, such as surviving as an organiza-
tion, increasing institutional professional reputation, improving facilities and faculty
and increasing enrolment and endowment (Ivy 2001; Nguyen and LeBlanc 2001;
Polat and Donmez 2010) and adapting to demands in the society (Caruana et al.

F. Muya (✉)
Department of Business and Entrepreneurship Studies, National Institute of Transport, Dar es
Salaam, Tanzania
e-mail: [email protected]
H. Tundui
School of Business, Mzumbe University, Morogoro, Tanzania
e-mail: [email protected]

© Springer Nature Switzerland AG 2023 409

1998). Furthermore, marketing orientation can increase accessibility to ﬁnancial

resources (Caruana et al. 1998; Narver and Slater 1990; Svenson and Wood 2007),
students’ retention (Ackerman and Schibrowsky 2007; Lichiello 2014; Shepherd
2008) and loyalty among students in the HLIs, hence lowering the costs of attracting
new ones (Rowley 2003). Despite all these marketing orientation–related benefits,
there is however no well-established, collectively agreed and acknowledged links
between marketing orientation and performance in the higher education context
(Eikenberry and Kluver 2004; Dolnicar and Lazarevski 2009; Rotfeld 2008).
Although studies on marketing orientation are many on industries such as
manufacturing, pharmaceutical, retailing and tourism, studies on HLI marketing
orientation in Tanzania are still very limited. In addition, the marketing orientation
studies carried out on HLI in different parts of the world dwelled on universities
(Anabila et al. 2019; Domański 2014; Khuwaja 2016; Khuwaja et al. 2015;
Niculescu et al. 2013; Sefnedi 2017; Rivera-Camino and Molero Ayala 2010; Starck
and Zadeh 2013), ignoring other HLIs such as technical institutions which also
operate under the same environment. Furthermore, many studies concentrated on
testing the direct links between marketing orientation and performance (Burgess and
Nyajeka 2007; Niculescu et al. 2013; Zebal and Goodwin 2012), ignoring the
contribution of indirect effects through mediating constructs in refining and under-
standing causal relationship (Wu and Zumbo 2008).
The effects of marketing orientation as revealed in the past studies cannot be
generalized across sectors and countries with different economic, political and
cultural structures, hence inspiring studies on country specific and sectoral basis,
especially in HLI contexts in Tanzania. This chapter, therefore, intends to fill the
identified gaps by assessing the direct and indirect relationship between marketing
orientation and performance in Tanzania HLIs, supporting the arguments that
continual survival and success of HLI hinge upon being marketing-oriented.

2 Literature Review

The foundation of this chapter is built on theories that view marketing as essential for
optimal organizational performance, especially marketing mix theory (Al-Fattal
2010; Baker and Saren 2010; Borden 1984; Filip 2012; Ivy 2008; Goi 2009),
which includes components of programme, place, promotion, price, process and
physical facilities, and market orientation theory (Felcman 2012; Kiessling et al.
2015; Kohli and Jaworski 1990; Narver and Slater 1990; Niculescu et al. 2013),
which includes components of administration, mentoring and intelligence genera-
tion. Wymer et al. (2006) viewed marketing orientation as an approach to managing
an organization through marketing tactics to achieve the organization’s goals and
objectives. According to Kurtinaitienė (2005), marketing orientation (sometimes
called market orientation) contributes to the understanding and implementation of
the marketing concept. These explanations relate to those provided by Perreault and
McCarthy (2002, p. 35) who view marketing orientation as “trying to carry out the
The Effects of Marketing Orientation on the Performance of Higher. . . 411

marketing concept instead of just trying to get customers to buy what the ﬁrm has
produced.” The relationship between marketing orientation elements and
performance-related aspects is diverse as hereunder explained.

2.1 Marketing Orientation and Retention

While exploring the application of marketing thinking to students’ retention, Shep-

herd (2008) maintained that marketing is essential for retaining students. This is
possible through persuading students using a variety of means. The findings
align with those in a study by Beck (2007) who found that market orientation
positively affected retention at two selected public Carnegie High Research Activity
Institutions. The findings imply that organizations committed to a market-oriented
culture will likely produce better retention results than those who are not.
In connection to the price element of marketing orientation, Polat and Donmez
(2010) found that price-related activities are essential in a firm–customer relation-
ship. Hanover Report (2014) found redesigning of scholarship and bursary as
essential for students’ retention. In the same vein, Hasan et al. (2009) found that
tuition fees and flexible tuition approaches are highly important students’ costs-
related aspects in higher education. Carabajal (2012) found that financial aid is a
number one factor influencing students’ decision as to whether to enrol to a partic-
ular university or not. The findings are in line with the findings in a study by
Ackerman and Schibrowsky (2007) who found that financial bonds, such as scholar-
ships, work–study opportunities, transportation, on-campus housing, tuition
discounting, financial aid package, subsidized student health insurance programmes
and no-interest emergency loans, are essential for students’ retention.
In connection to physical facilities as an element of marketing orientation,
Mokaya (2013) revealed that well-spaced classrooms, adequate and ample spacing
in the libraries and adequate science laboratories are among the key factors in
academic achievement and retention. Similar results are reported by Lau (2003)
who found that dormitories, study rooms, facilities for the disabled, career centres,
social and professional organizations and computer technology are essential physical
facilities in students’ retention in any education setting. In relation to the effects of
the programme element of marketing orientation, Mokaya (2013) considered ade-
quate participation in cocurricular activities as a very important aspect. These results
are also supported by Ivy (2008) who viewed aspects such as the design of the
degree, properly developed and adapted to meet students’ needs, curriculums and
programme duration as important programme elements in HLI contexts. In addition,
Rowley (2003) considered refocusing on the curriculum to make it more vocation-
ally relevant as an important factor in students’ retention. Furthermore, Ackerman
and Schibrowsky (2007) considered structural bonds such as policies that force
students to finish their years at the colleges and placement services for alumni as
important in students’ retention.
Another important element of marketing orientation is process. According to Ivy
(2008), the procedures at HLIs such as registration for the correct courses, correct
412 F. Muya and H. Tundui

calculation and entering of students’ marks or grades against their names, awarding
of correct qualification, finance system, accommodation, timetabling, library and
graduation are issues that must be in place throughout students’ life at the HLIs. All
these are important in order to ensure the highest level of student satisfaction. On the
effects of intelligence generation on reputation, Shapiro (1988) underscored the
importance of cross-functional information dissemination (ID) and inter-functional
coordination. The findings are in line with those of Zhang et al. (2017) who revealed
that information dissemination (ID) was a very important driver of marketing
orientation of Chinese firms. In another perspective, Ivy (2001) identified market
analysis as an important factor in strengthening the positions of universities. In the
same vein, Ramaseshan et al. (2002) also found that market information collection
and dissemination are critical to the overall success of new products that have an
implication on competitiveness and hence high reputation. These findings are
supported by Caruana et al. (1998) who considered responsiveness as very critical
in HLIs’ overall performance, just after information collection. Furthermore,
Hamadu et al. (2011) found a significant impact of marketing intelligence dissem-
ination on the performance of a business.
Using the components stated above, Muya and Tundui (2020) studied the rela-
tionship between marketing orientation and students’ retention in nine Tanzania
higher learning institutions. Marketing orientation was measured by 7 elements
(programme, people, place, promotion, price, process, and physical evidences/facil-
ities) with a total of 26 indicators measured at a 5-point Likert scale, retention
measured using 8 indicators. Data were collected from 373 students and analysed
using partial least squares structural equation modelling (PLS-SEM) technique with
the assistance of SmartPLS Version 3.0. The study established the ability of mar-
keting orientation to predict retention. A positive and significance relationship
between process, programme, price, promotion and place, and students’ retention
was established while insignificant relationship was found between people and
physical evidence against retention. Reputation was confirmed as an important
mediator in the research model. The study, though involved only students, shed
important light on the applicability of marketing approach in an education
environment.

2.2 Marketing Orientation and Competitiveness

Regarding the effects of marketing orientation on competitiveness, Hemsley-Brown

and Oplatka (2006) view marketing as essential for having a competitive edge and
gaining a larger share of the international and domestic markets. Their views are
supported by Maringe (2005) who agreed that higher education (HE) is experiencing
tougher competition in attracting students from rivals; thus, their managers and
academics need to consider marketing as a viable philosophy and strategy for
developing a HE sector that meets the needs of home-based and international
customers. Dwyer and Kim (2003) identiﬁed a number of marketing-oriented
The Effects of Marketing Orientation on the Performance of Higher. . . 413

components such as place (accessibility of destination), administration (skills of the

employees) and intelligence generation (information and intelligence system) as
essential for a firm’s competitiveness.
Furthermore, on the place element of marketing orientation, Aydın (2013) con-
cluded that the location of a HLI does influence students’ choice of a university.
Location was therefore a source of sustainable competitive advantage for the studied
universities. This finding is consistent with the finding in a study by Beneke and
Human (2010) who identified geographical location as among the most important
factors in a scholar’s decision of a place of further study. Huang (2012) found that
the right location is essential for attracting more students and revenues contributing
to HLI competitiveness. The findings are in line with the findings reported by Sezgin
and Binath (2011) who found that location, the proximity of the campus to the city
centre and exchange programmes are important factors in students’ decision of a
college of study. In views of Pringle and Huisman (2011), an institution located
along well-established public transit routes has a competitive over those located in
poor transit links. In another perspective relating to the programme element of
marketing orientation, Polat and Donmez (2010) found differentiation of products/
services from the products/services offered by competitors as important competi-
tiveness driver. The findings corroborate to those of Dwyer and Mellor (1992) who
found the importance of products developed types in firms’ competitiveness. In the
same vein, Sezgin and Binath (2011) found that curricula with novelty and flexibility
as well as the language of instruction are among the significant factors in a firm’s
competitiveness. Maringe (2005) considered all elements of the marketing mix as
important, suggesting that universities should try to blend all these elements into
their marketing strategies in order to remain competitive in the marketplace.
On the effects of marketing orientation element of intelligence generation on
competitiveness, Ross et al. (2013) asserted that free flow of information between
stakeholder groups is important so that everyone has knowledge that is consistent
and accurate across departmental/institutional boundaries, hence improving their
competitive advantage. Zebal and Goodwin (2012) also found that, for the private
university sector, gathering information about the current university environment on
issues such as changing regulations and an analysis of wider competitor pressures
that could be emerging is more important than the analysis of the strengths and
weaknesses of defined competitors. Dwyer and Kim (2003) found intelligence
system as an important aspect of a firm’s competitiveness.

2.3 Marketing Orientation and Reputation

Being marketing-oriented is essential for ﬁrms’ reputation through different chan-

nels. In relation to the promotion element of marketing orientation, Walsh and
Beatty (2007) established that persuasive communications can enhance a corporate
reputation to the extent of customers engaging in more positive word of mouth. This
is supported by Almquist (2007) who found that word of mouth is an important tool
414 F. Muya and H. Tundui

in customers and prospects perceptions of company’s values. In addition, Lloyd-

Smith and An (2019) found that advertising own firm’s corporate social responsi-
bility (CSR) activities and industry-level CSR spillover positively contribute to the
firm’s reputation. Han et al. (2019) found that social network services (SNS)’ sales
promotion results in a significant increase in loyalty. Place is another important
element of marketing orientation and can potentially contribute to institutions’
reputation. According to Bennett and Gabriel (2001), clear distribution policies
and channel cooperation can lead to corporate reputation for convenient availability
and positive channel. On the effects of pricing on institutions’ reputation, Campbell
(1999) views objective pricing policies as a means of building corporate reputation
for price fairness and value authentic.
On the effects of intelligence generation on reputation, Ivy (2001) asserted that it
is important for universities to conduct a market analysis to establish their market
position and to present an institutional image effectively. In the same vein, Nguyen
and LeBlanc (2001), based on theories developed by economists, claimed that a
consensus is that reputation is the result of past actions of an organization. On the
effects of administration and physical facilities as elements of marketing orientation,
Nguyen and LeBlanc (2001) argued that faculty members and facilities are critical
factors that helped to determine students’ perceptions of the image or reputation of a
higher education institution. These findings are supported by Rajh and Došen (2009)
who found a positive and strong effect of company employees on service brand
image. Castro et al. (2005) stressed the importance of employee in the market
orientation and performance of organizations. Elsewhere, Polat and Donmez
(2010) argued that effective marketing promises several benefits including enhanced
company image, improved customer loyalty and improved reputation.

2.4 Marketing Orientation and Quality Teaching

Marketing orientation is an important source of quality through a number of routes.

Polat and Donmez (2010) argued that effective marketing is essential for improved
total quality. Castro et al. (2005) emphasized the importance of coordinating the
company’s internal activities with the aim of continuously creating value and quality
for the customer. Ivy (2008) views aspects such as the design of the degree, which is
properly developed and adapted to meet students’ needs, curriculums and
programme duration as important programme elements in the HLI context. In
another study, Luk (1997) found a positive relationship between marketing culture
and service quality (SQ), a similar stance reached by Jaensson and Uiso (2015).
Crosby (1996) views human resource such as tutors in HLI as the prime focus in
service quality management and performance in service delivery. In not a quite
different perspective, Voon (2008) established that service-driven market orientation
(SERVMO) has a signiﬁcantly strong and positive relationship with service quality,
with employee orientation being the most dominant factor. The role of employees is
also recognized by Ye and Liang (2010) and Chigozirim and Mazdarani (2008). A
The Effects of Marketing Orientation on the Performance of Higher. . . 415

study by Matear et al. (2002) found that market orientation contributes to perfor-
mance through a dual mechanism in that it contributes both directly and with
innovation mediating the contribution. SeyedJavadin et al. (2012), while investigat-
ing the effects of internal marketing orientation (IMO) on service quality (SQ) and
the mediating role of organizational citizenship behaviour (OCB), found that internal
marketing activities are positively influencing OCB; OCB is having a meaningful
relationship with SQ and OCBs and, as being influential, playing a crucial mediation
role in the success of SQ and external marketing. The findings are supported by Edo
et al. (2015) who verified the positive influence of IMO on employees’ attitudes (job
satisfaction and trust) and, through them, on performance (service quality perceived
by customers that affect their level of satisfaction). From the foregoing discussion, it
can be hypothesized as follows:
H1: There is a positive relationship between mentoring (MENT) and students’
retention (RET).
H2: There is a positive relationship between physical facilities (PHYIC) and stu-
dents’ retention (RET).
H3: There is a positive relationship between price (PRIC) and students’ retention
(RET).
H4: There is a positive relationship between programme (PROG) and students’
retention (RET).
H5: There is a positive relationship between process (PROC) and students’ reten-
tion (RET).
H6: There is a positive relationship between place (PLAC) and competitiveness
(COMP).
H7: There is a positive relationship between programmes (PROG) and competitive-
ness (COMP).
H8: There is a positive relationship between intelligence generation (INTEL) and
competitiveness (COMP).
H9: There is a positive relationship between administration (ADMIN) and reputa-
tion (REP).
H10: There is a positive relationship between intelligence generation (INTEL) and
reputation (REP).
H11: There is a positive relationship between promotion (PROM) and reputation
(REP).
H12: There is a positive relationship between administration (ADMIN) and quality
teaching (TEAQ).
H13: There is a positive relationship between process (PROC) and quality teaching
(TEAQ).
H14: There is a positive relationship between reputation (REP) and competitiveness
(COMP).
H15: There is a positive relationship between quality teaching (TEAQ) and com-
petitiveness (COMP).
H16: There is a positive relationship between retention (RET) and competitiveness
(COMP).
416 F. Muya and H. Tundui

3 Methodology

3.1 Procedure and Participants

The chapter adopted a quantitative cross-sectional survey whereby data were col-
lected from nine HLIs in Tanzania mainland from July 2019 to January 2020. The
sample size for the survey was 800 respondents who were randomly and conve-
niently selected. However, the responses were obtained from 739 respondents,
which is equal to 92% responses rate. The chapter data were collected from staff
(49.5%) and students (50.5%) unlike previous studies that used only a single
group of respondents (Caruana et al. 1998—head of schools only; Rajh and Došen
2009—students only; Zhang et al. 2017—managers only), hence minimizing the
danger of biased results. Regarding the gender of the respondents, the majority
(53%) were males and the remaining respondents (46%) were females. Regarding
their age, the majority (58.1%) were between 20 and 31 years, 19.8% were between
31 and 40 years, 11.9% were between 41 and 50 years, and 5.4% were over 60 years
of age. Regarding their level of education, the majority (49.5%) had a bachelor’s
degree, followed by 20.2% who had a PhD, 20.2% had a master’s degree, 6% had a
postgraduate diploma, and ﬁnally a few (1.5%) had a diploma.

3.2 Variables and Measurements

Many previous scholars used the phrase market orientation, but their discussion
focus was entirely on the implementation of marketing principles. Such scholars
used either MKTOR (Narver and Slater 1990) or MARKOR (Kohli et al. 1993;
Kohli and Jaworski 1990) scale to measure market orientation. This chapter adopted
the phrase marketing orientation as used by McNeal and Lamb (1980), Gummesson
(1990), Kurtinaitienė (2005), Šályová et al. (2015), Sharp (1991) and Domański
(2014) to refer to the implementation of marketing principles. Marketing orientation
is measured by using nine exogenous constructs: six constructs (programme
[PROG], place [PLAC], promotion [PROM], price [PRIC], process [PROC] and
physical facilities [PHYIC]) are from the Education Marketing Mix (EMM) (Ivy
2008) and three constructs (administration [ADMIN], mentoring [MENT] and
intelligence generation [INTEL]) are from University MARKOR (UM) (Niculescu
et al. 2013) measured at a 5-point Likert scale.
The combination of the Education Marketing Mix and University MARKOR was
important to offset weaknesses available in some measure of marketing orientation.
According to Andreasen and Kotler (2014), using all elements of the “marketing
mix,” and not just communication is one of the important characteristics in an
organization that has fully adopted marketing orientation. The marketing mix frame-
work, however, suffers from the weakness of having too much internal orientation.
This weakness is against the proposition provided by Wymer et al. (2006) that an
The Effects of Marketing Orientation on the Performance of Higher. . . 417

organization that has a marketing orientation is able to focus on its various activities
and external communications to project a consistent image of itself and inﬂuence the
way the external world perceives it. The components of the University MARKOR
helped to compensate for the weaknesses on the marketing mix framework. It has to
be noted that a combination of Education Marketing Mix (EMM) and University
MARKOR (UM) gave rise to marketing orientation as the predictor construct.
Performance as an outcome variable was measured using four endogenous con-
structs of retention (RET) (Lichiello 2014; Niculescu et al. 2013), reputation (REP)
(Gajić 2012; Chun 2005; Conard and Conard 2000), competitiveness (COMP)
(Štimac and Šimić 2012) and quality teaching (TEAQ) (Zebal and Goodwin
2012); all are measured at 5-point Likert scale.

3.3 Data Analysis and Results

The partial least squares structural equation modelling (PLS-SEM) using SmartPLS
Version 3.0 was used to test the direct and indirect effects between marketing
orientation and performance through two major stages of assessment of the mea-
surement model followed by the assessment of the structural model. PLS-SEM was
used in this chapter because the main intention was to assess the prediction power of
the exogenous variables on endogenous variables and conﬁrmation of marketing
orientation theories (Hair et al. 2017). It was used also because of the complexity of
the structural model that has many constructs and many indicators. PLS-SEM is also
relevant for determining the direct and indirect effects of predictor variables on the
outcome variable(s) (Hair et al. 2016). Hair et al. (2017) indicated that PLS-SEM is
among the leading data analysis tools in marketing studies and leading journals. The
existence of user-friendly software like SmartPLS Version 3.0 also supported the
uses of PLS-SEM (Hair et al. 2011).

3.4 Assessment of Measurement Model

The assessment of the measurement model included examining the indicators

loadings, assessing internal consistency reliability, assessing the convergent validity
of each construct measure and ﬁnally assessing the discriminant validity (Hair et al.
2019). These results are presented in Fig. 1 and Table 1.
Table 1 shows internal consistency and convergent validity results as among the
key activities in the assessment of the measurement model. Due to its strength over
Cronbach’s alpha, composite reliability (CR) was used to measure the internal
consistency of items where all constructs of the research model were above 0.7
and less than 0.95 as recommended by Hair et al. (2019). These results, therefore,
revealed the presence of internal consistency of the scales and therefore higher levels
of reliability. The same results were also reported by Khuwaja (2016) who found a
418 F. Muya and H. Tundui

Fig. 1 Indicator’s loadings and coefﬁcient of determination

Table 1 Constructs’ reliability and validity

Constructs Composite reliability (CR) Average variance extracted (AVE)
ADMINI 0.861 0.556
COMP 0.858 0.548
INTEL 0.833 0.504
MENT 0.842 0.517
PHYIC 0.927 0.647
PLAC 0.783 0.555
PRIC 0.805 0.581
PROC 0.791 0.560
PROG 0.866 0.522
PROM 0.817 0.529
REP 0.873 0.579
RET 0.767 0.523
TEAQ 0.844 0.579
The Effects of Marketing Orientation on the Performance of Higher. . . 419

Table 2 Discriminant validity results

1 2 3 4 5 6 7 8 9 10
MENT 0.622
PHYIC 0.456 0.487
PLAC 0.293 0.453 0.355
PRIC 0.356 0.498 0.305 0.428
PROC 0.496 0.874 0.717 0.577 0.566
PROG 0.393 0.495 0.292 0.406 0.389 0.481
PROM 0.430 0.580 0.448 0.511 0.501 0.598 0.333
REP 0.307 0.504 0.365 0.385 0.240 0.523 0.275 0.355
RET 0.397 0.590 0.383 0.404 0.355 0.621 0.325 0.403 0.617
TEAQ 0.445 0.535 0.404 0.256 0.299 0.501 0.456 0.375 0.670 0.490

high level of internal consistency reliability for University MARKOR in the higher
education context. Using average variance extracted (AVE), it was found that all
constructs had a value of 0.50 and above which indicated that on average each
construct in our model explains more than half of the variance of its indicators
revealing the presence of convergent validity. Similar results are reported by
Khuwaja (2016) who found a high level of convergent validity for University
MARKOR.
Figure 1 shows that indicator’s loadings for the majority of indicators were above
0.70 as recommended, which implies that each construct explains more than 50% of
the indicator’s variance, thus providing acceptable item reliability (Hair et al. 2019).
Discriminant validity that is the extent to which a construct is truly distinct from
other constructs by empirical standards (Hair et al. 2019) was measured using the
heterotrait–monotrait ratio (HTMT). The results are presented in Table 2.
The results in Table 2 revealed the presence of discriminant validity as all values
were less than 0.9 as recommended. The presence of discriminant validity was also
observed in a study by Khuwaja (2016). Taking into account that the measurement
model was satisfactory, the next step was the assessment of the structural model
(Hair et al. 2019).

3.5 Assessment of Structural Model

Before assessing the structural relationships, collinearity must be examined to make

sure that it is not biased on the regression results (Hair et al. 2019). When collinearity
is not a problem, the assessment of the structural model follows, which include
considering the coefficient of determination (R2), f 2 effect size, the blindfolding-
based cross-validated redundancy measure (predictive relevance as measured by
Stone–Geisser’s Q2) as well as the statistical significance and relevance of the path
coefficients (Hair et al. 2019). Collinearity using variance inflation factor (VIF) was
examined after running the PLS algorithm, and the results are presented in Table 3.
420 F. Muya and H. Tundui

Table 3 Collinearity Constructs COMP REP RET TEAQ

assessment
ADMINI 1.383 1.313
COMP
INTEL 1.267 1.383
MENT 1.722
PHYIC 1.486
PLAC 1.149
PRIC 1.202
PROC 2.021 1.313
PROG 1.244 1.234
PROM 1.198
REP 1.602
RET 1.318
TEAQ 1.613

Table 4 Coefﬁcient of Endogenous constructs R2 R2 adjusted

determination
COMP 0.320 0.314
REP 0.229 0.226
RET 0.196 0.191
TEAQ 0.255 0.253

The VIF values revealed the presence of few indicators with collinearity and no
constructs suffered from collinearity problem as their variance inflated factors (VIF)
were lower than 3 as can be seen in Table 3. The other important activity in the
assessment of the structural model was assessing the coefficient of determination
(R2), for each endogenous variable as presented in Table 4.
Table 4 shows the coefficient of determination (R2), for each endogenous variable
implying the explanatory power of the exogenous variables on the endogenous
variables. Regarding f 2 effect size, the results revealed that the majority of f 2 values
(except REP on COMP = 0.181) were below 0.02, which implied small effects. On
predictive relevance, as measured by Stone–Geisser’s Q2 that postulates that the
model must be able to adequately predict each endogenous latent construct’s indi-
cator (Hair et al. 2011), the values of Q2 for each model’s construct and indicators
were above zero, thus providing support for the predictive relevance of marketing
orientation model for the four endogenous constructs. Furthermore, statistical sig-
nificance and relevance/size of the path coefficients were examined to assess the
extent to which the data reflect the hypothesized relationships, and the results are
presented in Table 5.
Table 5 shows that 11 relationships ( p < 0.05) were significant, implying that the
respective hypotheses are supported, and five relationships ( p > 0.05) were insig-
nificant, resulting in the rejection of those hypotheses. Table 5 also shows the
relative importance of the exogenous driver constructs in predicting the dependent
constructs. In predicting the dependent construct COMP, place element of the
marketing mix (PLAC = 0.115) is the most important among all constructs. In
The Effects of Marketing Orientation on the Performance of Higher. . . 421

Table 5 Size of path coefﬁ- Paths Path coefﬁcients p-values

cients and statistical
REP → COMP 0.444 0.000
signiﬁcance
ADMINI → REP 0.386 0.000
ADMINI → TEAQ 0.364 0.000
MENT → RET 0.231 0.000
PROC → TEAQ 0.214 0.000
PROC → RET 0.161 0.003
PROM → REP 0.144 0.001
PLAC → COMP 0.115 0.001
TEAQ → COMP 0.090 0.043
RET → COMP 0.087 0.018
PHYIC → RET 0.080 0.084
PRIC → RET 0.054 0.189
PROG → RET 0.040 0.320
INTEL → REP 0.038 0.352
INTEL → COMP 0.007 0.853
PROG → COMP -0.098 0.012

contrast, intelligence generation (INTEL = 0.007) has very little inﬂuence on

COMP, and the programme (PROG = -0.098) has negative effects on COMP.
On the other hand, mentoring (MENT = 0.231) is the most important among all
constructs in predicting the endogenous construct RET, followed by process (PROC
= 0.161). However, physical evidence (PHYIC = 0.080), price (PRIC = 0.054) and
programme (PROG = 0.040) had very little influence on RET. The relative impor-
tance of the exogenous driver constructs in predicting the endogenous constructs
reputation (REP) indicates that administration (ADMINI = 0.386) is the most
important followed by promotion (PROM = 0.144). However, intelligence genera-
tion (INTEL = 0.038) had very little influence on reputation. The results indicate
that administration (ADMINI = 0.364) is the most important factor in predicting the
endogenous construct quality teaching, followed by process (PROC = 0.214).
Finally, the results show that endogenous construct reputation (REP) is the primary
driver of HLI competitiveness (REP = 0.444) and quality teaching (TEAQ = 0.090),
but retention (RET = 0.087) had very little influence on COMP.

3.6 Mediation Effects

The relationship between marketing orientation and performance is sometimes

channelled via a number of variables (Maydeu-Olivares and Lado 2003; Kamboj
and Rahman 2017). This type of relationship was also assessed in this chapter
especially on the roles of reputation (Arikan et al. 2016; Sridhar and Mehta 2018)
and teaching quality (Osarenkhoe et al. 2017) as mediators. The results are presented
in Table 6.
422 F. Muya and H. Tundui

Table 6 Speciﬁc indirect effects

Relationships Coefﬁcients p-values
R1 ADMINI → REP → COMP 0.171 0.000
R2 INTEL → REP → COMP 0.017 0.351
R3 PROM → REP → COMP 0.064 0.001
R4 MENT → RET → COMP 0.020 0.031
R5 PHYIC → RET → COMP 0.007 0.223
R6 PRIC → RET → COMP 0.005 0.287
R7 PROC → RET → COMP 0.014 0.066
R8 PROG → RET → COMP 0.003 0.385
R9 ADMINI → TEAQ → COMP 0.033 0.058
R10 PROC → TEAQ → COMP 0.019 0.060

The results of mediation effects in Table 6 revealed the presence of mediation

between the constructs (R1, R3 and R4) and the absence of mediation for a number
of relationships (R2, R5, R6, R7, R8, R9 and R10). Among the major ﬁndings in this
analysis is that teaching quality does not mediate the relationship between many
constructs.

3.7 Comparison of Staff and Students’ Opinions

As the chapter findings come from both staff and students, a comparison was done
through multigroup analysis (MGA) to test whether the differences between
group-specific path coefficients are statistically significant. The PLS-MGA findings
(Column 2) revealed a significant difference between staff and students on the
relationship between (1) intelligence generation and reputation—INTEG → REP
(2) and between physical facilities and retention—PHYIC → RET. Similarly, the
Welch-Satterthwaite test findings (column 3) show significant differences between
staff and students on the relationship between (1) intelligence generation and
reputation—INTEG → REP and on the relationship between (2) physical facilities
and retention—PHYIC → RET, implying the absence of significant differences
between the groups. All these findings are presented in Table 7.
The general picture that can be deduced based on Table 7 is that there is absence
of significant differences between staff and students on the relationship between
marketing orientation and performance.

4 Discussion

Among the hypotheses in this chapter was the presence of a positive relationship
between mentoring (MENT) and students’ retention (RET). This hypothesis was
supported (β = 0.231, p < 0.05), revealing that mentoring is signiﬁcantly and
The Effects of Marketing Orientation on the Performance of Higher. . . 423

Table 7 PLS-MGA between staff and students

p-values p-values
Paths (1) (students vs. staff) (2) (students vs. staff) (3) Remarks (4)
ADMINI → COMP 1.000 0.996 Insignificant
ADMINI → REP 0.870 0.866 Insignificant
ADMINI → TEAQ 0.893 0.895 Insignificant
INTEL → COMP 0.273 0.278 Insignificant
INTEL → REP 0.002 0.002 Significant
MENT → COMP 0.376 0.384 Insignificant
MENT → RET 0.565 0.565 Insignificant
PHYIC → COMP 0.086 0.115 Insignificant
PHYIC → RET 0.044 0.037 Significant
PLAC → COMP 0.563 0.584 Insignificant
PRIC → COMP 0.937 0.935 Insignificant
PRIC → RET 0.947 0.978 Insignificant
PROC → COMP 0.378 0.396 Insignificant
PROC → RET 0.060 0.053 Insignificant
PROC → TEAQ 0.711 0.702 Insignificant
PROG → COMP 0.223 0.223 Insignificant
PROG → RET 0.171 0.182 Insignificant
PROM → COMP 0.316 0.323 Insignificant
PROM → REP 0.270 0.275 Insignificant
REP → COMP 0.908 0.909 Insignificant
RET → COMP 0.459 0.461 Insignificant
TEAQ → COMP 0.956 0.957 Insignificant

positively related to retention at a signiﬁcance level of 5%. These results portray a

very true picture as the presence of mentoring-related activities within a higher
learning institution may lead to higher education students’ academic performance
and proper decision-making, issues that are essential for their survival. These
ﬁndings are supported by Rowley (2003) who advocated mentoring-related activities
such as the use of personal tutor systems and encouraging electronic interaction
between staff and students. The ﬁndings were also supported by Ackerman and
Schibrowsky (2007) who emphasized that students’ retention is possible through
social bonding like all interpersonal interactions that exist such as student-advisor,
student-instructors and accessible faculty and support staff, all of which are
mentoring-related aspects. The importance of mentoring in students’ retention is
also underscored by the Hanover Report (2014) that advocated the presence of
academic advising and support (as the top factor), social connectedness, approach-
ability of faculty and staff, availability of mentoring and coaching programmes and
the presence of well-trained employees to provide accurate information and friendly
responses to enquiries. Fetsco et al. (2016) also found a positive contribution of
mentoring and advising on students’ retention. Hameed and Amjad (2011) found a
positive relationship between faculty, advisory staff and the classes with the
424 F. Muya and H. Tundui

student’s college experience. These results support the Tinto Model (Tinto 1993)
observations that the more integrated a student is into the social systems of the
institution, the more likely a student will be committed to the institution and to the
goal of graduation and, as a result, the more likely the student will persist to
graduation at that institution.
Another hypothesis in this chapter was the presence of a positive relationship
between physical facilities (PHYIC) and students’ retention (RET). This hypothesis
was rejected (β = 0.080, p > 0.05), which revealed that physical facilities are
positively but insignificantly related to retention. These findings are true due to the
fact that the presence of enough physical facilities may result in a high level of costs
that may discourage students from staying in an education institution due to failure to
bear the associated costs. Similar findings are also reported in a study by Anis et al.
(2018), Carter and Yeo (2016) and Muya and Tundui (2020). This finding is in
contrast to the findings from the previous studies by Mokaya (2013) and Lau (2003)
who found a positive role of physical facilities on students’ retention. On whether
there is a positive relationship between price and students’ retention, this hypothesis
was rejected (β = 0.054, p > 0.05), which implies that price is positively but
insignificantly related to students’ retention. These findings call for rethinking on
the price charged by HLIs whereby higher rates may jeopardize students’ continuity
with their studies. This shows the importance of sponsorship opportunities beyond
fees among students for enhancing their access to education and comfortable stay in
a HLI. The results on this hypothesis are close to those by Muya and Tundui (2020)
but far below those by Ackerman and Schibrowsky (2007), Hanover Report (2014),
Hasan et al. (2009), Carabajal (2012) and Polat and Donmez (2010), who found
price-related activities such as redesigning of scholarship, bursary, tuition fees
amount and flexible tuition approaches as essential for students’ retention.
Another hypothesis was whether there is a positive relationship between
programme and students’ retention. This hypothesis was rejected (β = 0.040, p >
0.05), which implies that programme is positively but insignificantly related to
students’ retention. The findings imply that students enrol at an institution because
there is a specific programme to fulfil their specific purposes. In course of proceeding
with studies, students may attain low performance if they joined wrong programme,
something which may evict the students from the HLI. In addition, if there is a
programme at an upper level such as a master’s degree, this may force the student to
stay in a particular HLI for continuing with studies. The findings in this hypothesis
do not support the previous research by Ivy (2008), Mokaya (2013), Muya and
Tundui (2020) and Rowley (2003) who revealed that programme-related activities
such as adequate participation in cocurricular activities, design of the degree,
properly developed and adapted curriculum to meet students’ needs, curricula and
programme duration and curriculum that is more vocationally relevant, as important
factors in students’ retention. In this chapter, it was also hypothesized that there is a
positive relationship between process and students’ retention. This hypothesis was
supported (β = 0.161, p < 0.05), which implies that the process is positively related
to students’ retention. These findings are in line with the findings reported in a study
by Ivy (2008) and Muya and Tundui (2020) who appreciated the importance of
The Effects of Marketing Orientation on the Performance of Higher. . . 425

processes at HLI such as registration for the correct courses, correct calculation and
entering of students marks or grades against their names, awarding of correct
qualification, finance system, accommodation, timetabling, library and graduation
as important throughout students’ life at the HLI.
In addition to the above hypotheses, in this chapter, it was explored whether there
is a positive relationship between place and competitiveness. This hypothesis was
supported (β = 0.115, p < 0.05), which implies that the place is positively related to
competitiveness. The results are supported by other studies such as Aydın (2013),
Beneke and Human (2010), Dwyer and Kim (2003), Hemsley-Brown and Oplatka
(2006), Maringe (2005), Sezgin and Binath (2011) and Pringle and Huisman (2011),
who found the importance of place-related aspects on firms’ competitiveness. On
whether there is a positive relationship between programmes and competitiveness,
although the path coefficient was statistically significant, this hypothesis was
rejected (β = -0.098, p < 0.05) as the sign is reversed thereby failing to support
the hypothesis that the programme is positively related to competitiveness. These
results are in contrast with the findings from other studies (i.e. Dwyer and Mellor
1992; Polat and Donmez 2010; Sezgin and Binath 2011) that found the importance
of programme-related aspects on firms’ competitiveness.
Another hypothesis in this chapter was the presence of a positive relationship
between intelligence generation and competitiveness. This hypothesis was rejected
(β = 0.007, p > 0.05), which implies that intelligence generation has an insignificant
relationship with competitiveness. These results differ from the findings from other
studies (i.e. Dwyer and Kim 2003; Ross et al. 2013; Zebal and Goodwin 2012) that
revealed that intelligence generation–related aspects such as free flow of information
between stakeholders, gathering information on issues like changing regulations and
an analysis of wider competitor pressure as important on the firms’ competitiveness.
The contradiction can be due to the nature of high learning institutions whose
vulnerability due to limited information is low in comparison with other for-profit
organizations. Furthermore, in this chapter, a positive relationship was hypothesized
between administration and HLI reputation. This hypothesis was supported (β =
0.386, p < 0.05), which implies that HLI administration is positively related to
reputation. The results are in line with the findings from other studies (Castro et al.
2005; Polat and Donmez 2010; Rajh and Došen 2009), which support the role of
administration-related aspects such as the best employee on the firms’ reputation.
In another attempt, it was hypothesized that there is a positive relationship
between intelligence generation and HLI reputation. This hypothesis was rejected
(β = 0.038, p > 0.05), which implies that intelligence generation is insignificantly
related to reputation. These results were surprising as many other studies
(i.e. Caruana et al. 1998; Castro et al. 2005; Hamadu et al. 2011; Shapiro 1988;
Zhang et al. 2017) acknowledged the impact of intelligence generation–related
aspects on the firms’ performance. The findings imply that even if higher learning
institutions are better on intelligence generation, other efforts are essential to
enhance their reputation. On whether there is a positive relationship between pro-
motion and HLI reputation, this hypothesis was supported (β = 0.144, p < 0.05)
which implies that HLI promotion is positively related to HLI reputation. The
426 F. Muya and H. Tundui

findings are in line with those from other studies (Almquist 2007; Lloyd-Smith and
An 2019; Han et al. 2019; Walsh and Beatty 2007), which reported the importance of
promotion-related aspects such as persuasive communication, word of mouth,
advertising and sales promotion on reputation.
Another hypothesis was on the presence of a positive relationship between
administration and quality teaching. This hypothesis was supported (β = 0.364,
p < 0.05), revealing that university administration positively and significantly
influences quality teaching. The findings corroborate with those by other researchers
(i.e. Crosby 1996; Edo et al. 2015; Jaensson and Uiso 2015; Polat and Donmez
2010; Voon 2008) who revealed the effects of administration aspects on quality
teaching. Another hypothesis was on the presence of a positive relationship between
process and quality teaching. This hypothesis was supported (β = 0.214, p < 0.05)
implying that university processes positively influence quality teaching. The find-
ings are supported by findings in a study by Yorke (1999), who established that
improving quality teaching requires a number of processes such as quality assurance
mechanisms both internally and externally and attaining certification such as
ISO 9000.
Among the hypotheses tested included whether reputation has a positive rela-
tionship with competitiveness (β = 0.444, p < 0.05). This hypothesis was supported
as the p values were less than the significance level of 0.05. The analysis revealed
further that reputation mediates the relationship between administration and com-
petitiveness (R1) (β = 0.171, p < 0.05) and between promotion and competitiveness
(R3) (β = 0.064, p < 0.05). As both indirect effects and direct effects are significant,
the presence of partial mediation (Hair et al. 2017) can be deduced. The effects of
reputation were also reported by Nguyen and LeBlanc (2001) and Hussin et al.
(2000) who revealed that institutional reputation contributed to improved students’
loyalty. Nuraryo et al. (2018) and Munisamy et al. (2014) found reputation as having
positive impacts on students’ retention. The findings on positive and significant
effects of reputation on satisfaction and retention are in line with the findings from a
study by Keh and Xie (2009) who found a positive relationship between reputation
and customer’s belief-consistent feelings of commitment. Similar findings are also
reported by Arikan et al. (2016) who revealed a positive and significant direct effect
of reputation on customer satisfaction and on customer loyalty. Furthermore, the
analysis revealed lack of mediation by reputation on the relationship between
intelligence generation and competitiveness (R2) (β = 0.017, p > 0.05).
Other hypothesis in this chapter was that quality teaching has a positive relation-
ship with competitiveness (β = 0.090, p < 0.05). The hypothesis was supported due
to the presence of a significant relationship between quality teaching and competi-
tiveness. Furthermore, the analysis revealed that there is no mediation by quality
teaching on the relationship between administration and competitiveness (R9) (β =
0.033, p > 0.05) and on the relationship between process and competitiveness (R10)
(β = 0.019, p > 0.05). The results were contrary to the findings of Sun and Pang
(2017) and Whang (2017) who found quality as essential in shaping firm’s compet-
itiveness both nationally and globally but also contrary to Hussin et al. (2000) who
found teaching quality as a very important variable in the higher education context.
The Effects of Marketing Orientation on the Performance of Higher. . . 427

The findings are however supported by Sari et al. (2018), who found an insignificant
relationship between service quality and students’ retention that matters for HLI
competitiveness. Finally, it was hypothesized that retention has a positive relation-
ship with competitiveness (β =0.090, p < 0.05). The hypothesis was supported due
to the presence of a significant relationship between retention and competitiveness.
Furthermore, the analysis revealed the presence of mediation by retention on the
relationship between mentoring and competitiveness (R4) (β = 0.020, p < 0.05).
These results corroborate with those by Gengeswari and Padmashantini (2013) and
Voss and Voss (2008). However, the results revealed lack of mediation by retention
on the relationship between physical facilities and competitiveness (R5) (β = 0.007,
p > 0.05), price and competitiveness (R6) (β = 0.005, p > 0.05), process and
competitiveness (R7) (β = 0.014, p > 0.05) and programme and competitiveness
(R8) (β = 0.003, p > 0.05).

5 Conclusion

This chapter has successfully assessed and presented the effects of marketing
orientation (MO) on performance in a higher education context using PLS-SEM;
the data were analysed using SmartPLS Version 3.0. The consideration of the
coefficient of determination (R2) of the endogenous variables revealed predictive
accuracy of MO on HLI performance. Such results refute the results of previous
studies that considered marketing orientation as not a suitable strategy for institu-
tions such as HLI. The path coefficient revealed the relative importance of exoge-
nous variables on the endogenous construct. The relationships between the
exogenous variables and endogenous variables were significant in the most hypoth-
esized relationships, which also implies the influence that marketing orientation has
on the performance of HLI. The relationship between MO and performance is
mediated by reputation and retention, aspects that strengthened the relationship
between the two. The results in this chapter lead to the conclusion that marketing
orientation (MO) is a relevant strategy for improving HLI performance through
various channels. This chapter recommends for the adoption of marketing orienta-
tion among HLIs as a key initiative for attaining optimal and sustainable
performance.
This chapter has some limitations worth noting. The chapter adopted a cross-
sectional time horizon, which limited the understanding of the effects of marketing
orientation on performance for a long span of years. Given enough time, longitudinal
studies can be undertaken to assess marketing orientation–performance relationship
for a long span of years. The chapter was written based on the data collected from the
education industry; future studies can be undertaken in other industries as similar
knowledge on other industries is limited in Tanzania. Due to time and resource
concerns, only some HLIs were involved in this chapter. For better generalizability
of results, future studies can expand the number of HLIs or involve all of them. As
the chapter used data obtained from a single country (intra-countries), for better
428 F. Muya and H. Tundui

generalization of results, future studies can be of inter-countries to establish market-

ing orientation–performance relationship across the countries.

Acknowledgments I express my heartfelt appreciation to the management of National Institute of

Transport (NIT) for ﬁnancing my PhD Studies, which resulted in coming up with this chapter.

References

Ackerman R, Schibrowsky J (2007) A business marketing strategy applied to student retention: A

higher education initiative. J Coll Student Retention Res, Theory Pract 9(3):307–336
Akonkwa DBM (2009) Is market orientation a relevant strategy for higher education institutions?
Context analysis and research agenda. Int J Qual Serv Sci 1(3):311–333
Al-Fattal A (2010) Understanding student choice of university and marketing strategies in Syrian
private higher education. Doctoral dissertation, University of Leeds, School of Education, Leeds
Almquist MDE (2007) An approach to mastering the marketing mix. Bus Strat Ser 8(2):122–131
Anabila P, Kastner ANA, Bulley CA, Allan MM (2019) Market orientation: A key to survival and
competitive advantage in Ghana’s private universities. J Mark High Educ 30(1):124–144
Andreasen AR, Kotler P (2014) Strategic marketing for non-proﬁt organizations, 7th edn. Pearson,
Edinburgh Gate
Anis A, Islam R, Abdullah NA (2018) Challenges faced by Malaysian private HLIs in providing
quality education: A thematic analysis. Qual Assur Educ 26(3):349–373
Arikan E, Kantur D, Maden C, Telci EE (2016) Investigating the mediating role of corporate
reputation on the relationship between corporate social responsibility and multiple stakeholder
outcomes. Qual Quant 50:129–149
Aydın OT (2013) Location as a competitive advantage to attract students: An empirical study from a
Turkish Foundation University. Int Rev Manag Mark 3(4):204–211
Baker MJ, Saren M (2010) Marketing theory: A student text, 2nd edn. SAGE, L.A.
Beck CN (2007) A Review of market orientation practices for selected public Carnegie high
research activity institutions. PhD dissertation, Montana State University, Bozeman, Montana
Beneke J, Human G (2010) Student recruitment marketing in South Africa: An exploratory study
into the adoption of a relationship Orientation. Afr J Bus Manag 4(4):435–447
Bennett R, Gabriel H (2001) Reputation, trust and supplier commitment: The case of shipping
company/seaport relations. J Bus Ind Mark 16(6):424–438
Borden NH (1984) The concept of the marketing mix. J Advert Res Classics, II(September):7–12
Burgess SM, Nyajeka P (2007) Market orientation and performance in low-income countries: The
case of Zimbabwean retailers. Adv Int Manag 20:215–257
Campbell MC (1999) Perceptions of price unfairness: Antecedents and consequences. J Mark Res
36(2):187–199
Carabajal K (2012) Noel-Levitz student satisfaction inventory report. Research, Planning and
Information Technology (Spring). https://wwwlassencollegeedu/about/institutional-effective
ness/Documents/Research/2012%20LCC%20SSI%20Springpdf Accessed 4 Mar 2021
Carter S, Yeo AC (2016) Students-as-customers’ satisfaction, predictive retention with marketing
implications: The case of Malaysian higher education business students. Int J Educ Manag
30(5):1–24
Caruana A, Ramaseshan B, Ewing MT (1998) Do universities that are more market oriented
perform better? Int J Public Sect Manag 11(1):55–70
Castro CB, Armario EM, del Río MES (2005) Consequences of market orientation for customers
and employees. Eur J Mark 39(5/6):646–675
The Effects of Marketing Orientation on the Performance of Higher. . . 429

Chandler N, Heidrich B, Szászvári K, Kása R (2021) Reframing market orientation: A comparative

study of the market orientation concept in the subcultures of university employees. Soc Econ
43(3):270–288
Chigozirim PS, Mazdarani E (2008) The impact of employee empowerment on service quality and
customer satisfaction: a case study of Länsförsäkringar AB. Bachelor thesis, Mälardalen
University, Västerås
Chun R (2005) Corporate reputation: Meaning and measurement. Int J Manag Rev 7(2):91–109
Conard MJ, Conard MA (2000) An Analysis of academic reputation as perceived by consumers of
higher education. J Mark High Educ 9(4):69–80
Crosby PB (1996) Quality is still free: Making quality certain in uncertain times. McGraw-Hill,
New York
Dolnicar S, Lazarevski K (2009) Marketing in non-profit organizations: An international perspec-
tive. Int Mark Rev 26(3):275–291
Domański T (2014) Marketing of higher education: Future challenges. Manag Bus Adm Cent Eur
22(4):118–132
Dwyer L, Mellor R (1992) The impact of new product strategies of Australian firms. Prometheus
10(2):285–299
Dwyer L, Kim C (2003) Destination competitiveness: Determinants and indicators. Curr Issue Tour
6(5):369–414
Edo VT, Monzonís JL, Miguel TAM, Sánchez-García JS (2015) The influence of internal market
orientation on external outcomes: The mediating role of employees’ attitudes. J Serv Theory
Pract 25(4):486–523
Eikenberry A, Kluver JD (2004) The marketization of the non-profit sector: Civil society at risk?
Public Adm Rev 64(2):132–140
Felcman M (2012) Reflections on the market-oriented theory in the behaviour of real organization.
Atl Mark J 1(2):49–62
Fetsco TG, Donnelly MAS, Tang WMA (2016) First-year retention programs: recommendations
for effective implementation. Arizona GEAR UP Research in Practice, Series V: Postsecondary
Transition, Paper No. 2. https://in.nau.edu/wp-content/uploads/sites/6/2018/08/Second-RIP-
Paper-on-College-Transitons-final-draft-ek.pdf. Accessed 11 July 2021
Filip A (2012) Marketing theory applicability in higher education. Procedia Soc Behav Sci 46:912–
916
Gajić J (2012) Importance of marketing mix in higher education institutions. Singidunum J 9(1):
29–41
Gengeswari K, Padmashantini P (2013) Impact of customer retention practices on firm performance
in the retailing sector: A comparison between a foreign and home grown retailer in Malaysia. Int
J Manag Inf Technol 6(1):672–688
Goi CL (2009) A review of marketing mix: 4Ps or more? Int J Market Stud 1(1):2–15
Gummesson E (1990) Marketing-orientation revisited: The crucial role of the part-time marketer.
Eur J Mark 25(2):60–75
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: Indeed, a silver bullet. J Mark Theory Pract
19(2):139–151
Hair JF, Hollingsworth CL, Randolph AB, Chong AYL (2016) An updated and expanded assess-
ment of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2017) A Primer on partial least squares structural
equation modeling (PLS-SEM), 2nd edn. SAGE, New Delhi
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hamadu D, Obaji R, Oghojafor B (2011) Intelligence information generation, dissemination, and
responsiveness in the performance of insurance business in Nigeria. J Market Dev Competi-
tiveness 5(7):53–62
Hameed A, Amjad S (2011) Students’ satisfaction in higher learning institutions: A case study of
COMSATS Abbottabad, Pakistan. Iran J Manag Stud 4(1):63–77
430 F. Muya and H. Tundui

Han H, Nguyen HN, Song H-J, Chua B-L, Lee S, Kim W (2019) Role of Social Network Services
(SNS) sales promotions in generating brand loyalty for chain steakhouses. J Qual Assur Hosp
Tour 20(1):617–645
Hanover Research (2014) Strategies for improving student retention—Report. https://www.
hanoverresearch.com/media/Strategies-for-Improving-Student-Retention.pdf. Accessed
18 Apr 2021
Hasan M, Siddiqui BA, Chowdhury SK (2009) Higher education marketing: 8S marketing mix for
private universities in Bangladesh. Bus Rev 1(1):101–113
Hemsley-Brown J, Oplatka I (2006) Universities in a competitive global marketplace: A systematic
review of the literature on higher education marketing. Int J Public Sect Manag 19(4):316–338
Hilman H, Kaliappen N (2014) Market orientation practices and effects on organizational perfor-
mance: Empirical insight from Malaysian hotel industry. SAGE Open 1–8
Huang HI (2012) An empirical analysis of the strategic management of competitive advantage: a
case study of higher technical and vocational education in Taiwan. Doctoral dissertation,
Victoria University
Hussin SR, Soon TH, Sidin S (2000) Marketing analysis of the higher education service sector in
Malaysia: Consumer perspective. Pertanika J Soc Sci Humanit 8(1):1–6
Ivy J (2001) Higher education institution image: a correspondence analysis approach. Int J Educ
Manag 15(6):276–282
Ivy J (2008) A new higher education marketing mix: the 7Ps for MBA marketing. Int J Educ Manag
22(4):288–299
Jaensson J, Uiso H (2015) Assessing the link between market orientation and poverty reduction in
the tourism industry: The case of Tanzania. Int J Bus Tour Appl Sci 3(1):57–64
Kamboj S, Rahman Z (2017) Market orientation, marketing capabilities and sustainable innovation:
The mediating role of sustainable consumption and competitive advantage. Manag Res Rev
40(6):698–724
Keh HT, Xie Y (2009) Corporate reputation and customer behavioral intentions: the roles of trust,
identification and commitment. Ind Mark Manag 38:732–742
Khuwaja FM (2016) The mediating role of innovation in the relationship between market orienta-
tion and university performance in Sind, Pakistan. Thesis, Universiti Utara, Malaysia
Khuwaja FM, Shari H, Abubakar LJ (2015) A conceptual paper on the impact of market orientation
and innovation on university performance of public sector universities of Pakistan, with specific
focus on student satisfaction. Int J Entrepren Small Medium Enterprise 7(4):419–436
Kiessling T, Isaksson L, Yasar B (2015) Market orientation and CSR: Performance implications. J
Bus Ethics 137(2):269–284
Kohli AK, Jaworski BJ (1990) Market orientation: The construct, research propositions, and
managerial implications. J Mark 54(April):1–18
Kohli AK, Jaworski BJ, Kumar A (1993) MARKOR: A measure of market orientation. J Mark Res
30(4):467–477
Kurtinaitienė J (2005) Marketing orientation in the European Union mobile telecommunication
market. Mark Intell Plan 23(1):104–113
Lau LK (2003) Institutional factors affecting students’ retention. Education 124(1):126–136
Lichiello PC (2014) The graduate retention relationship cycle: The role of performance, engage-
ment, satisfaction, and alienation. Theses and Dissertations, Lynchburg College
Lloyd-Smith P, An H (2019) Are corporate social responsibility and advertising complements or
substitutes in producing firm reputation? J Appl Econ 51(21):2275–2288
Luk STK (1997) An examination of the role of marketing culture in service quality. Int J Contemp
Hosp Manag 9(1):13–20
Maringe F (2004) Vice chancellor’s perceptions of university marketing: A view from universities
in a developing country. High Educ Rev 36(2):53–68
Maringe F (2005) Interrogating the crisis in higher education marketing: The CORD model. Int J
Educ Manag 19(7):564–578
The Effects of Marketing Orientation on the Performance of Higher. . . 431

Maringe F (2006) University and course choice: Implications for positioning, recruitment and
marketing. Int J Educ Manag 20(6):466–479
Matear S, Osborne P, Garrett T, Gray BJ (2002) How does market orientation contribute to service
firm performance? An examination of alternative mechanisms. Eur J Mark 36(9/10):1058–1075
Maydeu-Olivares A, Lado N (2003) Market orientation and business economic performance: A
mediated model. Int J Serv Ind Manag 14(3):284–309
McNeal JU, Lamb CW (1980) Marketing orientation in the non-profit sector: The case of hospitals.
J Acad Market Sci 8(1):26–32
Mok KH (2001) Academic capitalisation in the new millennium: the marketisation and
corporatisation of higher education in Hong Kong. Policy Polit 29(3):299–315
Mokaya ZM (2013) Influence of school infrastructure on students’ performance in public secondary
schools in Kajiado County, Kenya. Master thesis, University of Nairobi
Mokoena BA (2018) Exploring the applicability of the MKTOR Scale in predicting university
performance: A focus on universities of technology in South Africa International. J Soc Sci
Humanity Stud 10(1):1–16
Moorman C, Rust RT (1999) The role of marketing. J Mark 63(special issue):187–197
Munisamy S, Jaafar NIM, Nagaraj S (2014) Does reputation matter? Case study of undergraduate
choice at a premier university. Asia Pac Educ Res 23(3):451–462
Muya F, Tundui H (2020) Marketing orientation and students’ retention in Tanzania higher learning
institutions: Mediating role of institutional reputation. Int J Arts Humanit (IJAH) 9(2):214–232
Narver JC, Slater SF (1990) The effect of a market orientation on business profitability. J Mark 54
(October):20–35
Newman CN (2002) The current state of marketing activity among higher education institutions. J
Mark High Educ 12(1):15–29
Nguyen N, LeBlanc G (2001) Image and reputation of higher education institutions in students’
retention decisions. Int J Educ Manag 15(6):303–311
Nicolescu L (2009) Applying marketing to higher education: Scope and limits. Manag Market 4(2):
5–44
Niculescu M, Xu B, Hampton GM, Peterson RT (2013) Market orientation and its measurement in
universities. Adm Issues J Educ Pract Res 3(2)
Nuraryo I, Sumartias S, Umar H, Rahmat A (2018) The influence of corporate (University) identity
on student retention with corporate reputation and student satisfaction as mediating variables:
The Case of a Jakarta Business School. Soc Sci 13(8):1366–1372
Osarenkhoe A, Komunda MB, Byarugab JM (2017) Service quality as a mediator of customer
complaint behaviour and customer loyalty. Int Rev Manag Mark 7(1):197–208
Perreault WD, McCarthy JE (2002) Basic marketing: a global managerial approach, 14th edn.
McGraw-Hill/Irwin, New York
Polat G, Donmez U (2010) Marketing management functions of construction companies: Evidence
from Turkish contractors. J Civ Eng Manag 16(2):267–277
Pringle J, Huisman J (2011) Understanding universities in Ontario, Canada: An industry analysis
using Porter’s five forces framework. Can J High Educ 41(3):36–58
Rajh E, Došen DO (2009) The effects of marketing mix elements on service brand equity. Econ Res
22(4):69–83
Ramaseshan B, Caruana A, Pang LS (2002) The effect of market orientation on new product
performance: a study among Singaporean firms. J Prod Brand Manag 11(6):399–409
Rivera-Camino J, Molero Ayala V (2010) Market orientation at universities. Construct and explor-
atory validation. Innovar 20(36):125–138
Ross M, Grace D, Shao W (2013) Come on higher ed . . . get with the programme! A study of
market orientation in international student recruitment. Educ Rev 65(2):219–240
Rotfeld HJ (2008) The largest segments that should not be served: higher education marketing
serving the growing slacker segment. J Consum Mark 25(6):378–380
Rowley J (2003) Retention: Rhetoric or realistic agendas for the future of higher education. Int J
Educ Manag 17(6):248–253
432 F. Muya and H. Tundui

Šályová SJ, Táborecká-Petrovičová J, Nedelová G, Ďaďo J (2015) Effect of Marketing Orientation

on Business Performance: A Study from Slovak Foodstuff Industry. Proc Econ Finance 34:622–
629
Sari SW, Sunaryo S, Mugiono M (2018) The effect of service quality on customer retention through
commitment and satisfaction as mediation variables in java eating houses. J Appl Manag 16(4):
593–604
Sefnedi S (2017) The Market orientation and performance relationship: The empirical link in
private universities. Jurnal Manajemen Dan Kewirausahaan 19(1):28–37
SeyedJavadin S, Rayej H, Yazdani H, Estiri M, Aghamiri SA (2012) How organizational citizen-
ship behavior mediates between internal marketing and service quality: The case of Iranian GAS
company. Int J Qual Reliab Manag 29(5):512–530
Sezgin A, Binath A (2011) Determinants of university choice in Turkey. Int Higher Educ Congr
Proc 3(12):1651–1657
Shapiro BP (1988) What the hell is ‘market-oriented’? Harvard Business Review, November-
December, 119–125
Sharp B (1991) Marketing orientation: More than just customer focus. Int J Wine Mark 3(1):20–25
Shepherd IDH (2008) The application of marketing thinking to student retention. Planet 19(1):
26–30
Stachowski CA (2011) Educational marketing: A review and implications for supporting practice in
tertiary education. Educ Manag Adm Leadersh 39(2):186–204
Starck K, Zadeh SH (2013) Marketing within higher education institutions: A case study of two
private Thai universities. Master thesis, Mälardalen University
Štimac H, Šimić LM (2012) Competitiveness in higher education: a need for marketing orientation
and service quality. Econ Soc 5(2):23–34
Sridhar M, Mehta A (2018) The moderating and mediating role of corporate reputation in the link
between service innovation and cross-buying intention. Corp Reput Rev 21(2):50–70
Sun W, Pang J (2017) Service quality and global competitiveness: evidence from global service
firm. J Serv Theory Pract 27(6):1058–1080
Svenson G, Wood G (2007) Are university students really customers? When illusion may lead to
delusion for all. Int J Educ Manag 21(1):17–28
Tabaku E, Mersini (Zerellari) M (2013) An overview of marketing means used by non-profit
organizations: a detailed overview of NPOs operating in the District of Elbasan. J Mark
Manag 4(2):78–95
Tajeddini K, Trueman M, Larsen G (2006) Examining the effect of market orientation on innova-
tiveness. J Mark Manag 22:5–6
Tinto V (1993) Leaving college: Rethinking the causes and cures of student attrition, 2nd edn.
University of Chicago Press, Chicago
Tjahjadi B, Soewarno N, Jermias J, Hariyati H, Fairuzi A, Anwar DN (2022) Does engaging in
global market orientation strategy affect HEIs’ performance? The mediating roles of intellectual
capital readiness and open innovation. J Open Innov Technol Market Complex 8(29):1–18
Voon HB (2008) SERVMO: A measure for service-driven market orientation in higher education. J
Mark High Educ 17(2):216–237
Voss GB, Voss ZG (2008) Competitive density and the customer acquisition–retention trade-off. J
Mark 72:1–18
Walsh G, Beatty SE (2007) Customer-based corporate reputation of a service firm: Scale develop-
ment and validation. J Acad Mark Sci 35(1):127–143
Whang U (2017) Comparative advantage, product quality, and the competitiveness of firms. J
Korea Trade 21(3):174–190
Wu AD, Zumbo BD (2008) Understanding and using mediators and moderators. Soc Indicat Res
87:367–392
Wymer W, Knowels P, Gomes R (2006) Non-profit marketing: Marketing management for
charitable and nongovernmental organizations. SAGE, Thousand Oaks
The Effects of Marketing Orientation on the Performance of Higher. . . 433

Ye X, Liang Z (2010) The employees’ roles on service delivery: Case study on Haibin hotel &
Westin hotel in China. Bachelor Thesis, University of Gävle
Yorke M (1999) Assuring quality and standards in globalised higher education. Qual Assur Educ
7(1):14–24
Zebal MA, Goodwin DR (2012) Market orientation and performance in private universities. Mark
Intell Plan 30(3):339–357
Zhang L, Kara A, Spillan JE, Mintu-Wimsatt A (2017) Exploring market orientation among
Chinese small and medium-sized enterprises. Chin Manag Stud 11(4):617–636
Part VII
Health
Participative Leadership Is
the Discriminating Factor for Country’s
Performance During the COVID-19
Pandemic

Stephanie Dygico Gapud and George Faint

1 Introduction

The COVID-19 pandemic has impacted every nation in this globally interconnected
world. It seems that a country achieves superior performance in protecting its
citizens from the pandemic based not simply on the availability and access to
abundant resources (vaccine, health personnel, hospital beds, personal protective
equipment [PPEs], etc.) in a country. For example, the United States should have all
the technology and the knowledge at its disposal to combat the virus and should rank
very highly as one of the successful countries in combatting the pandemic. By the
time of writing this chapter, the United States ranked number 36 in Bloomberg’s
COVID-19 Resilience Ranking. At the same time, Chile and Vietnam (ranked
13 and 18, respectively) outranked the United States, while the top ﬁve went to
South Korea, the United Arab Emirates, Ireland, Norway, and Saudi Arabia (Hong
et al. 2022).

S. D. Gapud (✉)
Division of Business, Spring Hill College, Mobile, AL, USA
e-mail: [email protected]
G. Faint
Sorrell College of Business, Troy University, Dothan, AL, USA
e-mail: [email protected]

1.1 COVID-19 as the Unfolding External Jolt for the Natural

Experiment on Leadership

The COVID-19 pandemic has thrown the world into uncertainty and complexity
(Wernli et al. 2021) where the world’s resilience depends upon the participation of
multiple agents in different sectors of the society. We will use the global pandemic
caused by COVID-19 as the natural experiment (Dunning 2012; Sieweke and
Santoni 2020) to analyze the impact of leadership style, particularly shared or
participative leadership on the variation of COVID-19 outcome in different coun-
tries. The leadership style during times of crisis is the intangible property of the
country that is largely responsible for the COVID-19 outcome. How the leader
influences the behavior of the citizens was communicated in the country head’s
personal attitude, speeches, and policy during the pandemic (Usman et al. 2021).
Some denied the existence and potency of the virus, while others took it seriously
(Berdy 2020; Frieden 2021; Vinopal 2021). Women head of states like Prime
Minister Jacinda Ardern of New Zealand, Chancellor Angela Merkel of Germany,
Tsai Ing-wen of Taiwan, and Danish Prime Minister Mette Frederiksen made
headlines in popular media as well as became case studies for academic articles
for the effectiveness of their response (Mayer and May 2021; Olagnier and
Mogensen 2020; Tsai 2020; Vinopal 2021). Leading during a time of uncertainty
requires mindfulness of the complexity of the situation (Langer 2010; Weick and
Sutcliffe 2011).
Previous literature suggests that many factors influence societal compliance with
rules and guidelines implemented to combat the pandemic. Leadership, national
culture (Barney and Wright 1998; Gokmen et al. 2021; Huynh 2020; Lagman et al.
2021; Wang 2021), transparency, and trustworthiness of information are often cited
as factors that caused infodemic during the pandemic (Luengo and García-Marín
2020; Berdy 2020). Additionally, the likelihood that all COVID-19 funds and other
resources allocated actually go to the appropriate organizations and programs may
be measured by the Corruption Perception Index (CPI) for the country to estimate the
contribution to the strategy that the country takes on (Barney 1986; Corruption Cuts
2020; Cruz-Ozorio 2020; Rubio 2021; Transparency International 2021). Corruption
is defined as the abuse of entrusted power for private gain (Pozsgai-Alvarez 2020).
Some countries consolidated power using the COVID-19 pandemic as an excuse, yet
according to Transparency International (2021), their research shows that the same
government left citizens without emergency aid.

1.2 Proposed Theories to Explain Causality Observed

in the Leadership Natural Experiment

We followed the theorizing of Lord and Maher (1992) which inﬂuenced House et al.
(2004), when they embarked on the GLOBE project, and identiﬁed factors that
Participative Leadership Is the Discriminating Factor for Country’s. . . 439

explain how various societies behaved and responded to their leaders as inﬂuenced
by their cultural values. We further argue that the corruption perception partly
explains the ability of the leader to inﬂuence the community mobility during the
pandemic. Furthermore, we propose that the number of individuals infected and the
number of deaths at the time when there was still no vaccine available in the world
are largely impacted by the quarantine behavior in the community, which is
impacted by how the community perceive and process the information or messaging
by the leader.

1.3 Leadership: An Intangible Resource of an Organization

Borrowing from the resource-based view theory, we posit that leadership is a

valuable, invisible, and intangible resource tied to the firm, which is the differenti-
ating factor (Barney 1991; Barney and Arikan 2008; Peteraf 1993; Wernerfelt 1984)
that impacts the level of control an organization has during times of uncertainty.
Leadership is defined as the ability of an individual to influence, motivate, and
enable others to contribute toward the effectiveness and success of the organizations
of which they are members (House et al. 2004). When a leader enables others to
contribute toward the fulfillment of the organizational goal, the leadership style
should be participative or what they call shared leadership.
The ability of the leader to encourage experts to participate in the decision-
making is crucial to the effectiveness of the organization. When experts are allowed
to contribute and influence the strategy formulation of an organization, be it a small
team, a firm, or a country makes the organization adaptive even during times of
uncertainty (Weick and Sutcliffe 2011). Moreover, beyond command and control
(Pearce and Manz 2005), when leadership is shared by allowing experts to partici-
pate in the dynamic influencing processes during times of global crisis (Avolio and
Gardner 2005; Bligh et al. 2006; Pearce and Wassenaar 2014), innovative solutions
are created because talents are maximized, corruption is minimized, and the greater
good is served (Manz and Pearce 2017; Pearce et al. 2008). The participative global
leadership dimension score of several countries obtained from the GLOBE project
was used to determine how the impact of sharing the leadership in a country
influences the COVID-19 situation.
In their 2020 annual report on corruption perceptions around the world, Trans-
parency International noted that “Corruption and emergencies feed off each other,
creating a vicious cycle of mismanagement and deeper crisis. The large sums of
money required to deal with emergencies and the need for urgency in disbursing
these funds form a perfect storm for corruption” (Transparency International 2020).
To cope with the urgent need to provide relief during the pandemic, numerous
countries relaxed their public procurement regulations, which were in place “to
limit the discretion of public officials and promote competition among sellers in
order to avoid potential collusion between bureaucrats and contractors” (Gallego
et al. 2020, p. 1).
440 S. D. Gapud and G. Faint

Fig. 1 The proposed model. Notes: Hypothesized relationships of the constructs are illustrated. H1:
Corruption – !Participative leadership, H2: Culture + ! Participative leadership, H3: Participative
leadership + ! Community mobility, H4: Community mobility – ! New cases, H5: Community
mobility – !Total cases, H6: Community mobility – !Total deaths

Therefore, it is posited that the perception of corruption in a country erodes social

contract, and it impacts the leader’s strategy and therefore the ability to govern
during the pandemic. Moreover, it is argued that the ability of a leader to formulate a
successful strategy during a pandemic provides the resource heterogeneity that
determines the success of the nation’s strategy to curtail the impact of the
COVID-19. Furthermore, national culture and practices affect what leaders do
(House et al. 2002; Huynh 2020). We also propose that community mobility during
the initial part of the pandemic illustrates followership as impacted by the leadership
in a country is significantly related to the country’s pandemic situation. We used the
Google’s Community Mobility Report to determine the effectiveness of the influ-
ence of the participative leadership in a country as influenced by the national culture
and by the Corruption Perception Index.
In this chapter, a structural equation model was conceptualized (Fig. 1) based on
the findings of the GLOBE research that “the attributes and practices that distinguish
cultures from each other, as well as strategic organizational contingencies, are
predictive of the leader’s attributes and behaviors, and organizational practices,
that are most frequently enacted and are most effective” (House et al. 2002, p. 9).
The Corruption Perception Index was added as the other factor influencing the
leadership in a country aside from the dimensions of the national culture. The deaths
due to COVID-19 and current cases as the dependent variable are also explored.

2 Methodology

The model was evaluated using secondary data. The results of the GLOBE project
(https://globeproject.com/) were used for the participative leadership. The GLOBE
project identiﬁed six global culturally endorsed leadership dimensions (charismatic/
value-based, team-oriented, participative, humane-oriented, autonomous, and self-
Participative Leadership Is the Discriminating Factor for Country’s. . . 441

protective). In this chapter, the country data on participative global leadership

dimensions (GLDs) are used to test whether allowing experts in the decision-making
during the pandemic has an impact on how the pandemic situation varies in each
country. Participative leadership is defined as “the degree to which managers involve
others in making and implementing decisions” (House et al. 2004, p. 675). Leader-
ship is the ability of an individual to influence, motivate, and enable others to
contribute toward the effectiveness and success of the organizations of which they
are members (House et al. 2004). “Participative Leadership includes two primary
leadership dimensions labeled (a) non participative and (b) autocratic” (Dorfman
et al. 2012, p. 506).
Culture is defined as the collective programming of the mind that distinguishes
the members of one group or category of people from others (Hofstede 1984). There
are two sources of national cultural index, the GLOBE project and Hofstede’s
country insight. To avoid common method bias, the data for culture were taken
from the Hofstede’s country insight (2020). Hofstede and his coauthors’ continued
research on culture started in the 1980s and to date has covered 76 countries. All the
six Hofstede’s dimensions of culture (uncertainty avoidance, masculinity, power
distance, long-term orientation, collectivism, and indulgence) were incorporated in
the analysis. Hofstede (2011, p. 8) defined the cultural dimensions as follows:
1. Power distance, related to the different solutions to the basic problem of human
inequality
2. Uncertainty avoidance, related to the level of stress in a society in the face of an
unknown future
3. Individualism versus collectivism, related to the integration of individuals into
primary groups
4. Masculinity versus femininity, related to the division of emotional roles between
women and men
5. Long-term versus short-term orientation, related to the choice of focus for
people’s efforts: the future or the present and the past
6. Indulgence versus restraint, related to the gratification versus control of basic
human desires related to enjoying life
Data for the corruption perception in the model were taken from the Corruption
Perception Index (2020) published by the research arm of the Transparency Inter-
national. Transparency International (2020) is a global movement whose mission is
to stop corruption and promote transparency, accountability, and integrity at all
levels and across all sectors of society.
The data for the community movement statistics were taken from Google’s
COVID-19 Community Mobility Report (2020). Using similar technology and
process to develop their maps, Google aggregated and anonymized data to track
the movement of individuals during the pandemic. Google has publicly available
global data on movement trends during the pandemic over time by geography (i.e.,
cities and country). In this chapter, the reported changes in community movement
such as retail and recreation, groceries and pharmacies, parks, transit stations, and
workplaces were used in the analysis except for the residential. The data for the
442 S. D. Gapud and G. Faint

changes in movement from people’s home (residential) were not used as they seem
to sum up all the other movements. We can use either the residential data or the five
other categories to illustrate and differentiate the COVID-19 situations in various
countries on October 15, 2020. We chose to include more variables because doing so
will allow us to observe intricate widespread multivariate in the structural equation
model. The earliest data date February 15, 2020; however, Google stopped tracking
changes in community movement in October 15, 2022. The historical data can still
be downloaded as a csv file from the website.
The COVID-19 statistics (Worldometers 2020) were obtained on October
15, 2020. It is published by Dadax, a small media company that generates revenue
from advertising. During the pandemic, it reported real-time statistics on new cases,
total deaths, and total cases. Worldometers recently changed its name to
Worldometer, which tracks real-time statistics on various topics such as energy,
food, society and media, and world population. Worldometer is listed by the
American Library Association as an outstanding free reference website (RUSA
2011).
It is important to note that the COVID-19 and community mobility data were
taken when there was still no approved vaccine available on the market. Therefore,
our model is illustrating a paradigm of a snapshot in time of the relationships of the
variables without the effect of any vaccine in the pandemic situation in the world.
Furthermore, the country’s COVID-19 situation is largely impacted by the dyadic
relationship of the leader–follower dynamics as influenced by culture and perception
of corruption in a country.
Not all the countries with COVID-19 and CPI in the world are included in the
analysis. We are limited by the availability of the national culture and the leadership
data. Some countries have not yet been surveyed by Hofstede or by the GLOBE
project. We only included 59 countries with complete data from secondary sources.
SmartPLS Version 4.0.8.3 (Ringle et al. 2022) was the software used in the partial
least squares structural equation modeling (PLS-SEM). SmartPLS was selected
because it allowed easy creation, analysis, and validation of models (Sander and
Teh 2014).

3 Results

3.1 Measurement Model Conﬁrmation

We followed the series of steps in conﬁrmatory composite analysis (CCA) to

conﬁrm our partial least squares structural equation model. The CCA offers several
advantages over other approaches for conﬁrming measurement models consisting of
linear composites (Hair et al. 2020).
Participative Leadership Is the Discriminating Factor for Country’s. . . 443

3.1.1 Estimate of Loadings and Signiﬁcance

The outer model was assessed by examining the relationship between the constructs
and their indicators. The results show that the two reflectively measured constructs
(culture and community mobility)’s measures are reliable and valid (Table 1 and
Fig. 2). Three community mobility variables out of five were retained in the final
model. The community mobility construct in the model is measured by the move-
ment to the workplace (Mob_Work), travel to purchase and go to recreation places
(Mob_Ret_Rec), and the number of people used the available mass transportation,
that is, trains, airplanes, and bus (Mob_Travel). Moreover, three of the six
Hofstede’s cultural dimensions were retained in the model: masculinity (H_Masc),
uncertainty avoidance (H_UA), and power distance (H_PD).
The original sample (O) shows the original estimate of the outer weight, while the
sample mean (M) is the bootstrap-calculated value (Hair et al. 2017). The standard-
ized loadings of the variables should have a value of at least 0.708 and an associated
t-statistics above 1.96 to be significant for a two-tailed test at the 5% level (Hair et al.

Table 1 Outer loadings and signiﬁcance of the variables of latent constructs

Original Sample Standard t-statistics p-
sample (O) mean (M) deviation (SD) (|O/SD|) values
H_Masc ← National 0.70 0.53 0.35 2.02 0.04
culture
H_PD ← National culture 0.96 0.95 0.03 28.53 <0.01
H_UA ← National culture 0.77 0.65 0.24 3.16 <0.01
Mob_Retail_Rec ← 0.96 0.97 0.01 73.68 <0.01
Community mobility
Mob_Travel ← Commu- 0.95 0.96 0.03 36.56 <0.01
nity mobility
Mob_Work ← Commu- 0.84 0.89 0.06 14.82 <0.01
nity mobility
Total deaths ← Total 1.00 1.00 0.00
deaths

Fig. 2 The structural equation model

444 S. D. Gapud and G. Faint

2011). Other than the H_Masc (0.70) variable of the national culture construct, all
other loadings are above the threshold value of 0.708. Although H_Masc in the
model has slightly lower loading than the threshold, it was retained for construct
validity (Hair et al. 2017). The values of t-statistics in PLS-SEM are obtained by
executing a bootstrapping procedure (Hair et al. 2012). Squaring the individual
indicator loadings provides a measure of the amount of variance shared between
the individual indicator variable and its associated construct. This is referred to as
indicator reliability (Hair et al. 2019).

3.1.2 Reliability and Validity Measures

The reliability of the construct was measured in two ways—Cronbach’s alpha (α)
and composite reliability (CR). The rule of thumb for both reliability criteria is that
they should be above 0.70. The results of our analysis show that all of the reﬂectively
measured constructs’ measures are reliable and valid. Table 2 shows that all the
constructs in our model pass the reliability test.

3.1.3 Convergent Validity

Average variance extracted (AVE) was used to evaluate the convergent validity. The
AVE is obtained by averaging the indicator reliabilities of a construct. This metric
measures the average variance shared between the construct and the individual
indicators. The criterion for AVE is that its value should be 0.5 (50%) or higher.
All constructs (community mobility and national culture) on the model have con-
vergent validity (Table 2).

3.1.4 Discriminant Validity

Table 3 shows that the two latent constructs in the model (national culture and
community mobility) are distinct from each other. Moreover, the heterotrait–
monotrait (HTMT) criterion analysis indicated a lack of discriminant validity
between total deaths and total cases (Table 3). This very high (0.928) HTMT ratio
indicates the true correlation between total cases and total deaths (Hair et al. 2017;
Henseler et al. 2015). However, since both the variables are exact measures and not

Table 2 Reliability and convergent validity measures

Cronbach’s Composite Average variance extracted
alpha rho_A reliability (AVE)
Community 0.91 0.91 0.94 0.85
mobility
National culture 0.79 1.43 0.85 0.67
Participative Leadership Is the Discriminating Factor for Country’s. . . 445

Table 3 Discriminant validity assessment using the heterotrait–monotrait (HTMT) criterion

Community National New Total Total
mobility Corruption Leadership culture cases cases deaths
Community
mobility
Corruption 0.10
Leadership 0.34 0.45
National 0.34 0.26 0.33
culture
New cases 0.69 0.01 0 0.09
Total cases 0.86 0.07 0.15 0.08 0.81
Total deaths 0.83 0.07 0.18 0.07 0.64 0.93

constructs that need to be validated, this means that indeed the total number of deaths
is highly correlated with the total number of cases in a country. As it was expected,
the results are considered acceptable for this study. It is interesting to note that
community mobility is highly correlated with the country’s COVID-19 scenario
(new cases, 0.69; total cases, 0.86; total deaths, 0.83).
The results of the SEM analysis (Table 4) clearly illustrate that leadership
mediates the outcome of a country’s pandemic situation.

3.2 Structural Model Assessment

The final structural equation model in Fig. 2 shows that the dimensions of Hofstede’s
culture that are loaded significantly in the culture latent construct are masculinity
(H_Masc), power distance (H_PD), and uncertainty avoidance (H_UA). This means
that the differences in how each country scores on these three cultural dimensions
covary significantly with the various other constructs and indicators in the model.
Therefore, the dimensions of culture that sees inequality in the distribution of power
affect the society (power distance); the level of stress in a society in the face of
unknown future (uncertainty avoidance) and the division of emotional roles between
women and men have impact on the COVID situation in a country as explained by
the way the community behaves toward their perception of the way they are being
led (Hofstede 2011).

3.3 Predictive Validity of the Model: Within-Sample

Prediction Results

The coefﬁcient of determination (R-squared or R2), a measure of in-sample predic-

tion of all endogenous constructs (Shmueli et al. 2019), explains the percentage of
446 S. D. Gapud and G. Faint

Table 4 Path coefﬁcients and results of hypotheses

Original Sample Standard t- p-values
sample mean deviation statistics and
Hypothesized relationship (O) (M) (SD) (|O/SD|) significance
H1 Corruption perception –0.37 –0.29 0.17 2.17 0.03**
(–) ! Participative
leadership
H2 Culture (+) 0.28 0.36 0.12 2.29 0.02**
!Participative leadership
H3 Participative leadership 0.32 0.32 0.12 2.65 <0.01***
(+) ! Community
mobility
H4 Community mobility –0.65 –0.64 0.16 4.00 <0.01***
(–) ! New cases
H5 Community mobility –0.82 –0.78 0.14 5.68 <0.01***
(–) ! Total cases
H6 Community mobility -0.80 -0.76 0.14 5.71 <0.01***
(–) ! Total deaths
Notes: Critical t-values for a two-tailed test are 1.65 (significance level = 10%*), 1.96 (significance
level = 5%**), 2.58 (significance level = 1%***). The one-tailed significance of a directional
hypothesis at the level of 0.05 is 0.98

variance in the model and was used to assess the model’s in-sample fit (Table 5). R-
squared is the squared correlation of the actual and predicted values and is a measure
of in-sample predictive power because it includes all the data for model estimation to
judge the predictive power (Rigdon 2012; Sarstedt et al. 2014). In our model,
leadership is being impacted by culture and corruption at about 25% based on the
adjusted R2. However, the impact of all the independent variables (community
mobility, leadership, corruption, and national culture) in the model on the dependent
variables (i.e., total deaths, total cases, and new cases) is large (Cohen 1992). We
found that the variables relevant to country’s COVID-19 situation (i.e., new cases,
total cases, and total deaths) have high R2 values way above the large (0.26)
predictive accuracy threshold based on the rule of thumb established for the field
of psychology (Cohen 1992). However, it is important to note that the R2 results
explain the high predictive ability of the model only for the sample of data (59 coun-
tries) used in calculating the results. Furthermore, the high predictive property of the
model should not be inferred from the population (all of the countries in the world)
(Rigdon 2012; Sarstedt et al. 2014; Shmueli et al. 2019). All we can infer at this point
is that community mobility predicted 9% by the constructs preceding it in the model.
Leadership is 25% predicted. Of the COVID-19 metrics, total cases have the highest
coefficient of determination (R2 = 0.67). This indicates that the total cases are
predicted 67% by the model.
The effect size ( f2) is the second measure of predictive validity. It is an estimate
of the predictive ability of each independent construct in the model. The effect size is
also considered an in-sample predictive metric (Hair et al. 2020). As seen in Table 6,
Participative Leadership Is the Discriminating Factor for Country’s. . . 447

Table 5 R-squared values of endogenous latent constructs in the path model

R2 Psychology rule of Marketing R2 rule of
2
R adjusted thumb thumb
Leadership 0.27 0.25 Medium Weak
Community 0.11 0.09 Small Weak
mobility
New cases 0.43 0.42 Large Moderate
Total cases 0.68 0.67 Large Moderate
Total deaths 0.63 0.63 Large Moderate
Notes: Cohen (1992) classiﬁed the strength of the predictive accuracy (R2 values) as either small
(0.02), medium (0.13), or large (0.26). Marketing issues, are evaluated differently with R2 values
reported as substantial (0.75), moderate (0.50), or weak (0.25) (Sarstedt and Mooi 2014, p. 211)

Table 6 Within-sample ﬁt effect size: f 2 values of constructs

Community National New Total Total
mobility Corruption Leadership culture cases cases deaths
Community 0.75 2.08 1.73
mobility
Corruption 0.17
Leadership 0.12
National 0.10
culture
New cases
Total cases
Total deaths
Notes: The rule of thumb for assessing f 2 values follows Cohen (1992)—small (0.02), medium
(0.15), and large (0.35)

the within-sample ﬁt effect size of community mobility on new cases, total cases,
and total deaths is way above the large threshold. The corruption effect of leadership
is above the medium threshold level (0.17). The effect of leadership on community
mobility is below medium but above the small threshold (0.12). A similar level is
seen in the effect of national culture (0.10) to leadership.

3.4 Out-of-Sample Prediction Results: PLSpredict

In this section, we will determine whether the model held beyond the data used in the
analysis of the model. During times of uncertainty other than the COVID-19
pandemic, will the people in the community listen to the leader who makes decision
based on the advice of the experts? Moreover, will the people’s behavior be always
impacted by the Corruption Perception Index and the national culture particularly
masculinity, power distance, and uncertainty avoidance in the time of great
448 S. D. Gapud and G. Faint

Table 7 Comparison of the PLS-SEM with the LM outcomes of community mobility manifest
variables (MVs)
PLS-SEM LM
Manifest variables Q2predict RMSE MAE RMSE MAE
Mob_Retail_Rec 0.03 1,338,685.00 807,998.60 1,347,038.00 832,597.34
Mob_Travel 0.06 1,026,115.00 661,230.80 1,042,541.00 678,233.58
Mob_Work 0.04 1,856,358.00 939,073.10 1,895,001.00 988,259.43

Fig. 3 Comparison of the prediction errors generated in PLS and LM analysis

uncertainty (i.e., a nuclear disaster, hurricane mandatory evacuation, earthquake, or

wildfire) other than the COVID-19 pandemic? We used the PLSpredict to determine
the predictive ability of the model outside of the data-specific uncertainty situation
(COVID-19). PLSpredict is a recently developed out-of-sample prediction assess-
ment tools for composite-based PLS-SEM (Hair 2020).
To assess the results of a specific PLS path model, its predictive performance can
be compared against two naïve benchmarks (Hair 2020; Ringle et al. 2022; Shmueli
et al. 2019). The first benchmark is that if the Q2 value of community mobility is
positive (0.05), the prediction error of the PLS-SEM results is smaller than the
prediction error of simply using the mean values. In that case, the PLS-SEM models
offer a better predictive performance. The second benchmark is that the linear model
(LM) offers prediction errors and summary statistics that ignore the specified PLS
path model. Instead, the LM approach regresses all exogenous indicator variables on
each endogenous indicator variable to generate predictions. Thereby, a comparison
with the PLS-SEM results offers information whether using a theoretically
established path model improves (or at least does not worsen) the predictive perfor-
mance of the available indicator data. Table 7 shows the comparison of the
PLS-SEM with the LM outcomes. The PLS-SEM results have a lower prediction
error (e.g., in terms of root mean square error [RMSE] or mean absolute error
[MAE]) than the LM. Moreover, Fig. 3 shows that using the PLS-SEM model
generates lower prediction error than simply using the liner regression model.
We used big data to reveal which antecedent’s (culture or corruption) impact has
statistical significance on the country’s COVID-19 situation. We found that all the
paths in our model emanating from corruption have p-values less than 0.05.
Participative Leadership Is the Discriminating Factor for Country’s. . . 449

Moreover, the evaluation of within-sample and out-of-sample prediction metrics

shows that the model is predictive.

3.5 Importance–Performance Map Analysis (IPMA)

Modeling reality requires us to understand the intricacies of relationships between

the relevant indices and constructs and then interpret reality to reveal the truth.
Moreover, we analyzed in context how each variable impacts the performance of the
constructs in the model beyond p = 0.05 (Wasserstein et al. 2019). In this post hoc
analysis, we connect what we theoretically know and deep dive into the data and ﬁnd
out how each measured variable impacts the performance of the dependent variables
in the model. We determined the effect and importance of each preceding variable on
the performance of each dependent variable in the model. To do that, we conducted
another analysis available in the SmartPLS program called the importance–perfor-
mance map analysis.
Standard PLS-SEM analyses provide information on the relative importance of
constructs in explaining other constructs in the structural model. Information on the
importance of constructs is relevant for drawing conclusions. The importance–
performance map analysis (IPMA) extends the results of PLS-SEM by also taking
the performance of each construct into account. In the computation of IPMA, we
chose to evaluate the importance–performance map that includes all constructs in the
PLS path model that are indirect and direct predecessor constructs of the selected
target construct in the PLS path model. In Fig. 4 and Table 8, we identiﬁed that only
the Corruption Perception Index (CPI) has a positive effect on shaping the perfor-
mance of the total deaths as a variable. The goal of this analysis is to identify the
variable that has high importance but low performance. In this case, we see that CPI
has very high importance but a relatively low performance. In ceteris paribus

Fig. 4 Importance–performance map (total deaths, indicators, and standardized effects)

450 S. D. Gapud and G. Faint

Table 8 Indicators’ total effects and performance on total deaths

Indicators Performance on total deaths Total effects on total deaths
CPI_2020 52.45 0.09
H_Masc 76.20 -0.01
H_PD 78.70 -0.05
H_UA 81.71 -0.02
Mob_Retail_Rec 89.72 -0.30
Mob_Travel 87.80 -0.28
Mob_Work 93.78 -0.29
PartGLD_RCoded 47.48 -0.26

Fig. 5 Importance–performance map (community mobility, indicators, and standardized effects)

situation, a one-unit increase in the performance of CPI increases the performance of

total deaths. This means that an increase in corruption will result in an increase in
deaths.
To determine which variable impacts the performance of the community mobil-
ity, we again analyzed the model using IPMA; only this time the target is community
mobility. It is fascinating to see that leadership has a 32% total effect on the
performance of the variable community mobility. All the culture dimensions have
small effects, but they are high in terms of performance (Fig. 5 and Table 9).
Corruption has a negative effect (-0.118) and scored 50.447 on performance.
The results show that an increase in the participative leadership (PartGLD) will
have a corresponding positive change in the community mobility variable, while CPI
will have a negative impact on community mobility.
Corruption will have a negative effect on the performance of leadership (Fig. 6).
Power distance according to Hofstede (2011) is related to the different solutions to
the basic problem of human inequality as perceived by the citizens. H_PD has the
Participative Leadership Is the Discriminating Factor for Country’s. . . 451

Table 9 Indicators’ total effects and performances on community mobility

Indicators Total effect on community mobility Performance on community mobility
CPI_2020 -0.12 52.45
H_Masc 0.02 76.20
H_PD 0.07 78.70
H_UA 0.02 81.71
PartGLD_RCoded 0.32 47.48

Fig. 6 Importance–performance map (leadership, indicators, and standardized effects)

Table 10 Indicators’ total effects and performance on leadership

Indicator total effects on leadership MV performances
CPI_2020 -0.37 52.45
H_Masc 0.06 76.20
H_PD 0.20 78.70
H_UA 0.07 81.71

highest effect (20%), and any unit change in power distance will improve the
performance of the leadership manifest variable by 78.70% (Table 10). When
power is shared, there will be a corresponding decrease in the power concentration
on only one person. The leader will be perceived as empowering those below his
ranks when she/he allows experts to participate in leading during times of
uncertainty.
452 S. D. Gapud and G. Faint

4 Discussion

The data used to analyze the model were taken at the height of pandemic when there
was still no vaccine available. Therefore, “how individuals respond to advice on how
best to prevent transmission will be as important as government actions, if not more
important” (Anderson et al. 2020, p. 934). Moreover, the public’s perception of the
leaders’ communication strategies of how best to avoid infection was vital not only
in terms of health and life statistics but also in preventing or minimizing the
economic downturn. This, taking COVID-19 pandemic as the natural experiment
to test leadership theories, revealed that the presence of an autocratic and nonparti-
cipative global leadership dimension combined with high levels of corruption is the
perfect storm that potentially can cause damage of pandemic proportions when the
country is dealing with uncertainty (Burkle 2020). Moreover, the International
Monetary Fund explained that corruption indeed is the other pandemic that the
world is dealing with as revealed by the COVID-19 pandemic situation (Gaspar
et al. 2020).
In this chapter, we refer to participative leadership as a type of shared leadership.
Pearce et al. (2008) theorized that sharing the lead will deter corruption. Crevani
et al. (2007, p. 24) suggested that, in shared leadership, “employees should be
recognized as responsible and accountable co-leaders rather than as untrustworthy
subordinates, they should be entrusted to make decisions not only on operative
matters but also on governance matters.” Leadership style (participative/shared)
might be the variable that separates successful navigation during the pandemic.
When the leadership is participative or shared, the authority and decision-making
are allowed to migrate to the frontlines; therefore, there will be an inversion of tasks
(Crevani et al. 2007). The needs of the situation will be communicated up to the
chain of command. Therefore, tasks are delegated upward in the organizational
hierarchy. The resources will be allocated to where they will truly be needed. The
policies and strategies that will be formulated will match the changing needs of the
uncertain situation. The hierarchy will function to serve the needs of the common
good, and those who have authority will be seen as legitimately serving the purpose
of their ofﬁce. This type of shared leadership is observed when the manager is
mindful and practices mindful leadership (Eastburn 2018; Gapud 2019). Moreover,
shared leadership is prescribed to be the cure of the leadership disease—that of
concentrating power and authority on one person (Manz and Pearce 2018). As a
member of society, an individual leader should embrace the doctrine of social
responsibility and extend it to political mechanisms in combating common invisible
enemies, striving to achieve collective goals without resorting to deception or fraud.
This implies that individuals must conform and contribute to a broader social
purpose, as Friedman (2007) advocates.
During times of great uncertainty, leaders do matter (Azam et al. 2020; Jones and
Olken 2005; Sashkin and Sashkin 2003; Tomkins 2020). The executive has a
professional and moral responsibility to obtain information and make decisions to
prevent deaths in time of a disaster (Bishop 1991). Participative, a type of shared
Participative Leadership Is the Discriminating Factor for Country’s. . . 453

leadership, is key to responsible leadership (Pearce et al. 2014)—which is actualized

when country heads take on the “stakeholder approach to strategy” (Freeman 2016;
Freeman and McVea 2005) as leaders in the highest ofﬁce of the country engage in
an honest and open dialog to authentically care (Liedtka 2006, 2008) for its
constituents.

5 Conclusion

PLS-SEM is a very powerful tool indeed in revealing the causality in the secondary
data that are already available to describe the reality in the natural environment
where a phenomenon occurs. Organizing mindfully to be able to flexibly adapt and
achieve positive outcomes in a rapidly changing environment requires sharing the
lead to allow the participation of various experts to cater to the many battle fronts
presented by the pandemic. Therefore, participative leadership, when practiced by
the person occupying the highest office of command in the country, provides the
unique organizational capability that renders control over the country’s handling of
COVID-19. Moreover, the organizational outcome is largely influenced by differing
expectations of the society from their leaders, influenced by their cultural values, and
the perceived trustworthiness (or corruption) of the initiatives enacted to combat the
jolt that has disturbed the organizational environment.

6 Theoretical and Managerial Implications

Data that are publicly available are information describing reality. Structural equa-
tion modeling can help turn these data into knowledge if we combine the numbers
with the appropriate theories. “If a model is consistent with reality, then the data
should be consistent with the model. But if the data is consistent with the model, this
does not imply that the model corresponds to reality” (Bollen 1989). Following
Bollen and Pearl’s (2013) suggestion, other researchers should be able to suggest
alternative SEM models for the same set of data. Future studies should investigate
the possible impact of other global leadership dimensions before and after the global
introduction of vaccine. For example, future research can determine the impact of the
team orientation and the humane orientation on GLDs.
Results presented in this chapter determine the relationship between Hofstede’s
dimension of culture and the GLOBE project’s global leadership dimensions
(GLDs).
Back when there was no vaccine available, country leadership is an intangible
resource that mediates the organizational outcome, as evidenced by the spread of the
disease due to the movement of the community. Moreover, in this chapter, the
cultural dimensions that are loaded in the ﬁnal PLS-SEM algorithm are uncertainty
avoidance, masculinity, and power distance. It is very likely that changing the
454 S. D. Gapud and G. Faint

leadership dimension in the model might also result in a different set of cultural
dimensions that will be retained in the culture latent variable.
The Corruption Perception Index in the country has a great impact on national
outcomes. At the same time, leadership style (participative or shared) positively
impacts organizational performance, especially during times of uncertainty.
Allowing experts (those who have knowledge and experience relevant to the situa-
tion and organizational goals) to participate in strategy formulation provides orga-
nizational adaptability and curbs corruption.

References

Anderson RM, Heesterbeek H, Klinkenberg D, Hollingsworth TD (2020) How will country-based

mitigation measures inﬂuence the course of the COVID-19 epidemic? Lancet 395(10228):
931–934
Avolio BJ, Gardner WL (2005) Authentic leadership development: Getting to the root of positive
forms of leadership. Leadersh Q 16(3):315–338
Azam A, Kurmanaev A, Politi D, Londono E (2020) Inequality, poor leadership and weak health
systems bring disaster. The New York Times, June 24
Barney JB (1986) Organizational culture: can it be a source of sustained competitive advantage?
Acad Manag Rev 11(3):656–665
Barney JB (1991) Firm resources and sustained competitive advantage. J Manag 17(1):99–120
Barney JB, Arikan AM (2008) The resource-based view. In: Hitt AM, Freeman RE, Harrison JS
(eds) The Blackwell handbook of strategic management. Blackwell, Malden, MA, pp 123–182
Barney JB, Wright PM (1998) On becoming a strategic partner: The role of human resources in
gaining competitive advantage. Hum Resour Manag 37(1):31–46
Berdy M (2020) How Russia’s coronavirus crisis got so bad, 24 May 2020. https://www.politico.eu/
article/how-russias-coronavirus-crisis-got-so-bad-vladimirputin-covid19/. Accessed
14 Sept 2020
Bishop J (1991) The moral responsibility of corporate executives for disasters. J Bus Ethics 10(5):
377–383
Bligh MC, Pearce CL, Kohles JC (2006) The importance of self- and shared leadership in team
based knowledge work: A meso-level model of leadership dynamics. J Manag Psychol 21(4):
296–318
Bollen KA (1989) Structural equations with latent variables, vol 210. Wiley, New York
Bollen KA, Pearl J (2013) Eight myths about causality and structural equation models. In: Morgan
SL (ed) Handbook of causal analysis for social research. Springer, Dordrecht, pp 301–328
Burkle FM (2020) Declining public health protections within autocratic regimes: impact on global
public health security, infectious disease outbreaks, epidemics, and pandemics. Prehospital
Disaster Med 35(3):237–246
Cohen J (1992) A power primer. Psychol Bull 112(1):155–159
Corruption Cuts (2020) Global resources available to ﬁght COVID-19. Legal Monitor Worldwide,
May 2
Corruption Perception Index (2020) Transparency International. https://www.transparency.org/en/
cpi/2020 Accessed 9 Oct 2021
Crevani L, Lindgren M, Packendorff J (2007) Shared leadership: A post-heroic perspective on
leadership as a collective construction. Int J Leadersh Stud 3(1):40–67
Cruz-Ozorio J (2020) Corruption: the other global pandemic to eradicate. UNDP | United Nations
Development Programme. https://www.latinamerica.undp.org/content/rblac/en/home/
Participative Leadership Is the Discriminating Factor for Country’s. . . 455

blog/2020/corruption%2D%2Dthe-otherglobal-pandemic-to-eradicate.html Accessed
17 June 2021
Dorfman P, Javidan M, Hanges P, Dastmalchian A, House R (2012) GLOBE: A twenty year
journey into the intriguing world of culture and leadership. J World Bus 47(4):504–518
Dunning T (2012) Front matter. In: Natural experiments in the social sciences: A design-based
approach (strategies for social inquiry). Cambridge University Press, Cambridge, pp I–VI
Eastburn RW (2018) Organizational mindfulness in banking: A discriminating factor for firm
performance. J Manag Issues 30(1):122–127
Freeman RE (2016) A stakeholder theory of the modern corporation. In: Clarkson M (ed) The
corporation and its stakeholders. University of Toronto Press, Toronto, pp 125–138
Freeman RE, McVea J (2005) A stakeholder approach to strategic management. In: Hitt MA,
Freeman RE, Harrison JS (eds) The Blackwell handbook of strategic management. Blackwell,
Oxford
Frieden T (2021) Which countries have responded best to Covid-19? Wall Street J, Jan 2. Online
Edition
Friedman M (2007) The social responsibility of business is to increase its profits. In: Zimmerli WC,
Holzinger M, Richter K (eds) Corporate ethics and corporate governance. Springer, Berlin, pp
173–178
Gallego JA, Prem M, Vargas JF (2020) Corruption in the Times of Pandemia. Documentos de
Trabajo 018178, Universidad del Rosario
Gapud SD (2019) Leadership on fire: Mindfully shared. Doctoral Thesis, University of South
Alabama
Gaspar V, Mühleisen M, Weeks-Brown R (2020) Corruption and COVID-19. IMF Blog. Interna-
tional Monetary Fund, 28 July 2020. https://blogs.imf.org/2020/07/28/corruption-and-covid-19/
. Accessed 22 June 2021
Gokmen Y, Baskici C, Ercil Y (2021) The impact of national culture on the increase of COVID-19:
A cross-country analysis of European countries. Int J Intercult Relat 81:1–8
Google (2020) Google community mobility reports. https://www.google.com/covid19/mobility/.
Accessed 15 Oct 2020
Hair JF (2020) Next-generation prediction metrics for composite-based PLS-SEM. Ind Manag Data
Syst 121(1):5–11
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: Indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Sarstedt M, Ringle CM, Mena JA (2012) An assessment of the use of partial least squares
structural equation modeling in marketing research. J Acad Mark Sci 40(3):414–433
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2017) A primer on partial least squares structural
equation modeling (PLS-SEM), 2nd edn. Sage, Thousand Oaks, CA
Hair JF, Black WC, Babin B, Anderson RE, Tatham R (2019) Multivariate data analysis. Cengage,
London
Hair JF, Howard MC, Nitzl C (2020) Assessing measurement model quality in PLSSEM using
confirmatory composite analysis. J Bus Res 109:101–110
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Hofstede G (1984) Culture’s consequences: International differences in work-related values, vol
5. Sage, Thousand Oaks, CA
Hofstede G (2011) Dimensionalizing cultures: The Hofstede model in context. Online Read
Psychol Cult 2(1):2307–0919
Hofstede (Country) Insights (2020) Compare countries. https://www.hofstede-insights.com/
product/compare-countries/. Accessed 15 Oct 2020
Hong J, Lew L, Tan A, Gitau M, Varley K (2022) The Covid resilience ranking: The best and worst
places to be as the world enters next Covid phase. Bloomberg, Nov 24, webpage https://www.
bloomberg.com/graphics/covid-resilience-ranking. Accessed 17 Dec 2022
456 S. D. Gapud and G. Faint

House R, Javidan M, Hanges P, Dorfman P (2002) Understanding cultures and implicit leadership
theories across the globe: An introduction to project GLOBE. J World Bus 37(1):3–10
House RJ, Hanges PJ, Javidan M, Dorfman PW, Gupta V (eds) (2004) Culture, leadership, and
organizations: The GLOBE study of 62 societies. Sage, Thousand Oaks, CA
Huynh TLD (2020) Does culture matter social distancing under the COVID-19 pandemic? Saf Sci
130:104872
Jones BF, Olken BA (2005) Do leaders matter? National leadership and growth since World War
II. Q J Econ 120(3):835–864
Lagman JDN, Vergara RJD, Sarmiento PJD (2021) Culture as a contributory factor in combatting
the Covid-19 pandemic. J Public Health 43(2):e289–e290
Langer E (2010) A call for mindful leadership. Harv Bus Rev 28:60
Liedtka J (2006) Is your strategy a duck? J Bus Strateg 27(5):32–37
Liedtka J (2008) Strategy making and the search for authenticity. J Bus Ethics 80(2):237–248
Lord RG, Maher KJ (1992) Leadership and information processing: Linking perceptions and
performance. Routledge, London
Luengo M, García-Marín D (2020) The performance of truth: politicians, fact-checking journalism,
and the struggle to tackle COVID-19 misinformation. Am J Cult Sociol 8:405–427
Manz CC, Pearce CL (2018) Twisted leadership: How to engage the full talents of everyone in your
organization. Maven House Press, Lambertville, NJ
Mayer CH, May MS (2021) Women leaders transcending the demands of Covid-19: A positive
psychology 2.0 perspective. Front Psychol 12:1–12
Olagnier D, Mogensen TH (2020) The Covid-19 pandemic in Denmark: Big lessons from a small
country. Cytokine Growth Factor Rev 53:10–12
Pearce CL, Manz CC (2005) The new silver bullets of leadership: The importance of self- and
shared leadership in knowledge work. Organ Dyn 34(2):130–140
Pearce CL, Wassenaar CL (2014) Leadership is like fine wine. Organ Dyn 43(1):9–16
Pearce CL, Manz CC, Sims HP Jr (2008) The roles of vertical and shared leadership in the
enactment of executive corruption: Implications for research and practice. Leadersh Q 19(3):
353–359
Pearce CL, Wassenaar CL, Manz CC (2014) Is shared leadership the key to responsible leadership?
Acad Manag Perspect 28(3):275–288
Peteraf MA (1993) The cornerstones of competitive advantage: A resource-based view. Strateg
Manag J 14(3):179–191
Pozsgai-Alvarez J (2020) The abuse of entrusted power for private gain: Meaning, nature and
theoretical evolution. Crime Law Soc Chang 74(4):433–455
Rigdon EE (2012) Rethinking partial least squares path modeling: In praise of simple methods.
Long Range Plan 45(5–6):341–358
Ringle CM, Wende S, Becker J-M (2022) SmartPLS (4) [computer software]. SmartPLS GmbH
Rubio DF (2021) Corruption perceptions index 2020 media kit. https://www.transparency.org/en/
cpi/2020/media-kit. Accessed 17 Jan 2022
RUSA (2011) Best free reference web sites 2011 13th Annual List RUSA Emerging Technologies
in Reference Section (MARS), 15 June 2011. American Library Association. http://www.ala.
org/rusa/sections/mars/marspubs/marsbestfreewebsites/marsbestref2011. Accessed
16 Mar 2021
Sander T, Teh PL (2014) SmartPLS for the human resources field to evaluate a model. In:
International scientific conference new challenges of economic and business development,
8–10 May 2014, Riga, University of Latvia
Sarstedt M, Mooi E (2014) A concise guide to market research. The process, data, and methods
using IBM SPSS Statistics. Springer, New York
Sarstedt M, Ringle CM, Henseler J, Hair JF (2014) On the emancipation of PLS-SEM: A
commentary on Rigdon (2012). Long Range Plan 47(3):154–160
Sashkin M, Sashkin MG (2003) Leadership that matters: The critical factors for making a difference
in people’s lives and organizations’ success. Berrett-Koehler, Oakland, CA
Participative Leadership Is the Discriminating Factor for Country’s. . . 457

Shmueli G, Sarstedt M, Hair JF, Cheah JH, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Sieweke J, Santoni S (2020) Natural experiments in leadership research: An introduction, review,
and guidelines. Leadersh Q 31(1):101338
Tomkins L (2020) Where is Boris Johnson? When and why it matters that leaders show up in a
crisis. Leadership 16(3):331–342
Transparency International (2020) Corruption perceptions index. https://www.transparency.org/en/
cpi/2020. Accessed 22 June 2021
Transparency International (2021) What is corruption? https://www.transparency.org/en/what-is-
corruption. Accessed 22 June 2021
Tsai I-W (2020) President of Taiwan: How my country prevented a major outbreak of COVID-19.
Time, 16 Apr 2020. https://time.com/collection/finding-hope-coronavirus-pandemic/5820596/
taiwan-coronavirus-lessons/. Accessed 6 July 2021
Usman M, Ghani U, Cheng J, Farid T, Iqbal S (2021) Does participative leadership matters in
employees’ outcomes during COVID-19? Role of leader behavioral integrity. Front Psychol
12(1585):1–11
Vinopal C (2021) What we’ve learned about leadership from the COVID-19 pandemic. PBS News
Hour, Apr 6. https://www.pbs.org/newshour/world/what-weve-learned-about-leadership-from-
the-covid-19-pandemic Accessed 15 Oct 2021
Wang Y (2021) Government policies, national culture and social distancing during the first wave of
the COVID-19 pandemic: International evidence. Saf Sci 135:105138
Wasserstein RL, Schirm AL, Lazar NA (2019) Moving to a world beyond “p< 0.05”. Am Stat 73
(sup1):1–19
Weick KE, Sutcliffe KM (2011) Managing the unexpected: Resilient performance in an age of
uncertainty, vol 8. Wiley, New York
Wernerfelt B (1984) A resource-based view of the firm. Strateg Manag J 5(2):171–180
Wernli D, Tediosi F, Blanchet K, Lee K, Morel C, Pittet D et al (2021) A complexity lens on the
COVID-19 pandemic. Int J Health Policy Manag 11(11):2769–2772
Worldometers (2020) COVID-19 coronavirus pandemic. https://www.worldometers.info/coronavi
rus/#countries. Accessed 15 Oct 2020
https://globeproject.com/
A PLS-SEM Analysis of Consumer Health
Literacy and Intention to Use
Complementary and Alternative Medicine
in the COVID-19 Pandemic

Michael Christian, Henilia Yulita, Guan Nan, Suryo Wibowo,

Eko Retno Indriyarti, Sunarno Sunarno, and Rima Melati

1 Introduction

The emergence of several pandemics in various countries recently has also had an
impact on people’s actions to seek complementary or alternative medicine. This is an
act of self-care to protect oneself during social limitations and uncertainty about the
end of the pandemic. Searching for information through several media or word of
mouth often adds to the knowledge about complementary and alternative medicine

M. Christian (✉)
Department of Management, Faculty of Social and Humanities, Universitas Bunda Mulia,
Ancol, North Jakarta, Indonesia
H. Yulita
Department of Communication Studies, Faculty of Social and Humanities, Universitas Bunda
Mulia, Ancol, North Jakarta, Indonesia
G. Nan
Department of Chinese Literature, Faculty of Social and Humanities, Universitas Bunda Mulia,
Ancol, North Jakarta, Indonesia
S. Wibowo
Department of Biomedical and Bioengineering, Indonesia International Institute for Life
Sciences, East Jakarta, Indonesia
E. R. Indriyarti
Faculty of Economics and Business, Universitas Trisakti, Grogol, West Jakarta, Indonesia
e-mail: [email protected]
S. Sunarno
Psychological Science Doctoral Program, Universitas Persada Indonesia YAI, Kenari, Central
Jakarta, Indonesia
R. Melati
Faculty of Medicine, President University, Kota Jababeka, Cikarang Baru, Bekasi, Indonesia

(CAM). Several previous studies have explained the linkage of CAM in various
pandemics, such as influenza A virus subtype H1N1 (H1N1) (Arora et al. 2011; Tao
et al. 2013), Middle East respiratory syndrome coronavirus (MERS-CoV) (Hwang
et al. 2020), and COVID-19 (Charan et al. 2021; Lam et al. 2021). This explains the
role of CAM in the community in its function as complementary medicine or a
system of healing. Within CAM, a popularly used method is acupuncture (Greville-
Harris et al. 2016; Hopton et al. 2012; Pengpid and Peltzer 2018). Acupuncture, as
part of CAM, will be used in this study. Acupuncture in Chinese culture is known as
Zhēn cì (针刺), has been practiced for more than 3000 years, is well developed, and
is accepted by people and health-care professionals across the world as a natural
healing therapy (Zhuang et al. 2013). This development has an indirect impact on
public health literacy on acupuncture. Health literacy on CAM has several indica-
tors, such as knowledge and understanding of health information obtained (Smith
et al. 2016), benefits obtained, risk perception (Bains and Egede 2011), credibility
and expectations of the treatment performed (Wilhelm and Euteneuer 2021), and so
on. This explains why, despite the increase in use, information and perceptions that
exist in the community regarding acupuncture are still diverse. Some people con-
sider it a benefit based on beliefs and culture, but a significant number of people also
know the benefits based on medical science. However, as an alternative, acupuncture
cannot be separated from a series of opposing opinions on the benefits it provides.
This could be due to not many people knowing about acupuncture and the limited
availability of acupuncture as an alternative treatment for the community. Moreover,
during this COVID-19 pandemic when the public eagerly awaits and researchers are
trying to find drugs to prevent or cure diseases, there is still debate on the pros and
cons of treatment using, for example, complementary medicine (Nugraha et al. 2020;
Zhang and Liu 2020). Health literacy, for example, related to the level of good health
awareness can have an impact on better actions and attitudes toward preventive
efforts against COVID-19 even though in this case knowledge does not have a
significant impact (Silva and Santos 2021).
The above condition is common in Indonesia where existing knowledge about
acupuncture is not proportional to its benefits and risk perception factors. Although
the use of acupuncture is increasing in Indonesia, especially in Jakarta, not many
people know much about it (Wijanto et al. 2019). Most of the existing knowledge
about acupuncture is still based on perception, obtained from various sources,
including print, electronic, online, and word of mouth. Related to this, the problem
that arises next is whether the mostly perception-based knowledge can shape
people’s intentions to use acupuncture. Thus, this chapter aims to answer two
main problems, namely to measure how much people know about acupuncture
and the impact of the public’s knowledge about acupuncture on the intention to
use it. The current research mostly focuses on health literacy about acupuncture in
general (Barnes et al. 2018; Bishop et al. 2019; Gao et al. 2021; Greville-Harris et al.
2016; Hao and Mittelman 2014; Zhao et al. 2018), but there are not many studies that
explain knowledge about acupuncture related to aspects of risks or side effects
perceived by the community, especially in finding a complementary health system
during this COVID-19 pandemic. In addition, there are still fewer studies that focus
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 461

on people’s intentions to use acupuncture as an alternative treatment, especially in

Indonesia. The originality of this chapter lies in the use of two important aspects
related to alternative medicine that are not widely known, namely acupuncture health
literacy through knowledge and its usefulness aspects. In addition to the use of
acupuncture risk moderator variables perceived by the community, it completes the
measurement of public anxiety about alternative medicine that has not been widely
used. Data collection carried out during the COVID-19 pandemic is expected to be
able to enrich the results of this chapter, but also to be meaningful for further
research in other countries. This is because the study measured public perceptions
of alternative medicine in special conditions with the global, pro-longed COVID-10
pandemic.

2 Literature Review and Hypothesis Development

2.1 Knowledge on Acupuncture Health Literacy

Psychosocial aspects of health such as the ability or skill to ﬁnd out and understand
information and act on health decisions can explain an individual’s level of health
literacy (Mantwill et al. 2015; Sørensen et al. 2012). It is further explained that
literacy has various aspects, such as knowledge, behavior, attitudes, or psychosocial
temperament. Tuohetamu et al. (2017) explained that health education is useful for
increasing health awareness. This indirectly explains that health education is closely
related to knowledge of a series of health actions that need to be taken. Furthermore,
Dalrymple (2011) in the pyramid of informatics explained that health knowledge is
determined from the data and information obtained, which in the next stage will
result in a decision or an action to be taken. Acupuncture is still used by the
community as a complement or an alternative for self-protection, such as during
the MERS pandemic (Hwang et al. 2020) or COVID-19 as it is today (Lam et al.
2021; Paudyal et al. 2021). The number of COVID-19 cases in Indonesia, especially
in Jakarta, along with a slew of opinions on the general public’s mistrust of COVID-
19 calls for in-depth investigation. For this reason, this chapter proposes a hypothesis
(H) as follows:
H1: Knowledge about acupuncture affects the intention to use acupuncture.

2.2 Usefulness of Acupuncture

Due to the public’s apprehensions about the effectiveness of vaccines to ward off
COVID-19, many people continued to try alternative medicine, such as acupuncture.
This placed acupuncture between the beneﬁts and risks as perceived by society. The
intention to use alternative medicine is motivated by several aspects such as
462 M. Christian et al.

attitudes, views, and experiences (Kristoffersen et al. 2018). On the other hand,
Sirois et al. (2016) explained that apart from the effectiveness aspect of alternative
medicine and the values or views of alternative medicine itself, the aspect of user
commitment is a determining factor that governs alternative medicine user behavior.
In this case, there are still a series of perceptions that are in effect public concerns
about the risks of acupuncture, which has proven beneﬁts. Therefore, related to this,
health literacy is also quite dependent upon several other speciﬁc aspects such as
individual character, environment, social and individual burden (Martin et al. 2011),
and so on. A study conducted by Luh and Eckstrom (2021) in Oregon explained that
most acupuncturists and hospital staff agree or somewhat agree that knowledge is an
important aspect of therapeutic modalities. In addition, it was explained that acu-
puncture can help provide quality of life for hospital patients even though it can be
affected by various knowledge factors. There is still a perception of acupuncture risk
that is moderated by various levels of knowledge that can affect people’s behavior in
using acupuncture. Therefore, it is necessary to conduct more diverse studies. Based
on these results, this chapter proposes the following hypotheses:
H2: Usefulness affects the intention to use acupuncture.
H3: Perceived risk affects the intention to use acupuncture.
H4: Perceived risk moderates knowledge about acupuncture on the intention to use
it.
H5: Perceived risk moderates the usefulness of the intention to use acupuncture.
Thus, based on the explanation of the development of the hypotheses above, the
conceptual model is shown in Fig. 1.

3 Methodology

The quantitative method adopted in this study used the survey and questionnaire as
instruments for collecting data from participants during February–April 2021. The
analytical method of this chapter is divided into two stages of analysis. The ﬁrst
analysis uses frequency statistics to explain how much participants know about
acupuncture using 10 statements measured on a “yes (true)” and “no (false)” scale.
These statements adapt the measurements of Greville-Harris et al. (2016), which
consist of two statements to measure participants’ knowledge about the beneﬁts of
acupuncture, three statements to measure participants’ knowledge about the side
effects of acupuncture, and four statements to measure participants’ knowledge
about aspects of acupuncture’s technique. In addition, this chapter also felt the
need to measure the participants’ knowledge of one statement about the legality of
medical acupuncture in health care in Indonesia.
The second method of analysis in this chapter measured the impact of health
literacy on acupuncture on the intention to use it with a Likert scale from 1 (strongly
disagree) to 5 (strongly agree). The statement on the variables of this chapter adapts
the indicators from Hwang et al. (2020) and Omar and Putit (2012), in which the
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 463

Moderating Effect
1

Knowledge

Usefulness Intention to use

Risk

Moderating Effect
2

Fig. 1 Conceptual model—PLS-SEM

intention to use acupuncture variable summarizes four statements that explain

interest or desire to use acupuncture either shortly or in the future, no desire to
use, and the desire to recommend acupuncture to others. The knowledge about
acupuncture variable consists of three statements, namely knowledge to distinguish
alternative and complementary medicine, knowledge of whether scientiﬁc validation
is necessary, and the interest of pharmaceutical companies in acupuncture. The
usefulness variable consists of four statements that explain the validity of acupunc-
ture as an alternative treatment, function of health in quality of life, ability to reduce
side effects of conventional treatment, and source of more personalized/humane
treatment. Furthermore, the risk variable as a moderator in this chapter consisted of
three statements that explained the dangers of using acupuncture and an assessment
of whether acupuncture could hinder the conventional treatment that the patient was
undergoing.
The criterion for participants in this chapter was people in the Jakarta area who
knew acupuncture as an alternative/complementary treatment, whether or not they
had used the method. The random sampling technique was chosen, and the sample
size was determined by multiplying the total items/indicators by 5–10. The tech-
nique popularized by Hair et al. (2014) has also been widely used in various studies,
464 M. Christian et al.

such as Christian et al. (2022, 2023) and Indriyarti et al. (2022). Furthermore, the
total sample collected after ﬁltering according to the criteria described above
consisted of 185 participants. The analysis in this chapter was performed by partial
least square structural equation modeling (PLS-SEM) using the SmartPLS software.
The concept of a structural model with this analytical tool is generally used to
analyze causal models with relatively small sample sizes (Wolf et al. 2013).
PLS-SEM is also a prominent method for estimating path models with latent vari-
ables and their relationships (Sarstedt et al. 2021).

4 Results and Discussion Hypothesis Development

4.1 Distribution of Participants’ Proﬁles

As shown in Table 1, of the total participants (N = 185), women were more than
70% and men were less than 30%. Furthermore, in terms of age, ages under 27 years
(80%) dominated, which was also corroborated by the parameter of background as a
student (more than 70%). More than half of the total participants involved did not
have comorbidities, whereas those who were not sure whether they had
comorbidities or not were quite high, almost 35%. Looking at these conditions, it
can be said that when viewed in terms of age and educational background, it was
quite possible that participants were not aware of their health conditions in terms of
comorbidities. Therefore, it is necessary to increase awareness and health literacy for
personal well-being in the future. Speciﬁcally, this chapter also explains that more
than 80% of participants had never tried acupuncture. However, those who had tried
but stopped were less than 16% and those who continued to use acupuncture until
now were less than 2%. Furthermore, participants already had information about
alternative acupuncture treatment that they had obtained from various sources, for
example, information from family or friends. It amounted to almost 50% of the total
participants in this study, whereas those who obtained information from Internet
sources comprised more than 30% and from television were less than 13%.

4.2 Health Literacy on Acupuncture

Participants in this chapter had good knowledge in recognizing acupuncture based

on its benefits. As shown in Fig. 2, of the total participants, the majority answered
correctly, that is, that acupuncture can provide benefits for natural pain relief in the
brain (75.14%) and pain relief (90.27%). This result is in line with that of Greville-
Harris et al. (2016), in which research conducted in the UK also explains that both
aspects of these benefits are well known to the public.. Thus, this chapter can explain
that regardless of whether the individual has used acupuncture or not, its benefits are
familiar. Interestingly, there are still many participants who do not know that
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 465

Table 1 Distribution of participants’ proﬁles

Proﬁle N %
Gender
Female 132 71.4
Male 53 28.6

Age
<27 years old 148 80
27–40 years old 7 3.8
41–56 years old 27 14.6
>56 years old 3 1.6
Job
Student 144 77.8
Private employees 18 9.7
Entrepreneur 9 4.9
Housewife 6 3.2
Teacher 6 3.2
Government employee 1 0.5
Clergy 1 0.5
Do you have comorbidities?
No 116 62.7
Probably 60 32.4
Yes 9 4.9
Have ever used acupuncture?
Never 153 82.7
Yes, but stopped 29 15.7
Yes, until now 3 1.6
Source of acupuncture information
Family/friend 90 48.6
Internet 72 38.9
Television 23 12.4

Acupuncture can release natural pain relieving chemicals in the

brain. 139 46
Pain can be relieved with acupuncture. 167 18
Acupuncture is not included in the field of specialization of 105 80
medical science in Indonesia.
Acupuncturists are not allowed to give acupuncture to several 102 83
patients at the same time.
Dizziness may occur in some patients on the first experience of 66 119
acupuncture.
Acupuncture never uses disposable needles. 69 116

Acupuncture needles and injection needles are the same size. 55 130

Bruising may occur for some acupuncture patients. 61 124

Acupuncture needles are the size of a human hair. 138 47

Scarcity of side effects due to acupuncture. 130 55

Yes No

Fig. 2 Distribution of participants’ acupuncture health literacy

466 M. Christian et al.

acupuncture falls under medical care in Indonesia. More than half (56.76%) of the
total participants did not know that the medical acupuncture has been recognized
within the medical field in Indonesia. Referring to the official website of medical
acupuncture in Indonesia, the Indonesian Doctors Association (IDI) has, since 2002,
endorsed the management of the Indonesian Association of Acupuncture Experts
(PDAI). Pengpid and Peltzer (2018) in a study of traditional and complementary
medicine in Indonesia summarized that in 2012 the Indonesian Ministry of Health
had received registrations of more than 280,000 traditional and alternative medicine
practitioners, which included acupuncture treatment techniques or methods.
Furthermore, from the aspect of knowledge wherein participants judged the side
effects of acupuncture, the majority rated two out of three statements correctly.
About 70.27% correctly judged that, in general, the side effects of acupuncture were
rare. The participants dominantly (67.03%) correctly judged that acupuncture did not
cause bruising. However, in patients who are using acupuncture for the first time,
side effects such as dizziness are very likely. The participants in this statement are
more dominant in judging the opposite (64.32%). So, in this case, the participants
have less knowledge of the side effects of dizziness, especially for people who had
tried acupuncture for the first time. The next statement given to the participants was
general knowledge about the acupuncture method. Three statements regarding
participants’ knowledge of needles used in acupuncture were rated correctly by the
participants in this chapter. About 74.59% correctly judged that the size of the
acupuncture needles used was approximately the size of a human hair. About
70.27% of participants also had a good knowledge in assessing acupuncture needles
differently from syringes in general. No less important is that 62.70% of participants
have good knowledge in assessing that acupuncture needles are not disposable
needles. Knowledge of whether an acupuncturist is allowed to give acupuncture
simultaneously with several patients (multi-bed) is uncertain where 55.14% of
participants are still mistaken in assessing “no.” However, acupuncture can be
performed simultaneously on several patients with certain position settings.

4.3 Reliability and Validity Test

To account for the reliability and validity, we checked if outer loadings (OL),
composite reliability (CR), and average variance extracted (AVE) meet the mini-
mum thresholds as suggested in the literature (e.g., Barati et al. 2019; Christian et al.
2021a, b; Memon and Rahman 2014): OL > 0.7, CR > 0.7 and AVE > 0.5. In the
ﬁrst measurement, several items do not meet the requirements so that they must be
eliminated and retested. Furthermore, the results of CR and AVE in Table 2 show
that each variable has a CR > 0.7 and an AVE > 0.5. With these results, it can be
explained that all variables are reliable and valid.
As stated in Table 3, this chapter also looks at discriminant validity using the
heterotrait–monotrait ratio of correlations (HTMT) < 0.9. Based on the ﬁndings, it
can be concluded that all variables matched the criteria for discriminant validity.
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 467

Table 2 Reliability and validity test

Variable Item OL CR AVE
Knowledge of acupuncture KNOW1 0.771 0.777 0.636
KNOW3 0.824
Usefulness USEF1 0.797 0.850 0.653
USEF2 0.841
USEF4 0.786
Intention to use INTE1 0.861 0.875 0.701
INTE2 0.844
INTE4 0.806
Notes: There is only one item (RISK1) in the risk variable that meets the requirements, so it is not
shown in this table

Table 3 Discriminant validity

Variable Intention to use Knowledge about acupuncture Risk
Knowledge about acupuncture 0.451
Risk 0.417 0.690
Usefulness 0.711 0.738 0.596

Table 4 PLSpredict
RMSE MAE MAPE Q2predict
PLS- PLS- PLS- PLS-
Description SEM LM SEM LM SEM LM SEM LM
INTE1 0.782 0.797 0.594 0.615 25.809 26.705 0.200 0.168
INTE2 0.926 0.976 0.710 0.738 30.425 31.953 0.197 0.108
INTE3 1.079 1.084 0.719 0.852 48.931 49.443 0.201 0.025
INTE4 0.792 0.805 0.610 0.619 21.613 22.306 0.281 0.258

Table 4 shows a predictive relevance test that employs the PLSpredict. If the root
mean square error (RMSE), the mean absolute error (MAE), and the mean absolute
percentage error (MAPE) values corresponding to the PLS-SEM model are below
the ones of the linear model (LM), the PLS-SEM model is good. The Q2predict value
of the PLS-SEM must then be greater than that of the LM (Hair et al. 2019, 2022).
The results in Table 4 demonstrated that the PLS-SEM model is good.

4.4 Hypothesis Testing

Table 5 shows that the standard deviation on the path of knowledge about acupunc-
ture, risk, and moderation effect 1 on the intention to use is greater than the mean.
This explains that the data used vary, whereas the other path explains the opposite
result. Furthermore, by using the 95% conﬁdence interval, Table 5 also shows that
468 M. Christian et al.

Table 5 Hypothesis testing

Standard Sample t- p-
Path deviation mean statistics value Remark
Knowledge about acupuncture ! 0.071 0.020 0.002 0.998 Reject
Intention to use H1
Usefulness ! Intention to use 0.060 0.603 10.189 0.000 Accept
H2
Risk ! Intention to use 0.074 0.028 0.359 0.719 Reject
H3

Table 6 R2 value
Variable R2 R2 adjusted
INTE1 0.407 0.390

for the results of hypothesis testing based on the results of t-statistics and p-value to
support the existing hypothesis, the t-statistics must be > 1.96 and the p-value must
be < 0.05. Based on these results, there are four rejected hypotheses and one
accepted hypothesis. Knowledge of the intention to use acupuncture is not proven,
or in other words, Hypothesis 1 is rejected. Similarly, the perception of risk toward
the use of acupuncture is not proven to affect the intention to use (does not support
H3). Meanwhile, the usefulness aspect can explain its effect on the intention to use
(supports H2). In both moderating effects, public perception of the risks of using
acupuncture did not moderate knowledge about acupuncture and usefulness of the
intention to use acupuncture. With this result, H4 and H5 are rejected.

4.5 Coefﬁcient Determination

These results explain that the exogenous variables used in this chapter were able to
explain the endogenous variables by 40.7%. This ﬁgure is sufﬁcient to contribute to
research in explaining the aspects of the intention to use acupuncture in the partic-
ipants of this study. This also explains that the value of R2 is close to moderate (Hair
et al. 2019) (Table 6).

4.6 Discussion

The results of this chapter explain that public knowledge of acupuncture does not
affect the intention to use it. These results are in line with the research of Greville-
Harris et al. (2016), which emphasizes that the level of knowledge about acupunc-
ture does not have an impact on decision-making actions on using it. On the other
hand, Bishop et al. (2019) explained that the level of public knowledge in receiving
information about acupuncture, for example through online sources, can affect the
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 469

intention to use it. Similar to other studies (Hao and Mittelman, 2014; Gao et al.
2021), this chapter confirms variations in the public’s knowledge about acupuncture,
particularizing on the knowledge a) about acupuncture as CAM compared to other
types of treatment, b) of whether scientific validation is necessary, and c) about the
interest of pharmaceutical companies in acupuncture.
The desire to use acupuncture (both soon and in the future) by the participants of
this study is considered to indicate aspects of urgent need or spontaneous use. This is
based on the perception that exists in a society where acupuncture is better known for
its function as a complementary or companion treatment. Public health conditions
can greatly determine whether acupuncture is used as an urgent necessity (Barnes
et al. 2018; Zhao et al. 2018) or just a trial (Jiao et al. 2020). In other words, the level
of public knowledge of acupuncture does not shape people’s actions as a willingness
to use it. The current COVID-19 pandemic has greatly affected the health conditions
of both those who are exposed and those who are not exposed to COVID-19. Seeing
this condition, acupuncture can be one of the companion options for the community
(Badakhsh et al. 2021; Paudyal et al. 2021). In addition, several other conditions
(difficulty in obtaining medical drugs, limited access to a doctor’s examination, or
concerns about conducting medical checkups in hospitals) that occurred owing to the
pandemic made the aspect of using acupuncture as a necessity or experiment even
higher. Different levels of community knowledge are not an obstacle for people’s
intentions to use acupuncture.
It is no secret that medical acupuncture can provide benefits for the health of the
human body. The results of this chapter explain that the perceived benefits of
acupuncture affect the intention to use acupuncture. This chapter succeeded in
explaining that the benefits of acupuncture such as health support for a better quality
of life and reducing the side effects of using conventional medicine to get more
personalized treatment have been formed in public perception. Several studies also
explained the general benefits of acupuncture, which also have an impact on
behavior to use acupuncture, such as relieving pain (Sadeghi et al. 2014), improving
health (Sayampanathan et al. 2015), or reducing the risk of depression and anxiety in
migraine sufferers (Liao et al. 2020). The COVID-19 pandemic has shaped people’s
behavior patterns to independently seek CAM based on the perception of the benefits
obtained. As mentioned above, acupuncture can function as a prophylactic against
COVID-19. Acupuncture can be useful for helping the recovery process of positive
patients. In addition, from a psychological point of view, acupuncture can also be
useful in the management of post-traumatic stress disorder (Chen 2020).
This chapter also proves that the risk of acupuncture on participants’ perceptions
has no impact on the intention to use it. Furthermore, the results of this chapter
explain that the perception of risk also does not act as a moderator of the relationship
between knowledge about acupuncture and usefulness to the intention to use acu-
puncture. The research of Chan et al. (2016) explained that the perception of
acupuncture risks only applies to non–acupuncture users. Risk perceptions that are
formed include the assumption that acupuncture is not clinically based so that it can
create concerns about its use. In addition, acupuncture is considered to have a high
risk. Another concern that poses a risk in the perception of nonusers is about the
470 M. Christian et al.

acupuncture treatment itself, which is quite dependent on the expertise of medical

practitioners. This is very likely because health knowledge about acupuncture for
nonusers is still not sufficiently formed, so the risk perception factor becomes the
more dominant influencing factor. In addition, Bishop and Lewith (2013) explained
that risk also appears in the user’s first experience. This is due to concerns that the
expectations of acupuncture treatment will be met. Aspects of the benefits in this
chapter as described above can shape the behavior of participants to act on the
intention to use acupuncture. On the other hand, the risk did not affect the relation-
ship between the benefits and the intention to use acupuncture. Participants who
already know the benefits of acupuncture are not affected by the perceived risk of
using acupuncture. In this regard, participants who had experience using acupunc-
ture felt more benefit than perceived risk. To be able to optimally spread the benefits
of acupuncture and reduce the risk perception that is formed in the community,
Bishop and Salmon (2013) explained that information about acupuncture can be
conveyed through leaflets but must ensure that the information is ethical, scientific,
informative, and accurate and includes the risks. This is also in line with the current
state of the COVID-19 pandemic, in which the dissemination of information,
especially that related to health literacy, is expected to emphasize good and positive
things to build public trust.

5 Conclusion and Implications for Theory and Practice

This chapter explains that the intention to use acupuncture is not affected by
knowledge about it or risks. Participants view that the usefulness aspect dominates
the behavior of using acupuncture during this pandemic. Although participants have
different levels of knowledge about acupuncture, they agree that acupuncture treat-
ment has good health benefits. Therefore, during this pandemic, acupuncture can be
used for the community. This chapter has limitations, namely the first being the
uncertain conditions of the COVID-19 pandemic, which make policy adjustments
and social restrictions that affected the speed of the research process. Second, the
variables used in this chapter focused on only the personal aspects of the partici-
pants, namely knowledge about acupuncture, benefits, and risks, whereas other
aspects, such as financial conditions during difficult times during the pandemic,
are also important. Third, the sample size and the research area taken are relatively
small, so they need to be expanded in future studies.
As a theoretical implication, this chapter emphasizes the importance of under-
standing people’s intentions to use acupuncture in the health care system, especially
during a pandemic. Under normal conditions, risk factors and customer knowledge
about acupuncture determine the use of health services. When social limitations in
obtaining health services become an obstacle, the intention to use acupuncture is
focused on perceived usefulness. Due to COVID-19 concerns, it is necessary to
maintain health in various ways. Thus, perceived usefulness is shaped by personal
motivational factors.
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 471

As a practical implication, this chapter highlights that due to the COVID-19

pandemic, people are still looking for a series of alternative medicine treatments.
This is because the public can still maintain and improve their immunity as a self-
protection measure against COVID-19. However, acupuncture has barriers to use,
mainly due to perceptions that exist in society about its risks. In this circumstance,
acupuncture business actors can connect with the public as regular or new customers.
This can be performed by using social media or collaborating with widely used
health service apps during the pandemic.

References

Arora R, Chawla R, Marwah R, Arora P, Sharma RK, Kaushik V, Goel R, Kaur A, Silambarasan M,
Tripathi RP, Bhardwaj JR (2011) Potential of complementary and alternative medicine in
preventive management of novel H1N1 flu (swine flu) pandemic: thwarting potential disasters
in the bud. Evid Based Complement Altern Med ECAM 58650:1–16
Badakhsh M, Dastras M, Sarchahi Z, Doostkami M, Mir A, Bouya S (2021) Complementary and
alternative medicine therapies and COVID-19: a systematic review. Rev Environ Health 36(3):
443–450
Bains SS, Egede LE (2011) Association of health literacy with complementary and alternative
medicine use: a cross-sectional study in adult primary care patients. BMC Complement Altern
Med 11(138):1–7
Barati M, Taheri-Kharameh Z, Farghadani Z, Rásky É (2019) Validity and reliability evaluation of
the Persian version of the heart failure-specific health literacy scale. Int J Community Based
Nurs Midwifery 7(3):222–230
Barnes MA, Bennett J, Ross J, Kraemer K, Cottion S (2018) Single-point acupuncture for treatment
of urge incontinence in women: a pilot nonrandomized trial. Med Acupunct 30(1):21–24
Bishop FL, Lewith GT (2013) Patients’ preconceptions of acupuncture: a qualitative study explor-
ing the decisions patients make when seeking acupuncture. BMC Complement Altern Med
13(102):1–10
Bishop FL, Salmon C (2013) Advertising, expectations, and informed consent: the contents and
functions of acupuncture leaflets. Acupunct Med 31(4):351–357
Bishop FL, Greville-Harris M, Bostock J, Din A, Graham CA, Lewith G, Liossi C, O’Riordan T,
White P, Yardley L (2019) Supporting informed choice in acupuncture: effects of a new person-,
evidence- and theory-based website for patients with back pain. Acupunct Med 37(2):98–106
Chan K, Siu JY-M, Fung TKF (2016) Perception of acupuncture among users and non-users: A
qualitative study. Health Mark Q 33(1):78–93
Charan J, Bhardwaj P, Dutta S, Kaur R, Bist SK, Detha MD, Kanchan T, Yadav D, Mitra P, Sharma
P (2021) Use of complementary and alternative medicine (CAM) and home remedies by
COVID-19 patients: a telephonic survey. Indian J Clin Biochem 36(1):108–111
Chen Y (2020) The role of acupuncture during the COVID-19 pandemic: From the historical
perspective to practical application. Tradit Med Mod Med 3(1):27–35
Christian M, Dewi D, Rembulan GD, Indriyarti ER, Wibowo S, Yuniarto Y (2021a) Business
performance determinants of salted fish distribution in Kapuk during the COVID-19. J Distrib
Sci 19(6):29–39
Christian M, Haris K, Indriyarti ER, Wibowo S, Sunarno S (2021b) Service distribution strategy on
business performance of Padang restaurants in North Jakarta. J Distrib Sci 19(12):57–69
Christian M, Pardede R, Indriyarti ER (2022) Generation Z in Jakarta’s attitude towards COVID-19
ad distribution on YouTube. J Distrib Sci 20(3):13–22
472 M. Christian et al.

Christian M, Wibowo S, Indriyarti ER, Sunarno S, Yuniarto Y (2023) Do service quality and
satisfaction affect the intention of using application-based land transportation? A study on
generation YZ in Jakarta BT. In: Hamdan A, Shoaib HM, Alareeni B, Hamdan R (eds) The
implementation of smart technologies for business success and sustainability. Springer,
New York, pp 737–746
Dalrymple PW (2011) Data, information, knowledge: the emerging field of health informatics. Bull
Am Soc Inf Sci Technol 37(5):41–44
Gao Z, Zhang J, Liu G-F, Ji L-X (2021) Research trends from 2010 to 2020 for pain treatment with
acupuncture: a bibliometric analysis. J Pain Res 14:941–952
Greville-Harris M, Hughes J, Lewith G, Liossi C, White P, Graham CA, Bishop FL (2016)
Assessing knowledge about acupuncture: A survey of people with back pain in the
UK. Complement Ther Med 29:164–168
Hair JF, Anderson RE, Black WC (2014) Multivariate data analysis, 7th edn. Pearson, Upper Sdler
River
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair J, Hult GTM, Ringle C, Sarstedt M (2022) A primer on partial least squares structural equation
modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks
Hao JJ, Mittelman M (2014) Acupuncture: past, present, and future. Global Adv Health Med 3(4):
6–8
Hopton AK, Curnoe S, Kanaan M, MacPherson H (2012) Acupuncture in practice: mapping the
providers, the patients, and the settings in a national cross-sectional survey. BMJ Open 2:1–9
Hwang JH, Cho HJ, Im HB, Jung YS, Choi SJ, Han D (2020) Complementary and alternative
medicine use among outpatients during the 2015 MERS outbreak in South Korea: a cross-
sectional study. BMC Complement Med Ther 20(147):1–10
Indriyarti ER, Christian M, Yulita H, Ruminda M, Sunarno S, Wibowo S (2022) Online food
delivery app distribution and determinants of Jakarta’s gen Z spending habits. J Distrib Sci
20(7):73–86
Jiao R, Huang M, Zhang W, Liu Z (2020) Efficacy of acupuncture in improving the symptoms and
the quality of life of patients with moderate or severe acne vulgaris: study protocol for a
randomized controlled trial. Trials 21(563):1–8
Kristoffersen AE, Stub T, Musia F, Fønnebø V, Lillenes O, Norheim AJ (2018) Prevalence and
reasons for intentional use of complementary and alternative medicine as an adjunct to future
visits to a medical doctor for chronic disease. BMC Complement Altern Med 18(109):1–8
Lam CS, Koon HK, Chung VC-H, Cheung YT (2021) A public survey of traditional, complemen-
tary, and integrative medicine use during the COVID-19 outbreak in Hong Kong. PLoS One
16(7):1–15
Liao C-C, Liao K-R, Lin C-L, Li J-M (2020) Long-term effect of acupuncture on the medical
expenditure and risk of depression and anxiety in migraine patients: a retrospective cohort study.
Front Neurol 11(321):1–8
Luh C, Eckstrom E (2021) Perceptions that influence acupuncture use in hospice settings: results of
a state-wide survey study. J Altern Complement Med 27(9):760–770
Mantwill S, Monestel-Umaña S, Schulz PJ (2015) The relationship between health literacy and
health disparities: a systematic review. PLoS One 10(12):1–12
Martin LT, Schonlau M, Haas A, Rosenfeld L, Derose KP, Buka SL, Rudd R (2011) Patient
activation and advocacy: which literacy skills matter most? J Health Commun 30(16):177–190
Memon AH, Rahman IA (2014) SEM-PLS analysis of inhibiting factors of cost performance for
large construction projects in Malaysia: perspective of clients and consultants. Sci World J 1–9
Nugraha RV, Ridwansyah H, Ghozali M, Khairani AF, Atik N (2020) Traditional herbal medicine
candidates as complementary treatments for COVID-19: a review of their mechanisms, pros and
cons. Evid Based Complement Altern Med 1–12
Omar UH, Putit L (2012) Consumer behavioral intention to use complementary alternative med-
icine. Int Proc Econ Dev Res 46(22):116–120
A PLS-SEM Analysis of Consumer Health Literacy and Intention to. . . 473

Paudyal V, Sun S, Hussain R, Abutaleb MH, Hedima EW (2021) Complementary and alternative
medicines use in COVID-19: A global perspective on practice, policy, and research. Res Soc
Adm Pharm 18(3):2524–2528
Pengpid S, Peltzer K (2018) Utilization of traditional and complementary medicine in Indonesia:
Results of a national survey in 2014–15. Complement Ther Clin Pract 33:156–163
Sadeghi R, Heidarnia MA, Tafreshi MZ, Rassouli M, Soori H (2014) The reasons for using
acupuncture for pain relief. Iran Red Crescent Med J 16(9):1–7
Sarstedt M, Ringle CM, Hair JF (2021) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg AE (eds) Handbook of market research. Springer, Cham
Sayampanathan AA, Koh THB, Kong KH, Low YP (2015) Factors affecting decision-making of
patients choosing acupuncture in a public hospital. Ann Transl Med 3(19):1–12
Silva MJ, Santos P (2021) The impact of health literacy on knowledge and attitudes towards
preventive strategies against COVID-19: a cross-sectional study. Int J Environ Res Public
Health 18(5421):1–12
Sirois FM, Salamonsen A, Kristoffersen AE (2016) Reasons for continuing use of Complementary
and Alternative Medicine (CAM) in students: a consumer commitment model. BMC Comple-
ment Altern Med 16(75):1–9
Smith CA, Chang E, Brownhill S, Barr K (2016) Complementary medicine health literacy among a
population of older Australians living in retirement villages: a mixed methods study. Evid Based
Complement Alternat Med 1–9
Sørensen K, Broucke SVD, Fullam J, Doyle G, Pelikan J, Slonska Z, Brand H, (HLS-EU)
Consortium Health Literacy Project European (2012) Health literacy and public health: A
systematic review and integration of definitions and models. BMC Public Health 12(80):1–13
Tao Z, Yang Y, Shi W, Xue M, Yang W, Song Z, Yao C, Yin J, Shi D, Zhang Y, Cai Y, Tong C,
Yuan Y (2013) Complementary and alternative medicine is expected to make greater contribu-
tion in controlling the prevalence of influenza. Biosci Trends 7(5):253–256
Tuohetamu S, Pang M, Nue X, Mahemuti M, Mohemaiti P, Qin Y, Peng Z, Zheng J, Yu H, Feng L,
Feng Z (2017) The knowledge, attitudes, and practices on influenza among medical college
students in Northwest China. Hum Vaccines Immunother 13(7):1688–1692
Wijanto C, Tamtomo D, Joebagyo H (2019) Recognizing acupuncture therapist and services. J
Health Policy Manag 4(2):128–138
Wilhelm M, Euteneuer F (2021) Does health literacy make a difference? Comparing the effect of
conventional medicine versus homeopathic prescribing on treatment credibility and expectancy.
Front Psychol 12:581255
Wolf EJ, Harrington KM, Clark SL, Miller MW (2013) Sample size requirements for structural
equation models: an evaluation of power, bias, and solution propriety. Educ Psychol Meas
73(6):913–934
Zhang L, Liu Y (2020) Potential interventions for novel coronavirus in China: A systematic review.
J Med Virol 95(4):479–490
Zhao Y, Zhou J, Mo Q, Wang Y, Yu J, Liu Z (2018) Acupuncture for adults with overactive
bladder. Medicine (Baltimore) 97(8)
Zhuang Y, Xing J-J, Li J, Zeng B-Y, Liang F-R (2013) Chapter one—History of acupuncture
research. Int Rev Neurobiol 111:1–23
Patient Satisfaction of Brazilian Military
Health-Care System: An Exploratory Study
by Multivariate Analysis

Ari Melo-Mariano, Simone Borges Simão Monteiro,

Ronaldo Rodrigues Pacheco, and Marcelo Ladeira

1 Introduction

A quick search over the past 3 years for “Health Services” and “Satisfaction” in
crucial databases such as Medline, Scopus, and Web of Science immediately returns
results related to the adoption of information systems, telemedicine, and the Internet
of things (Haddad et al. 2019; Pal et al. 2018).
Although current results are related to the interaction of technology with patient
satisfaction and health care, some places, such as Brazil, still need to focus on more
elementary topics, such as patient satisfaction with the service provided. Currently,
45% of Brazilians claim to be dissatisfied with the country’s public health care,
being this sector a problem ahead of unemployment, corruption, and education for a
significant portion of the population (Silva and Lima 2019).
However, due to the continental dimensions of the Brazilian territory, not all
localities have the same available care network as the large centers, reducing the
options as they move away from them. Although the military hospitals have a
defined public, in the Brazilian territory, they assume a fundamental role as health-
care providers for the military and the population in areas where there is no other
option.

A. Melo-Mariano (✉) · S. B. S. Monteiro · R. R. Pacheco

University of Brasília, Production Engineering Department, DataLab, University Campus Darcy
Ribeiro, Brasília-DF, Brazil
e-mail: [email protected]; [email protected]
M. Ladeira
University of Brasília, University Campus Darcy Ribeiro, Brasília, DF, Brazil
e-mail: [email protected]

The lack of studies published in indexed databases on the satisfaction of the

services provided in the military hospitals makes this theme a significant shortcom-
ing and is, therefore, of great interest for the study.
Thus, this chapter seeks to explore the question: How to identify the services that
add the most value to the user satisfaction of Brazilian military hospitals?
Patient satisfaction is a critical indicator in assessing health-care service quality.
Thus, many organizations seek their customer’s voices through satisfaction surveys,
most often conducted with questionnaires (Grenier et al. 2010). Unfortunately,
results from these surveys generally allow health-care professionals to identify
which factors ought to be improved other than allowing managers to understand
the patient’s actual needs, enabling them to make informed investment decisions to
deliver better quality services (Batbaatar et al. 2017).
Thus, understanding the factors influencing the satisfaction of health-care service
users in the military hospitals is vital to contribute to improvements and strategies for
public hospitals in similar situations.
Therefore, this chapter aims to identify priority actions to improve the overall
satisfaction of services provided in the military hospitals.
This chapter is scientifically and socially justified. This study presents a scientific
interest by addressing a topic of great academic interest in a context still little
explored, without results found in indexed scientific databases. Socially, it is also
important because the results aim to help health managers to improve the health
services provided to users.
This chapter is structured as follows: Section 2 presents the background of this
study, Section 3 presents the methodology applied to achieve the proposed objec-
tives, Section 4 presents this chapter’s results, and Section 5 presents this study’s
conclusions.

2 Background

Since the 1990s, Latin American countries have begun to undertake reforms to
strengthen their health-care systems, aiming at universal citizen coverage through
tax financing (Atun et al. 2015).
Thus, in 1990, the Unified Health System in Brazil (SUS laws 8,080/90 and
8,142/90) was regulated, which comprises a complex chain of service providers and
buyers divided into three subsectors: the public sector, divided into federal, state, and
municipal levels, encompassing the Unified Health System and military health-care
services; the private subsector, where services are provided using public or private
resources; and the supplementary health-care subsystem, which encompasses the
different types of private health-care plans and insurance policies, including some
tax incentives (Coutinho 2017).
Thus, although access to health is a fundamental right, and public service should
guarantee universality and equality to all individuals (Brazilian Constitution 1988),
Patient Satisfaction of Brazilian Military Health-Care System:. . . 477

the services provided do not always meet the expectations and actual needs of the
population.
Mendes et al. (2018) explained that this quality-of-service assessment aims to
identify, measure, and analyze the results that will serve as a basis for decision-
makers seeking to promote the improvement in services and the institution’s
objective.
Patient satisfaction is a critical indicator in assessing health-care service quality.
The results from these studies generally allow health-care professionals to identify
which factors ought to be improved, other than allowing managers to understand the
patient’s actual needs, enabling them to make informed investment decisions to
deliver better quality services (Batbaatar et al. 2017).
Vaitsman and Andrade (2005) warned that measuring satisfaction and using their
health-care indicators are helpful to integrate clients in the process of improving
services, as well as a possibility to transform care into something more humane.
However, measuring the quality of services is not always straightforward because
they are characterized by their intangibility and heterogeneity, making it difficult to
measure consumer satisfaction. Mendes et al. (2018) explained that the subjectivity
of satisfaction surveys is often criticized because clients lack concrete indicators to
evaluate the services they are receiving.
Focusing on service quality measurement, many studies were carried out in the
1980s and 1990s, giving rise to some instruments to assist in service quality
measurement (Pérez et al. 2007).
One of the scales that achieved more excellent usability in research was the result
of the study by Parasuraman et al. (1988), where the authors present a 22-item tool
divided into five underlying dimensions (tangibility, reliability, responsiveness,
assurance, and empathy) to assess customer perceptions of service quality in service
providers (SERVQUAL) and to measure satisfaction firmly.
Although it emerged in a nonhospital context, SERVQUAL has been adapted to
health-care versions (HEALTHQUAL). Several health-care institutions and/or inter-
national accreditation and certification systems have also introduced different quality
measurement items, such as the Joint Commission International (JCI), the leading
health-care services accreditation agency, Accreditation Canada International (ACI),
and the Australian Council for Healthcare Standards International (ACHSI), coming
in different models (Lee 2017). Despite the vast literature on this subject, results are
often contradictory or do not apply to certain organizations due to specific issues,
such as demographic aspects and others (Batbaatar et al. 2017).
This flexibility in measuring health-care service satisfaction makes researchers
and organizations adopt different constructs for their respective instruments as an
artifact. According to Dresch et al. (2015), artifacts are the interface between the
internal and external environments and are designed to serve a specific purpose.
Thus, adopting a tool that allows measuring composite constructs is necessary.
According to Henseler (2017), partial least squares structural equation modeling
(PLS-SEM) offers the possibility of modeling composite constructs and presenting
ingredients for the composition of a construct, which may be different for each
researcher and purpose.
478 A. Melo-Mariano et al.

3 Research Methodology

This is an exploratory study with a quantitative approach using structural equations

via variance. The study was conducted in Brazil. The patients of the military
hospitals present in the national territory were included in this study. The military
has 25 hospitals, 4 polyclinics, and 29 medical posts.
The Brazilian Army, through its specialists, created an annual assessment of its
users’ satisfaction, organized into five composite variables: medical services, service
tangibility, accessory services, dental services, and overall satisfaction.
The study sample comprised 2,510 respondents from 54 different units, including
all states of Brazil and its Federal District. The data refer to the period from 2015 to
2018, characterizing the study as longitudinal.
Three meetings were held with specialists and unit participants to compose the
construction of the model according to the view of the military hospitals.
The medical and dental services were defined and represented throughout the
service process, from the appointment to the waiting and attendance. The tangibility
of services concerns humanization on arrival by security/support/reception staff and
information, hospitality, food at the hospital, and the possibility of referral to another
hospital (usually done in person).
In ancillary services are laboratory, radiology, and pharmacy services and overall
satisfaction with the quality of emergency care and general care (Table 1).
As it is a composite construct as formative, structural equations appear as the
most appropriate solution since it fits the premises of application (Hair et al. 2022) as
experimental models, where the theory is not consolidated, besides presenting metric
scale and a complex model, composed and formative.
After selecting the relevant categories for the study, the data were cleansed by
eliminating data noise, such as typing errors. The respondents’ answers were then
turned into numerical data: fully satisfied = 5; satisfied = 4; unobservable or
indifferent = 3; dissatisfied = 2; and very dissatisfied = 1. Later, they were stored
in an Excel worksheet, a compatible format with the SmartPLS Version 3 software
(Ringle et al. 2015).
Among the data characteristics, we can highlight the predominant age range of
the study between 18 and 25 years, with 40% of the total respondents between
30 and 49 (28%), 18–29 (23%), more than 50–59 (9%), and more than 60 (1%). The
data were collected by the unit responsible for measuring patient satisfaction. About
57% of the interviews were in-person at the service unit, 25.5% over the Internet, and
17.7% over the phone. Regarding the gender of respondents, about 93% are male
and about 7% female, a proportion considered representative of the Brazilian
military population.
Patient Satisfaction of Brazilian Military Health-Care System:. . . 479

Table 1 Data collection instrument

Variable Question description ID
Medical Medical appointments process (ﬁrst appointment and/or return) Servm1
services Waiting time from appointment to attendance Servm2
Quality during the appointment with health-care professionals (care, Servm3
attention, promptness, and clarity in the information/service and reso-
lution of problems)
Service Humanization of the gateway reception/information; Tang1
tangibility Hospitalization (hospitality, food, medical, and nursing care); Tang2
(Referral to Civil Health Organization/Autonomous Health Profes- Tang3
sional when unable to make an appointment at Military Health
Organization)
Accessory (Care in Clinical Analysis Laboratory [waiting time for collection, the Serva1
services time between the examinations and the delivery of results and quality
reports])
(Radiology [waiting time for the appointment, waiting time for exam- Serva2
inations, the time between examinations and results delivery, and
quality of reports])
(Attendance at the outpatient pharmacy [available medications, pro- Serva3
fessional service, and billing])
Dental (Quality in consultation with a dentist [care, attention, promptness, and Sodon1
services clarity of explanations and resolubility])
(Process of scheduling dental appointments) Sodon2
(Waiting time from appointment to dental appointment) Sodon3
Global (Emergency and urgent care) Satg2
satisfaction (Quality of care to military health care, in general) Satg1

4 Model and Hypotheses

The hypotheses and model of this research were formulated from the meetings held
and the collection instrument used by the Military Health Organizations in Brazil
(Table 2).

Table 2 Hypotheses
Hypotheses Relationship
H1 Medical services → Global satisfaction
H2 Accessory services → Global satisfaction
H3 Accessory services → Medical services
H4 Accessory services → Dental services
H5 Dental services → Global satisfaction
H6 Service tangibility → Global satisfaction
H7 Service tangibility → Medical services
H8 Medical services → Global satisfaction
480 A. Melo-Mariano et al.

Fig. 1 Proposed structural model

From the hypothesis presented and the relationships established by the latent
variables, the structural model in this chapter is proposed, as shown in Fig. 1.

5 Results and Discussions

5.1 Valuation of the Measurement Model

The valuation of the measurement model can change depending on the type of the
model. For the formative models, two analyses are performed, the valuation of
multicollinearity and the valuation of the relevance and meaning of the weights.

5.2 Multicollinearity Test

Multicollinearity can be a possibility in tests involving regressions. However, some

tests enable the possibility to notice multicollinearity; the most widely used is called
the variance inﬂation factor (VIF) (Diamantopoulos 2011). This test aims to identify
indicators measuring the same object, causing instability and possible errors in the
study results.
Patient Satisfaction of Brazilian Military Health-Care System:. . . 481

The SmartPLS Version 3 program offers the calculation of two VIFs, as it is the
valuation of the formative measurement model; the external VIF will be used,
leaving the internal VIF for the valuation of the structural model later.
Although some studies in the 1990s point that the maximum value for the VIF
index in a model with no multicollinearity is 10, recent studies propose that the
maximum acceptable limit for the VIF index is 3.3; the lower the VIF index, the
more robust the model (Ringle and Sarstedt 2016; Hair et al. 2022). Aiming to
evaluate whether the model’s variables present multicollinearity, the VIF indexes
were measured.
Initially, the indicators Sodon2 (5.920) and Sodon3 (5.051) presented
multicollinearity as they presented VIF values above the minimum allowed. So, as
suggested by Hair et al. (2022), the Sondon2 indicator, due to its higher than
permitted VIF value, was excluded from the model, and the test was once again
performed; the result is given in Table 3.
Thus, all VIF values were below 3.3, ensuring no collinearity at the indicator
level.
The second test of the measurement model concerns the valuation and meaning of
the weights performed item by item. Again, a two-tailed bootstrapping was
performed, with 5000 subsamples and 5% confidence margin to calculate the
relevance and significance of the weights (Table 4).
Bootstrapping is a nonparametric procedure that allows testing the statistical
significance of various results of PLS-SEM, such as path coefficients (beta) and
R2. As for the test’s parameters, the Student’s t-distribution was two-tailed (n-1
degrees of freedom) as the direction between hypotheses was not specified. Hair
et al. (2022) proposed that the number of subsamples (n) was 5000, and the
significance level was equal to 0.05. Thus, if the practical value of t (0.05; 4999)
= 1.96 is more significant than 1.96, the hypothesis is supported.

Table 3 Variance inﬂation factor

Indicators VIF
Satg1 1.251
Satg 1.251
Serva1 1.278
Serva2 1.488
Serva3 1.243
Servm1 2.500
Servm2 2.505
Servm3 1.355
Sodon1 2.623
Sodon3 2.238
Tang1 1.273
Tang2 1.178
Tang3 1.310
482 A. Melo-Mariano et al.

Table 4 Valuation and meaning of the weights

Path t-value p-value
Satg1 → Global satisfaction 0.805 40.658 <0.001
Serva1 → Accessory services 0.611 22.672 <0.001
Sodon3 → Dental services 0.537 12.298 <0.001
Sodon1 → Dental services 0.534 12.166 <0.001
Tang1 → Service tangibility 0.533 23.259 <0.001
Servm3 → Medical services 0.520 22.620 <0.001
Tang3 → Service tangibility 0.498 21.511 <0.001
Servm1 → Medical services 0.407 12.508 <0.001
Serva2 → Accessory services 0.344 10.380 <0.001
Serva3 → Accessory services 0.334 11.992 <0.001
Satg2 → Global satisfaction 0.333 12.037 <0.001
Tang2 → Service tangibility 0.267 12.156 <0.001
Servm2 → Medical services 0.261 8.400 <0.001

It can be noticed that the relevant weights have a structural path greater than 0.2
and a t-value greater than 1.96 and will be signiﬁcant with a p-value smaller than
0.05, according to Hair et al. (2019). It can be perceived that all items are relevant
and signiﬁcant for the structural model. Satg1, Serva1, Sodon3, Sodon1, Tang1, and
Servm3 are being the most important.
Once the measurement model has been valued and approved, the next step is the
valuation of the structure model.

5.3 Structural Model Valuation

The model’s evaluation consists of identifying the proposed model’s explanation

rate. In order to do so, both the coefficients of determination (R2) and beta index (β)
were used, as suggested by Hair et al. (2019).
The variable R2 represents the explanatory power from 0 to 1 and is explained by
its explanatory variables. The higher the value of R2, the more significant the model’s
capacity to explain the said variable. As suggested by Falk and Miller (1992) and
Hair et al. (2022), the values of the coefficient of determination should be high
enough so that the model attains minimal explanatory power.
Meanwhile, the beta index (β) values are expressed in degrees and reflect each
construct’s specific contribution to the coefficient of determination (R2). Chin (1998)
proposed that in order to be deemed significant, values ought to reach a value of
|β| ≥ 0.2, ideally being above 0.3.
Figure 2 presents the final form of the model in which R2 is represented by the
values within the circles. In contrast, if the values bind the variables, adjusted R2 was
used.
Patient Satisfaction of Brazilian Military Health-Care System:. . . 483

Servm1
0.407
Servm2 0.261 0.470
0.520
Servm3
Medical Services

0.166 0.376 0.568

Serva1 Satg1 Tang1

0.611 0.805 0.533
Serva2 0.344 0.139 0.560 0.262 0.267 Tang2
0.334 0.333 0.498
Serva3 Satg2 Tang3
Accessory Services Global Satisfaction Service Tangibility

0.372 0.118 0.205

Sodon1
0.534
0.277
0.537
Sodon3
Dental Services

Fig. 2 Calculated structural model

The defined variable medical service was explained by 47%, global satisfaction
was explained by 56%, and dental services were explained by 27.7%. Overall
satisfaction was explained by 56%. The variables that most influenced were medical
services and services tangibility. When it is a formative model, the constructs are
represented by the leading indicators that form it.
In this case, medical services are expressed by Servm3—Medical appointments
process (first appointment and/or return) and Servm1—Quality appointment during
an appointment with health-care professionals (care, attention, promptness, and
clarity in the information/service and resolution of problems).
The medical service, in turn, was explained by 47%. The variable that most
influence is service tangibility, represented by its most significant indicators
Tang1—humanization of the gateway reception/information and Tang 3—referral
to Civil Health Organization/Autonomous Health Professional. When unable to
make an appointment at Military Health Organization.
Finally, dental services are explained by 27.7%, and accessories services are the
most influential variable. They are represented by the Serva1 indicator referring to
the time of performance/waiting for examinations necessary for the completion
of care.
Thus, explaining the variables that most influence overall satisfaction and medical
and dental services, the significance test via bootstrapping will be performed with
5000 subsamples and 5% confidence interval in a two-tailed test. All values of
t reached higher than 1.96, confirming that the analysis supports the raised hypoth-
eses. Table 5 presents all the hypotheses and corresponding beta (β), values of the
Student’s t, and p-value indexes.
According to Hair et al. (2019), to test the hypothesis, the Student’s t-value must
be greater than 1.96, with a p-value less than 0.05. Thus, as given in Table 5, all the
hypotheses are supported for the present study as they all meet the required param-
eters established by the literature.
484 A. Melo-Mariano et al.

Table 5 Hypotheses
Hypotheses Beta %β t-value p-value
H1—Medical services → Global satisfaction 0.376 25.49 13.56 0.000
H2—Accessory services → Global satisfaction 0.139 7.76 5.89 0.000
H3—Accessory services → Medical services 0.166 8.75 6.89 0.000
H4—Accessory services → Dental services 0.372 18.76 14.31 0.000
H5—Dental services → Global satisfaction 0.119 5.60 5.89 0.000
H6—Services tangibility → Global satisfaction 0.262 17.16 10.08 0.000
H7—Services tangibility → Medical services 0.568 38.23 27.22 0.000
H8—Services tangibility → Dental services 0.205 9.06 7.81 0.000

Additionally, the beta decomposition was performed to know the percentage of

inﬂuence of each of the variables in a speciﬁc way. Thus, global satisfaction is
explained by 56%. The most critical variable was medical service (25.49%),
followed by service tangibility, contributing 17.16%. The medical services are
explained by 38.23% by service tangibility: Finally, the dental services by 18.76%
by accessory services.

5.4 Practical Implications

The SmartPLS Version 3 has a tool that assists in formulating practical implications.
The importance–performance map analysis (IPMA) allows the researcher to prior-
itize activities that require performance improvements. It presents by the priority of
importance versus performance of the variables that must be worked on ﬁrst helping
in business decisions (Fig. 3). Ringle and Sarstedt (2016) proposed that the variables

Importance-Performance Map
100

Global Satisfaction 50

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
Total Effects
Accessory Services Dental Services Medical Services Service Tangibility

Fig. 3 Importance–performance map analysis

Patient Satisfaction of Brazilian Military Health-Care System:. . . 485

are treated in the right to the left quadrant and then the lower to the upper quadrants,
ensuring the most effective intervention possible.
According to the IPMA matrix in Fig. 3, the variable services tangibility emerges
as the essential variable toward reaching the client’s global satisfaction, followed by
medical services.
The factor that best conforms to the tangibility service is Tang1—humanization
of the gateway reception/information. According to Carvalho and dos Santos (2018),
the practice of welcoming is the time to stagger a relationship of trust with the client
requiring qualified care on the patient’s arrival and generating a connection that
allows him to share his fears and expectations, increasing the satisfaction.
Boeger (2005) explained that having different types of receptions for specific care
linked to a central screening reception is a good strategy. Thus, it is necessary to have
a reception team always qualified to establish a welcome for the patient at this time
so sensitive to it.
The Tang3 (ease to find a referral to Civil Health Organization/Autonomous
Health Professional, when unable to make an appointment at Military Health
Organization) indicator is a matter of care and bureaucracy, which should be
circumvented through process modeling with the possibility of removing activities
that do not add value and the possibility of using an information system in order to
automate the process (Seraphim et al. 2010).
The third item Tang2—hospitalization (hospitality, food, medical, and nursing
care) will run in the new concept of the hospital, such as hospitality. Guimarães
(2002) explained that an environment of good services is an adequate reception
expanded for humanization and increases customer and companion satisfaction
(when is the case). Thus, this topic will be the one with the first assuming the need
for a hotel behavior in hospitalization formed by everything that happens with the
most tangible hotel experience such as daily food monitoring and the storage of
the room.
It is suggested that the “humanization of the gateway reception/information”
(Tang1) has presented greater importance because the health-care system in Brazil
is free; once hospitalized the patient knows he will be well treated, but the long lines
and the initial difficulty are always preponderant factors.
The indicators that form medical services are important: Servm3—quality during
appointments with health-care professionals; Servm1—medical appointments pro-
cess (first appointment and/or return); and Servm2—waiting time from the appoint-
ment to attendance.
The quality of medical care is this variable’s first item to be worked on. In a study
by Iliescu and Carauleanu (2014), 320 opinions were collected, mapping the char-
acteristics to be explored in a physician-considered quality. The results showed a
division between technical and affective qualities.
The best techniques represented are professional competence, attention, patience,
and consistency factors in the SERVQUAL scale. In addition, the affective charac-
teristics appear to be love for people’s modesty, honesty, and ethics. Thus, creating
facilities for learning these characteristics through courses and reflections on the
486 A. Melo-Mariano et al.

positive results of the humanization of care. This also helps for the immediate
improvement of patients.
In Servm1 and Servm2, the process is the same and is directly related. The
construction of medical consultations should comply with a composition of ﬁrst
(longer) consultations and return consultations (less time-consuming), favoring a
less patient waiting time. The composition of the hospital as a hotel also collaborates
with a minor waiting sensation.
From this perspective, the steps can be separated into three moments: in the short
term with the factors related to the service tangibility; in the medium term regarding
the issues of transformation of the hospital environment to a hotel environment
together with the training of service provision offered by the doctor; and ﬁnally in
the long term supporting the changes related to the affective characteristics well
valued by the patients.
It can be perceived that most of the actions to be performed are to humanize
patient care, making them as comfortable as possible.

6 Conclusion

This chapter’s objective was to identify priority actions to improve the overall
satisfaction of services provided in the military hospitals. The results obtained by
the IPMA analysis present the service’s tangibility highlighting services related to
the humanization of gateway reception and the ease of finding an external referral.
According to the results expressed in this chapter, these should be the services that
are prioritized the most. Medical services, consultation quality, consultation process,
and waiting time were the most important factors according to the IPMA matrix.
The relationships between the constructs presented in the data collection instru-
ment were modeled as a secondary goal of this chapter to identify each explanatory
variable’s influence on the model’s dependent variables. The results in Table 5 point
out that services tangibility is the variable that influences the intervener variable
medical services the most, which is the variable that influences global satisfaction
the most. The proposed model (Fig. 2) shows the evidence of its ability to explain a
significant portion of global satisfaction (56%) of military health-care services to
patients besides explaining medical services in 47% and dental services in 27.7%.
Regarding the reliability of the findings of this chapter, the model was validated
for the VIF index and valued by the coefficient of determination (R2) and beta index
(β). Likewise, all the hypotheses were tested and validated by the Student’s t-test and
p-value.
This chapter’s limitations are the lack of similar studies in the military context to
allow the results presented in this chapter to be compared with those within the
literature. However, the lack of similar studies reiterates the relevance of the present
study, which features comprehensive and pioneering research on overall patient
satisfaction in the Brazilian military health-care sector.
Patient Satisfaction of Brazilian Military Health-Care System:. . . 487

Besides contributing to filling the identified literature gap, this chapter aims to aid
managers in prioritizing investments destined to improve health-care services’
quality to show which services provide the most value as perceived by the end
customer. This allows efforts to be concentrated more assertively in search of better
health outcomes for patients and better quality of life for the community.
In future works, the proposed structural model will be applied to specific military
health units to compare the results.

References

Atun R, de Andrade LOM, Almeida G, Cotlear D, Dmytraczenko T, Frenz P et al (2015) Health-

system reform and universal health coverage in Latin America. Lancet 385(9974):1230–1247
Batbaatar E, Dorjdagva J, Luvsannyam A, Savino MM, Amenta P (2017) Determinants of patient
Satisfaction: a systematic review. Perspect Public Health 137(2):89–101
Boeger MA (2005) Gestão em Hotelaria Hospitalar, 2nd edn. Atlas, São Paulo
Brazilian Constitution (1988) Constituição da república federativa do Brasil. Senado Federal,
Centro Gráfico, Brasília
Carvalho SS, dos Santos JIL (2018) Contribuições do setor de acolhimento com classificação de
risco às gestantes. Revista Brasileira de Pesquisa em Saúde/Brazilian Journal of Health Research
20(3):53–59
Chin WW (1998) The partial least squares approach to structural equation modeling. Mod Methods
Bus Res 295(2):295–336
Coutinho GTB (2017) Qualidade nos serviços de saúde: Um estudo exploratório no Distrito
Federal. Tese (Graduação em Engenharia de Produção), Universidade de Brasília
Diamantopoulos A (2011) Incorporating formative measures into covariance-based structural
equation models. MIS Q 35:335–358
Dresch A, Lacerda DP, Júnior JAVA (2015) Design science research: método de pesquisa para
avanço da ciência e tecnologia. Bookman, Brasil
Falk RF, Miller NB (1992) A primer for soft modeling. University of Akron Press, U.S.
Grenier C, de Jésus A, Farsi F, Marx G, Brédart A, Peixoto O et al (2010) L’annonce de la maladie:
satisfaction des patients et démarche qualité dans les Centres de lutte contre le cancer. Bull
Cancer 97(10):1163–1172
Guimarães ER (2002) Business management in health – hospitality, Belo Horizonte. [s.n.] Apostille
of the Qualification Course in Health Management
Haddad SM, Souza RT, Cecatti JG (2019) Mobile technology in health (mHealth) and antenatal
care–searching for apps and available solutions: a systematic review. Int J Med Inform 127:1–8
Hair JF, Hult GT, Ringle CM, Sarstedt M, Castillo-Apraiz J, Cepeda Carrion G, Roldán JL (2019)
Manual de partial least squares structural equation modeling (PLS-SEM), 2nd edn.
OmniaScience, Terrassa, Spain
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2022) A primer on partial least squares structural
equation modeling (PLS-SEM), 3rd edn. Sage, Thousand Oaks
Henseler J (2017) Bridging design and behavioral research with variance-based structural equation
modeling. J Advert 46(1):178–192
Iliescu ML, Carauleanu A (2014) The portrait of a good doctor: conclusions from a patients and
medical students survey. Revista de Cercetare si Interventie Sociala 47:261–271
Lee D (2017) HEALTHQUAL: a multi-item scale for assessing healthcare service quality. Serv Bus
11(3):491–516
488 A. Melo-Mariano et al.

Mendes IAC, Trevizan AM, de Godoy S, Nogueira PC, Ventura CAA, Bis Furlan CE (2018)
Expectations and perceptions of clients concerning the quality of care provided at a Brazilian
hospital facility. Appl Nurs Res 39:211–216
Pal D, Funilkul S, Charoenkitkarn N, Kanthamanon P (2018) Internet-of-things and smart homes
for elderly healthcare: an end user perspective. IEEE Access 6:10483–10496
Parasuraman A, Zeithaml VA, Berry LL (1988) Servqual: a multiple-item scale for measuring
consumer perceptions of service quality. J Retail 64(1):12–40
Pérez SM, Abad CGJ, Carrillo MMG, Fernández SR (2007) Effects of service quality dimensions
on behavioural purchase intentions: a study in public-sector transport. Manag Serv Qual 17(2):
134–151
Ringle CM, Sarstedt M (2016) Gain more insight from your PLS-SEM results: the importance-
performance map analysis. Ind Manag Data Syst 116(9):1865–1886
Ringle CM, Wende S, Becker J (2015) SmartPLS 3 [computer software]. SmartPLS GmbH. www.
smartpls.com. Accessed 15 May 2022
Seraphim EC, Silva ÍBD, Agostinho OL (2010) Lean Ofﬁce em organizações militares de saúde:
estudo de caso do Posto Médico da Guarnição Militar de Campinas. Gestão & Produção 17(2):
389–405
Silva BTD, Lima IMSO (2019) 15a Conferência Nacional de Saúde: um estudo de caso. Saúde e
Sociedade 28(3):97–114
Vaitsman J, Andrade GRBD (2005) Satisfação e responsividade: formas de medir a qualidade e a
humanização da assistência à saúde. Ciência & Saúde Coletiva 10:599–613
Part VIII
Other Fields of Business Research
and Economics
FinTech Loan Continuance Intention: How
Far Can Self-Efﬁcacies Go?

Samuel Danilola, Adewumi Odeniran, and Adewumi Otonne

1 Introduction

Recently, the number of firms that facilitate financial activities using technology has
risen considerably, causing the financial industry to experience significant disruption
(PWC 2016). For instance, a PWC global survey of customers in 2016 found that
80% of FinTech customers believe FinTech firms are more customer-centric,
whereas only 53% of bank customers believe banks are customer-centric. This
customer centricity has a far-reaching effect as it depletes traditional banks’ market
share. FinTech makes financial services, especially banking, more streamlined and
accessible as users are now more easily managed via technology, which enhances
service delivery. This innovation promotes access to financial services, reduces
transaction costs, grows users’ convenience and efficiency, and ultimately enhances
the ability to manage spending and budgeting (Ryu 2018).
FinTech creates a diverse group of financial services for customers, including
payment technology, crowdfunding, wealth management, and lending (Shiau et al.
2020). More specifically, FinTech lenders are a critical part of the finance industry.
Their vital role is offering a loan request and delivery process that does not require
physical presence—everything starts and ends online. The hurdle-free and enhanced
processes of FinTech lenders create further incentives on the demand side. Busi-
nesses and individuals find FinTech firms the go-to lenders to escape bottlenecks,
such as paper documentation and need for collateral, usually experienced with many
African banks (Hope 2019). As this demand continues unabated, FinTech’s

S. Danilola (✉) · A. Odeniran · A. Otonne

Department of Economics and Finance, Lagos Business School, Pan-Atlantic University, Ajah,
Lagos, Nigeria
e-mail: [email protected]

relevance for accessing personal loans and other financial services cannot be
understated.
With the growing investment in FinTech loan services, their survival depends on
their ability to service users effectively and ensure their continuous usage. Active
subscribers, market share (especially in comparison with formal lenders), and
revenue of FinTech firms hinge on the sum of the number of their initial adopters
and continued subscribers. Ensuring continued usage of FinTech loans is more
critical to the success of information systems (IS) than their initial adoption (Zhou
et al. 2018). With the growth in the FinTech space, the convenience of digital
technology, and increasing competition, gaining customer loyalty and commitment
has become more difficult because of the diverse options available at a low cost
(Wang et al. 2019). Retaining customers is crucial for FinTech loan service providers
to sustain growth. It, therefore, becomes necessary to advance our knowledge of
customers’ continuance intention in FinTech, especially for loan services.
Given the importance of using FinTech, this chapter examines users’ continuance
intentions in securing FinTech loans. It achieves this by extending the expectation
confirmation theory of information system continuance (ECT-IS), as done by Shiau
et al. (2020). The study by Shiau et al. (2020) extends the literature by focusing on
the ECT-IS model, paying attention to the antecedents of the ECT-IS constructs. The
study mentions that inherent motivation of users is also necessary for determining
continuance intention (Nascimento et al. 2018). The study highlighted the relevance
of self-efficacy theory in explaining the intention of customers to continue using
FinTech for wealth management. However, the study is limited because it cannot be
applied to all other financial applications like loans (Shiau et al. 2020). For instance,
the inherent self-beliefs needed to handle loan processing and wealth management
are somewhat different. Highlighting this difference is critical for FinTech firms as it
will be useful when designing products that meet the needs of each customer
segment and ensure their retention. This chapter hopes to address that difference
by showing the effect of the self-efficacy theory on customer reuse of FinTech loans.
Furthermore, the growing FinTech ecosystem in Africa calls for an inquiry into
the chances of its continued existence. Some studies have paid attention to contin-
uance intention in an African context. They include Franque et al. (2021), Akinwale
and Kyari (2022), and Warsame and Ireri (2021). Although these studies examined
FinTech use in Africa, none used a combination of self-efficacy and ECT-IS to
investigate users’ continuance intention for FinTech loan services. Therefore, it is
difficult to tell how the reuse of FinTech platforms to obtain loans can be enhanced.
As a result, this leaves a gap that this chapter addresses by integrating ECT-IS with
self-efficacy theory to show the factors affecting the reuse of FinTech for obtaining
loans in an African context.
The rest of the chapter is organized as follows. Section 2 reveals the theoretical
background based on the literature. Section 3 reviews the research model and
hypotheses. Section 4 describes the research methodology, and Section 5 details
the data analysis and results. Section 6 discusses the chapter’s results, and Section 7
concludes the chapter, showing its implications, limitations, and avenues for future
research.
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 493

2 Theoretical Background

2.1 Expectation Conﬁrmation Theory of IS Continuance

Expectation confirmation theory has been used widely to predict the repurchase of
different products and services (Bhattacherjee 2001). However, the theory was
limited in integrating possible consumer expectation changes after using the product.
This post-adoption experience could shape users’ satisfaction with the product, but
the model failed to capture it adequately. So Bhattacherjee (2001) stretched the ECT
to include IS continuance. According to ECT-IS, a user’s IS continuance intention
depends on his/her satisfaction level, the level to which his/her expectation is
confirmed, and the perceived usefulness of the IS. ECT-IS has been used in different
contexts since it was proposed, and the constructs and findings from those studies
vary with contexts and the studies’ objectives (Shiau et al. 2020).

2.2 Self-Efﬁcacy

Self-efﬁcacy depicts an individual’s subjective judgment of their capacity to do a

particular task required to realize the desired outcome (Bandura 1991). Self-efficacy
is of utmost importance because it tells us a great deal about how far people are
willing to go to achieve a particular goal despite the prevailing obstacles. Previous
research has incorporated self-efficacy in different contexts (Shiau et al. 2020; Choi
2018; Kim et al. 2016). However, the literature shows that domain-specific self-
efficacy predicts a particular phenomenon more than general self-efficacy (Cassar
and Friedman 2009). Based on the specific elements of accessing FinTech loans,
users measure their ability to understand loan terms and use technology
(smartphone) to acquire the loan. Before deciding on the loan, the customer might
have to choose the appropriate loan terms from competing platforms to meet their
needs based on their understanding of the loan, hence measuring financial self-
efficacy. Complementarily, the customer presumably assesses their self-efficacy in
using a technological platform—a smartphone—to complete the task of securing the
loan. So, it is critical to understand self-efficacy in accessing FinTech loans, specif-
ically financial and technological self-efficacy.

3 Literature Review and Hypotheses Development

FinTech arose from digitalization, which is the use of technology to improve

corporate processes. It provides ﬁnancial services using a technology-enabled or
digital process (Wang et al. 2019). As the intent to avail loans using FinTech
involves an amalgam of technology and loan knowledge, it is necessary to include
494 S. Danilola et al.

the internal motivations of the users. To better show FinTech continuance intentions
for loan services, this chapter combines the ECT-IS and self-efficacy theories to
reveal the impact of self-efficacy on continuance intention. In this chapter, self-
efficacy is divided into technological and financial self-efficacies.

3.1 Financial Self-Efﬁcacy and Technological Self-Efﬁcacy

When individuals remark that they possess high financial self-efficacy, they are
usually capable of managing any financial tasks (Farrell et al. 2016; Asebedo and
Payne 2019). More specifically, the financial flexibility of a user usually increases
with upticks in their financial self-efficacy. This is because they most likely make
decisions after weighing the potential benefits of such actions and are usually more
optimistic about the outcomes of those decisions. As Farrell et al. (2016) showed,
this attitude potentially results in more favorable financial outcomes. For loan
services, people with high financial self-efficacy could be more eager to take loans
using FinTech because they can handle the loan acquisition process and manage
other challenges that might arise when using FinTech in loan-related activities,
thereby perceiving it as useful. This led to the following hypothesis (H):
H1: Financial self-efficacy positively affects users’ perceived usefulness of FinTech.
The status of financial self-efficacy in determining the expectations and perfor-
mance of consumers has been documented (Shiau et al. 2020). This is because
individuals with higher financial self-efficacy put in tremendous effort to attain
their financial goals (Bandura 1986). As these individuals achieve their goals, their
expectations are confirmed. The higher the user’s financial self-efficacy, the more
likely they will exceed their expectations; hence, the positive relationship between
financial self-efficacy and confirmation. Therefore, this chapter proposes the follow-
ing hypothesis:
H2: Financial self-efficacy positively affects users’ confirmation of FinTech use.
Similar to financial self-efficacy, the role of technological self-efficacy on the
perceived usefulness of innovations has been highlighted in several studies (Holden
and Rada 2011; Pan 2020). Ma and Liu (2005) showed that Internet self-efficacy
significantly affects the perceived ease of use, usefulness, and behavioral intentions
to use electronic medical records. However, for mobile banking products, Alalwan
et al. (2016) showed the crucial role of technological self-efficacy in determining
perceived usefulness. Technological self-efficacy measures an individual’s percep-
tions of smartphones as smartphones become advanced. Users can complete loan
purchase using smartphones. Therefore, users with high technological self-efficacy
could find FinTech more useful. Hence, this chapter proposes the following
hypothesis:
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 495

H3: Technological (smartphone) self-efﬁcacy positively affects users’ perceived

usefulness of FinTech.
The literature has also revealed that people with high self-efficacy put in more
effort toward ensuring that they achieve the desired results, confirming their expec-
tations. For example, Shiau et al. (2020) stated that smartphone self-efficacy is a
significant determinant of confirmation when smartphones are used for wealth
management. However, although the nature of the relationship in the case of
FinTech loans is not clear, these beliefs may also lead to a more robust capability
for a user to perform a particular task using their smartphones for accessing loan
services and invariably making their perceived performance exceed their expecta-
tions. As a result, this chapter proposes the following hypothesis:
H4: Technological (smartphone) self-efficacy positively affects users’ confirmation
of FinTech use.

3.2 Conﬁrmation, Perceived Usefulness, and Satisfaction

The ECT-IS model proposes that individuals retain the use of an IS product based on
their level of satisfaction with the product, the extent of confirmation of their
expectations, and their post-usage perceived usefulness. Using this model, several
studies have identified how confirmation significantly predicts perceived usefulness
(Nascimento et al. 2018; Shiau et al. 2020). As the initial expectation of IS use may
vary, users’ initial perceived usefulness may be unstable and hence will be adjusted
by confirmation experience. Thus, users’ confirmation of expectations positively
affects perceived usefulness, enhancing customers’ satisfaction and the intention to
continue usage (Bhattacherjee 2001). When applying ECT-IS continuance to
smartphones for wealth management, Shiau et al. (2020) revealed the effect of
confirmation on perceived usefulness. Yuan et al. (2016) and Susanto et al. (2016)
also showed that confirmation of users’ mobile banking expectations significantly
affects the perceived usefulness of those services. Following these studies, it could
be inferred that perceived usefulness is positively affected by the confirmation of
FinTech use. Hence, this chapter proposes the following hypothesis:
H5: Confirmation positively affects users’ perceived usefulness of FinTech use.
Furthermore, the significance of users’ perceived usefulness and confirmation as
joint determinants of users’ satisfaction has been revealed by Bhattacherjee (2001).
Explaining this relationship for mobile banking services, Yuan et al. (2016) and
Susanto et al. (2016) confirmed a direct path from perceived usefulness to satisfac-
tion using the ECT model. Therefore, this chapter proposes the following
hypothesis:
H6: Perceived usefulness positively affects users’ satisfaction with FinTech use.
496 S. Danilola et al.

It has been established that conﬁrmation is a signiﬁcant determinant of satisfac-

tion (Hossain and Quaddus 2012). The consensus is that confirmation results in
satisfaction provided the initial expectation has at least been confirmed. For FinTech
use, confirmation depicts users’ perception of whether their expectation of FinTech
use equals or exceeds its actual performance. Thus, when the users’ initial expecta-
tions for the loans are confirmed, they find FinTech use satisfactory. As they use the
product continually, they keep updating their expectations. If FinTech continues to
exceed their expectations, their post-adoption expectations are confirmed. Therefore,
this chapter proposes the following hypothesis:
H7: Confirmation positively affects satisfaction with FinTech use.

3.3 FinTech Continuance Intention

In many contexts, the status of the perceived usefulness in explaining behavioral

intentions has been documented (Yuan et al. 2016; Nascimento et al. 2018; Shiau
et al. 2020). Hence, the more valuable users perceive a FinTech product, the more
likely they will continue to use it. Therefore, this chapter proposes the following
hypothesis:
H8: Perceived usefulness positively affects FinTech continuance intention.
According to ECT-IS, users’ continuance intention is influenced by the satisfac-
tion derived from using IS, and this relationship has been confirmed in the FinTech
services setting (Nascimento et al. 2018; Lim et al. 2018; Shiau et al. 2020). For
instance, Warsame and Ireri (2021) confirmed the positive effect of perceived
satisfaction on the reuse of mobile money lending services. Thus, this chapter
hypothesizes the following:
H9: Satisfaction positively affects FinTech continuance intention.

3.4 The Mediating Role of Conﬁrmation

Financial self-efficacy influences financial behavior positively (Farrell et al. 2016). It

also has the most extensive effect on satisfaction by confirming expectations (Joo
and Grable 2004). Users are confident about taking on a financial task once they
perceive themselves as having high financial self-efficacy. Once they perceive their
performance to be high, then the confirmation of their expectation is also increased.
This expectation confirmation is predicted to flow through to the perceived useful-
ness. In the cases of using smartphones to manage wealth, for instance, the mediating
effect of confirmation in the link between financial self-efficacy and perceived
usefulness has been confirmed by Shiau et al. (2020). Therefore, financial self-
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 497

Fig. 1 The research model (adapted from Shiau et al. 2020). Notes: H10. Confirmation mediates
the effect of financial self-efficacy on perceived usefulness under the full model. H11. Confirmation
mediates the effect of technological self-efficacy on perceived usefulness under the full model

efficacy can affect confirmation and the perceived usefulness of FinTech. Thus, this
chapter proposes the following hypothesis:
H10: Confirmation mediates the association between financial self-efficacy and the
perceived usefulness of FinTech under the entire model.
On the other hand, technology self-efficacy emphasizes how individuals perceive
their effectiveness at using smartphones. It depicts individuals’ ability to download
apps, install them, and accomplish other smartphone tasks (Alalwan et al. 2016). As
stated earlier, people with greater self-efficacy are more readily engaged in functions
related to that efficacy. Therefore, this belief encourages positive behavior, propel-
ling them to make the necessary efforts to confirm their expectations. Confirmation,
therefore, gives details into an expanded perception of usefulness. Shiau et al. (2020)
reveal that when using FinTech for wealth management, confirmation partially
mediates the relationship between technology self-efficacy and perceived usefulness.
Therefore, this chapter proposes the following hypothesis:
H11: Confirmation mediates the impact of technological self-efficacy on the per-
ceived usefulness of FinTech under the entire model.
The hypotheses define the research model following the theories in the FinTech
loan context. The research model to be tested is presented in Fig. 1.

4 Research Methodology

4.1 Instrument and Measurement Items

This study combines self-efﬁcacy and ECT-IS theories to explain and predict IS
continuance intention. It uses scales whose operational deﬁnitions were adapted
498 S. Danilola et al.

from the literature (Shiau et al. 2020). The constructs were modeled as reﬂective,
following Jarvis et al. (2003) and Shiau et al. (2020). The chapter also used a seven-
point Likert scale ranging from “strongly disagree” (1) to “strongly agree” (7) to
measure each item of the construct. The operational deﬁnitions and the scale items of
each construct are provided in the Appendix.

4.2 Sample and Data Collection

The respondents were individuals who had accessed loans using FinTech apps. The
content validity of the questionnaire was confirmed using a pilot test that gauged the
appropriateness, ease of understanding, and use and task relevance of the questions.
The pilot was conducted on respondents with a history of using FinTech loans. They
contributed to some of the questions, and a few were modified to ensure content
validity. The data used for the final analysis were elicited using an online survey. The
study adopted this method based on its cost-saving attribute, rapid response rate, and
widespread coverage.
The study employed partial least squares structural equation modeling
(PLS-SEM). PLS-SEM was favored over the covariance-based structural equation
modeling (CB-SEM) for the following reasons. First, the data collected using Likert
scale are usually skewed; PLS-SEM is the best tool for checking causation as it
relaxes the normality assumption of the variables (Hair et al. 2019; Khan et al. 2019;
Shiau et al. 2019), ensuring that the results are not affected by the data type.
Secondly, the model is complex and has many indicators and constructs (Ringle
et al. 2020; Gefen et al. 2011). Finally, the study adopted a predictive scope, and
PLS-SEM effectively maximizes the variances in the predictands to be explained by
the predictors (Ringle et al. 2020; Hair et al. 2016). Due to the reasons above,
PLS-SEM is the most appropriate model for the analysis.

5 Data Analysis and Results

5.1 Descriptive Statistics

The main survey was conducted from December 2021 to March 2022, receiving
273 responses in Nigeria. The data were cleaned, and invalid (respondents who had
not used FinTech loans) responses were removed; only 129 valid responses were
retained for data analysis. The sample size was sufﬁcient for the analysis after
following the conservative 10-time rules. The rule states that the sample size should
be 10 times the total number of inner models in the study (Schreiber et al. 2006). In
this study, the inner models are seven, so the minimum recommended sample size is
70 respondents, and we used 129. Of the valid responses, 63.57% were men. Most
respondents were between the ages of 25 and 34 years, inclusive (73.64%), and most
FinTech Loan Continuance Intention: How Far Can Self-Efﬁcacies Go? 499

Table 1 Descriptive statistics of respondents’ characteristics

Demographic Characteristics Frequency Percent (%)
Gender Male 82 63.57
Female 47 36.43
Age (years) 18–24 20 15.50
25–34 95 73.64
35–44 12 9.30
45–54 2 1.55
Education Secondary 6 4.65
Bachelor/higher national diploma 90 69.77
Postgraduate 32 24.81
Professional certiﬁcation 1 0.78
Employment status Student 10 7.75
Unemployed 4 3.10
Self-employed 28 21.71
Employed (part-time) 12 9.30
Employed (full-time) 75 58.14
Period (years) <1 80 62.02
1–3 44 34.11
4–6 6 3.88

had less than one year of experience taking FinTech loans (62.02%). The results are
shown in Table 1.

5.2 Common Method Bias

When a study collects measures of two or more latent variables from the same
respondent at a point in time using the same method, there could be the common
method bias (CMB), which could negatively affect the research’s validity (MacKen-
zie and Podsakoff 2012). Assessing the possibility of CMB, the chapter used
Harman’s one-factor test (Podsakoff and Organ 1986). We conducted a principal
component analysis of all the indicators for this purpose. The result revealed a ﬁve-
factor structure with no factor accounting for more than 50% of the covariance.
Hence, no general factor was apparent (Mattila and Enz 2002; Shiau et al. 2020). The
combined factors accounted for 75.28% of the total variance; the largest factor was
40.61%. In addition to the Herman one-factor test, an inspection of the variance
inﬂation factors (VIFs) is also a valuable tool to ascertain the likelihood of CMB
(Kock 2015). As all the VIFs resulting from the full collinearity test are lower than
3.3, the data could be considered free of CMB.
500 S. Danilola et al.

5.3 Measurement Model

The study assessed the measurement models’ reliability, convergent validity, and
discriminant validity using factor loadings, composite reliability (CR), and average
variance extracted (AVE) for the convergent validity. The results in Table 2 confirm
that the factor loadings of all the items exceed the recommended 0.6 (Chin et al.
2008). Furthermore, Cronbach’s alpha, rho_A, and CR values, which measure the
degree to which the items measure the latent construct, which exceeded 0.7 (Sarstedt
et al. 2017). However, the AVE, which tells the total variance in the construct
indicators measured by the latent construct, exceeds the recommended floor of 0.5
(Hair et al. 2013).
After verifying the convergence of the model, Fornell and Larcker (1981) and the
heterotrait–monotrait (HTMT) ratio of correlations were used to investigate the level
to which each construct does not identify other variables—the discriminant validity.
As each construct’s correlation with another construct is less than the square root of
the construct’s AVE, as shown in Table 3, the model’s discriminant validity is
confirmed (Fornell and Larcker 1981).
As a result of criticisms about the adequacy of Fornell and Larcker’s (1981)
criteria, Henseler et al. (2015) proposed an alternative method called HTMT. The

Table 2 Scale properties of the measurement model

Item Cronbach’s Composite
Constructs Items loadings alpha rho_A reliability AVE
Confirmation CON2 0.924 0.921 0.924 0.950 0.864
CON3 0.950
CON4 0.913
FinTech continuance FC1 0.911 0.925 0.931 0.952 0.870
intention FC2 0.938
FC4 0.948
Financial self-efficacy FS1 0.793 0.889 0.992 0.913 0.677
FS2 0.866
FS3 0.793
FS4 0.831
FS5 0.830
Perceived usefulness PU1 0.910 0.924 0.925 0.946 0.815
PU2 0.911
PU3 0.930
PU4 0.859
Satisfaction SAT1 0.966 0.958 0.961 0.973 0.922
SAT2 0.972
SAT3 0.943
Technical self-efficacy TSE1 0.884 0.791 1.028 0.864 0.681
TSE3 0.760
TSE4 0.849
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 501

Table 3 Discriminant validity: Fornell–Larcker criterion

1 2 3 4 5 6
1. Confirmation 0.930
2. Financial self-efficacy 0.342 0.823
3. FinTech continuance intention 0.589 0.176 0.933
4. Perceived usefulness 0.795 0.345 0.665 0.903
5. Satisfaction 0.756 0.293 0.631 0.832 0.960
6. Technical self-efficacy 0.123 0.330 -0.043 0.103 0.038 0.832
Notes: Values on the diagonal (in bold) are the square roots of the AVE, whereas the off-diagonals
are correlations.
AVE average variance extracted

Table 4 Discriminant validity: heterotrait–monotrait (HTMT)

1 2 3 4 5 6
1. Confirmation
2. Financial self-efficacy 0.332
3. FinTech continuance intention 0.638 0.164
4. Perceived usefulness 0.861 0.335 0.716
5. Satisfaction 0.802 0.265 0.668 0.884
6. Technical self-efficacy 0.127 0.363 0.066 0.121 0.038

HTMT criteria recommend values less than 0.90 for structural models with con-
structs (Henseler et al. 2015). In this study, the values ranged from 0.038 to 0.884, as
revealed in Table 4, supporting the constructs’ discriminant validity.

5.4 PLSpredict for Model Assessment

PLSpredict directly provides better predictive power. Thus, it is used in this study
(Shmueli et al. 2016). Using the coefﬁcients of determination (R2), the study showed
how much each explanatory variable would explain the dependent variables. After
testing four competing models (Fig. 2), Models 1 to 4 explain 46.2% of the variation
in the FinTech loan app continuance intention. But to measure the out-of-sample
predictive power (i.e., the capacity of the models to predict the values of new cases
that were outside the estimation process), R2 falls short. Hence, Cepeda Carrion et al.
(2016) and Shmueli et al. (2016) revealed that the models should be estimated on a
training (analysis) sample. Then their predictive performances are assessed based on
other data besides the training data.
The predictive error matrices are also presented. The analysis focused on the
model’s target construct and continuance intention; however, the result also reports
other exogenous constructs’ indicators. As seen in Table 5, ﬁrstly, all the endoge-
nous constructs’ indicators of Models 2 to 4 outperform the most naive linear
regression model (LM) benchmark, as their Q2predict values were above zero. The
502 S. Danilola et al.

Fig. 2 The four competing models

Table 5 PLSpredict assessments among four models
RMSE MAE MAPE Q2predict
M1 M2 M3 M4 M1 M2 M3 M4 M1 M2 M3 M4 M1 M2 M3 M4
CON2 1.815 1.808 1.820 1.817 1.499 1.493 1.501 1.498 63.460 63.252 63.484 63.514 0.042 0.046 0.037 0.046
CON3 1.920 1.916 1.924 1.925 1.618 1.615 1.617 1.618 74.528 74.457 74.482 74.708 0.088 0.092 0.084 0.088
CON4 1.771 1.769 1.776 1.769 1.494 1.494 1.499 1.493 62.641 62.660 62.739 62.661 0.084 0.089 0.081 0.090
FC1 2.050 2.053 2.052 2.047 1.754 1.758 1.762 1.760 94.365 94.508 94.841 94.719 0.017 0.019 0.018 0.019
FC2 2.152 2.151 2.147 2.144 1.885 1.885 1.893 1.890 97.608 97.578 98.010 97.898 0.002 0.004 0.007 0.009
FC4 2.019 2.018 2.015 2.012 1.755 1.757 1.757 1.756 93.065 93.030 93.207 93.156 0.010 0.010 0.011 0.013
PU1 1.730 1.720 1.734 1.726 1.401 1.392 1.410 1.400 52.742 52.384 52.865 52.596 0.098 0.110 0.097 0.103
PU2 1.741 1.731 1.737 1.736 1.418 1.411 1.420 1.415 57.966 57.518 57.978 57.862 0.079 0.089 0.082 0.088
PU3 1.763 1.754 1.757 1.752 1.456 1.449 1.459 1.454 58.838 58.431 58.997 58.886 0.056 0.064 0.063 0.068
PU4 1.835 1.827 1.831 1.824 1.425 1.418 1.429 1.423 61.882 61.687 61.998 61.794 0.046 0.051 0.047 0.055
SAT1 1.803 1.799 1.805 1.804 1.453 1.450 1.460 1.461 71.119 70.953 71.629 71.716 0.090 0.094 0.084 0.089
SAT2 1.772 1.767 1.773 1.772 1.431 1.430 1.440 1.439 68.396 68.256 68.996 69.062 0.069 0.072 0.064 0.069
FinTech Loan Continuance Intention: How Far Can Self-Efﬁcacies Go?

SAT3 1.777 1.767 1.764 1.760 1.465 1.457 1.455 1.453 70.075 69.718 69.918 69.840 -0.001 0.006 0.007 0.014
The bold values signify the model with the least error for each variable
503
504 S. Danilola et al.

Table 6 PLSpredict assessment—PLS versus LM

RMSE MAE MAPE Q2predict
LM PLS LM PLS LM PLS LM PLS
CON2 1.857 1.808 1.516 1.493 63.289 63.252 -0.006 0.046
CON3 1.909 1.916 1.544 1.615 69.876 74.457 0.098 0.092
CON4 1.792 1.769 1.488 1.494 61.376 62.660 0.065 0.089
FC1 2.127 2.053 1.763 1.758 94.996 94.508 -0.054 0.019
FC2 2.227 2.151 1.914 1.885 97.901 97.578 -0.068 0.004
FC4 2.085 2.018 1.775 1.757 94.140 93.030 -0.056 0.010
PU1 1.721 1.720 1.359 1.392 51.615 52.384 0.109 0.110
PU2 1.738 1.731 1.398 1.411 55.616 57.518 0.082 0.089
PU3 1.762 1.754 1.464 1.449 57.989 58.431 0.055 0.064
PU4 1.863 1.827 1.454 1.418 61.272 61.687 0.013 0.051
SAT1 1.803 1.799 1.407 1.450 66.772 70.953 0.090 0.094
SAT2 1.835 1.767 1.430 1.430 68.542 68.256 -0.001 0.072
SAT3 1.771 1.767 1.430 1.457 68.233 69.718 0.001 0.006

test for the prediction errors reveals that Model 2 produces the best predictive power
compared to the other two models. More specifically, most of the indicators of
Model 2 have lower prediction errors measured in terms of root mean square error
(RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE).
Furthermore, most of the Q2predict values for the indicators of Model 2 are above zero
and outperform those of other competing models. Therefore, Model 2 has the largest
prediction power among the four models. In addition, examining the predictive
measures for the mediators reveals similar results. The four models equally explain
confirmation and satisfaction.
Finally, comparing the errors from the PLS model analysis with the naive LM
benchmark, as shown in Table 6, PLS-SEM analysis produced lower prediction
errors for Model 2, as most of the Q2predict values of the constructs were better than
those of the linear model. More specifically, many of Model 2’s RMSE, MAE, and
MAPE values are less than the naive LM of Model 2. Thus, Model 2 effectively
explains users’ continuance intention toward taking loans using FinTech, and the
path is shown in Fig. 3.

5.5 Structural Model

Before the structural model assessment, the study checked Model 2 for collinearity
issues by investigating the variance inflation factor (VIF) values of all the predictor
constructs in the model (Hair et al. 2017). The VIF values are below the
recommended conservative threshold of 3.3 (Diamantopoulos and Siguaw 2006);
hence, collinearity is not at critical levels. The study conducted a bootstrap technique
specifying 5000 subsamples to test the path coefficients of the model. The results of
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 505

Fig. 3 Results of the structural model for Model 2. Notes: **p < 0.05; ***p < 0.001

the structural model analysis for the best model are presented in Fig. 3, revealing the
significance and relationship of each hypothesized relationship and the variance
explained.
The model explains 46.2%, 71.6%, 63.9%, and 11.7% of the variation in the
FinTech continuance intention, satisfaction, perceived usefulness, and confirmation,
respectively. Financial self-efficacy has significant positive effects on confirmation
(β = 0.338, p < 0.001) but a non-significant positive effect on perceived usefulness
(β = 0.083, p > 0.05). However, for technology self-efficacy, it has a non-significant
positive effect on confirmation (β = 0.012, p > 0.05) and no relationship with
perceived usefulness. Confirmation positively affects perceived usefulness
(β = 0.767, p < 0.001). Perceived usefulness has significantly positive effects on
satisfaction (β = 0.627, p < 0.001), and confirmation has very positive effects on
satisfaction (β = 0.257, p < 0.05). Perceived usefulness has significantly positive
effects on FinTech continuance intention (β = 0.452, p < 0.001), and satisfaction
has very positive effects on FinTech continuance intention (β = 0.313, p < 0.05).
Therefore, only seven of the hypotheses were supported.

5.6 Analysis of Mediation

The study proposed H10 and H11, which suggest the mediating role of confirmation
on how financial and technology self-efficacy affects perceived usefulness. The
mediation test proposed by Zhao et al. (2010) was used to test for mediation. As
shown in Table 7, firstly, the indirect effect of financial self-efficacy on perceived
usefulness (a × b) is significant (β = 0.087, t = 2.274). Secondly, the direct impact of
financial self-efficacy on perceived usefulness (c) is not significant (β = 0.083,
t = 1.493). As the indirect effect is significant, but the direct impact is not, t-only
mediation is confirmed. However, the indirect effect of technological self-efficacy
506 S. Danilola et al.

Table 7 Inner VIF values

1 2 3 4 5 6
1. Confirmation 1.132 2.723
2. Financial self-efficacy 1.132 1.132
3. FinTech continuance intention
4. Perceived usefulness 3.240 2.723
5. Satisfaction 3.240
6. Technical self-efficacy 1.132

on perceived usefulness (a × b) is insigniﬁcant (β = 0.009, t = 0.136). Since the

direct result of technological self-efficacy on perceived usefulness does not exist,
there is no effect. The study supports that H10—financial self-efficacy’s impact on
perceived usefulness is indirect and mediated by confirmation and rejects H11.

6 Discussion

The chapter seeks to provide FinTech firms and financial institutions with informa-
tion to help them understand how self-efficacy influences users’ continuance inten-
tion by combining self-efficacy and ECT-IS theories. Like previous studies (Susanto
et al. 2016; Yuan et al. 2016; Shiau et al. 2020; Warsame and Ireri 2021), this chapter
shows that users’ FinTech continuance intentions are significantly affected by their
perceived usefulness and satisfaction. Users are likely to continue using FinTech to
access loans when their loan needs are met; hence, they perceive it as helpful.
Suppose FinTech increases the ease with which they can access loans and improves
their effectiveness in managing their finances, then it is doubtful that they will
discontinue using FinTech for obtaining loans (Table 8).
Similar to perceived usefulness, satisfaction is another major factor crucial to
continuance intentions (Bhattacherjee 2001). The importance of satisfaction is clear
because it adapts to the effect of uncertainties around using FinTech. Although
FinTech is convenient and fast, being relatively new, users can always rely on the
satisfaction they derived from the experience with its use to determine continuance
intention (Shiau et al. 2020). Thus, users consistently use FinTech to access loans,
provided they feel content with their previous use. Moreover, as revealed by
Bhattacherjee (2001), Nascimento et al. (2018), and Shiau et al. (2020), the centrality
of perceived usefulness and satisfaction to continuance intention was consistent,
which implied that users place a premium on the satisfaction derived from using
FinTech for accessing loans as well as the perceived usefulness of FinTech when
making continuance decisions.
Regarding satisfaction, the result revealed the significant positive influences of
users’ perceived usefulness and the confirmation of their expectations on satisfac-
tion. The link between satisfaction and intention has been validated in consumer
behavior research over a wide range of products and services contexts (Bhattacherjee
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 507

Table 8 Structural model assessment for direct and indirect effects

Standard Confidence
Hypotheses β t-value interval Remark
H1 Financial self-efficacy → Per- 0.083 1.493 [-0.022, 0.194] Not
ceived usefulness supported
H2 Financial self-efficacy → 0.338 4.784*** [0.187, 0.467] Supported
Confirmation
H3 Technological self-efficacy → – – No
Perceived usefulness relationship
H4 Technological self-efficacy → 0.012 0.136 [-0.246, 0.127] Not
Confirmation supported
H5 Confirmation → Perceived 0.767 15.939*** [0.665, 0.852] Supported
usefulness
H6 Perceived usefulness → 0.627 6.925*** [0.445, 0.799] Supported
Satisfaction
H7 Confirmation → Satisfaction 0.257 2.510** [0.052, 0.454] Supported
H8 Perceived usefulness → Con- 0.452 3.849*** [0.208, 0.673] Supported
tinuance intention
H9 Satisfaction → Continuance 0.255 2.001** [0.011, 0.512] Supported
intention
Indirect effects
H10 Financial self-efficacy → 0.259 4.476*** [0.144, 0.368] Supported
Confirmation → Perceived
usefulness
H11 Technological self-efficacy → 0.009 0.136 [-0.192, 0.096] Not
Confirmation → Perceived supported
usefulness
Notes: **p < 0.05; ***p < 0.001

2001; Warsame and Ireri 2021), and its revalidation in the FinTech context further
buttresses the robustness of the link. However, by comparing the path coefficients of
confirmation-satisfaction (β = 0.257) and perceived usefulness-satisfaction
(β = 0.627), this study reveals that perceived usefulness has a more significant
influence on satisfaction than confirmation. This result is consistent with Warsame
and Ireri (2021), who found a more significant impact of perceived usefulness of
FinTech mobile money lending apps than confirmation on satisfaction. However, the
result of this study differs from Shiau et al.’s (2020), which showed that confirmation
has a higher effect than perceived usefulness in their research on wealth manage-
ment. Therefore, this chapter shows that users of FinTech apps for loan services and
users for general financial management could differ significantly, and this difference
should be considered. Users of FinTech for loan services give credence to the
usefulness of the FinTech app in determining how satisfied they will be and the
chances of reusing the product.
This chapter shows that confirmation significantly influences perceived useful-
ness, whereas financial self-efficacy has an insignificant positive effect on perceived
usefulness in the model. As self-efficacy depicts people’s self-beliefs about their
508 S. Danilola et al.

abilities, people with higher financial abilities believe they can make financial
decisions for themselves. They are usually more willing to believe that the financial
tools they use will help them address their financial needs. However, in accessing
loans through FinTech apps, the result reveals that users do not perceive the need for
complex financial ability to determine a loan app’s perceived usefulness. Why is
this so? Loan platforms mostly give loans with a specified interest rate and payback
period and leave the borrower with no choice but to accept or decline the loan—no
bargaining is involved; hence, no deep financial knowledge is required to obtain
FinTech loans. The same can also be said of technological self-efficacy, which has
no relationship with perceived usefulness in the model.
Nevertheless, consistent with Shiau et al. (2020) and Nascimento et al. (2018), as
confirmation of expectations increases, the perceived usefulness of the app increases,
and this effect depicts the possibility of users’ perception of FinTech being adjusted
with the magnitude to which their expectations are confirmed. Confirmation in
FinTech is the cognitive belief formed by the amount to which a user’s expectation
is met during the use of FinTech (Bhattacherjee 2001). Users have some expecta-
tions about the FinTech loan app, primarily speedy delivery of loan facilities. When
realized, the customer sees the product as relevant to their needs.
The chapter also finds that users’ confirmation positively relates to their financial
and technological self-efficacy. Although the financial self-efficacy–confirmation
relationship is significant, the technological self-efficacy-confirmation is not signif-
icant. Individuals with higher financial self-efficacy make better financial
decisions—like choosing the best loan service to avail and when—and putting in
more significant efforts to achieve their desired financial goals (Bandura 1986). This
finding is similar to that reported by Farrell et al. (2016) about how financial self-
efficacy leads to positive financial behavior in taking loans. Therefore, as these
individuals attain their goals, their expectations are confirmed.
However, this result about the insignificant positive effect of technological self-
efficacy on confirmation differs from that of Shiau et al. (2020). As taking loans
using a FinTech app is primarily limited to the basic installation of the apps and then
giving necessary permissions, users do not perceive this ability as critical to deter-
mining whether their expectation of securing the loan has been confirmed. Instead,
they pay more attention to actualizing their financial goal from the level of financial
knowledge they have as more relevant to determining the confirmation of their
expectations. This is also reflected in the path coefficients, as financial self-efficacy
has a higher path coefficient in its relationship with confirmation than technological
self-efficacy.
Finally, confirmation’s mediating role between financial self-efficacy and per-
ceived usefulness is confirmed in the case of the FinTech loan app. Financial self-
efficacy only affects perceived usefulness through confirmation, meaning users only
perceive the FinTech loan app as helpful after confirming their expectation of the
use. This is true because users with high financial ability are more likely to only
engage in positive financial behaviors such as accessing loans using FinTech
because of speed and convenience. As these users exceed their expectations after
accessing the loans, they will feel a sense of confirmation, reinforcing their already
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 509

established personal beliefs and enhancing their perception of its usefulness. The
result is consistent with Shiau et al.’s (2020) finding that the mediation effect of
confirmation in the relationship between self-efficacy and perceived usefulness was
confirmed for wealth management platforms. As a result, users’ financial self-
efficacy is considered an essential factor in FinTech continuance intention for
FinTech loans; however, this only occurs through the indirect mediating role of
confirmation. On the other hand, the same cannot be said of the mediating role of
confirmation between technological self-efficacy and perceived usefulness. These
results are significant because, although domain-specific self-efficacy is critical for
continuance intention, FinTech companies need to pay attention to ensure the user’s
expectation concerning the FinTech product, especially for loan services, is
confirmed.

7 Conclusion

FinTech, combining digital technologies and financial services, has gained much
attention from scholars and businesses. This chapter examines the factors that drive
continuance intention when FinTech is used for accessing loans by incorporating
self-efficacy and ECT-IS theories. As the chapter emphasizes the explanatory and
predictive abilities of the model, it used PLSpredict to select the best model from
four competing models. The best model explains 46.2% of the FinTech continuance
intention. The result shows that users’ perceived usefulness is affected directly by
confirmation and indirectly by financial self-efficacy. Therefore, in addition to
identifying the importance of confirmation, perceived usefulness, and satisfaction,
as highlighted in previous studies, firms that provide FinTech services for loans
should also consider financial self-efficacy and ensure the products are well designed
to meet the user’s expectations. Doing so could increase users’ satisfaction and
continuance intentions, enhancing the profitability of FinTech companies.

7.1 Implications for Research and Practice

The chapter has two theoretical implications. First, by integrating self-efﬁcacy

theory and ECT-IS to understand users’ FinTech continuance intention for loans,
it showed that although financial self-efficacy is an essential driver of continuance,
technological self-efficacy does not have any impact, unlike Shiau et al. (2020) who
showed the relevance of both efficacies for FinTech wealth management apps. This
goes on to say that not all FinTech platforms are the same as they vary in the
products offered and hence also vary in principal continuance intention drivers.
Furthermore, the theoretical and empirical validation of the effects of financial self-
efficacy is also a significant research contribution of this chapter.
Second, this chapter advances our knowledge of FinTech consumers’ behavior in
Africa, especially for loan services. Prior research on FinTech in Africa remains
510 S. Danilola et al.

sparse despite FinTech growing importance. Former studies in Africa have focused
on mobile money using the E-Procurement Adoption Model (EPAM) model. This
chapter enriches the literature about FinTech by showing how the users’ reuse
intention for FinTech loans is affected by financial self-efficacy, perceived useful-
ness, confirmation, and satisfaction. It provides evidence in an area that is sparse in
the literature and takes us closer to understanding FinTech customers’ continuance
behavior.
In practice, this chapter has two implications. First, this study stresses the
relevance of the financial self-efficacy of FinTech users for loan services. The
chapter confirms that financial self-efficacy affects perceived usefulness indirectly
through the mediation of confirmation. The implication is that financial self-efficacy
should be addressed when designing products for users as it can indirectly affect
perceived usefulness when the expectation of the users has been confirmed. In taking
loans using FinTech, users are usually concerned about the facility’s convenience,
speed, and cost. Although the ability to use a smartphone seems essential, it does not
carry significant weight in determining perceived usefulness, so the focus should be
on financial self-efficacy through meeting the user’s expectations. Therefore,
FinTech firms and financial institutions could disaggregate users based on their
financial self-efficacies and design commensurate services for each segment to
enhance their financial knowledge while still focusing on meeting their expectations.
They should pay attention to users with higher financial self-efficacy as they are
more likely to perceive the usefulness of FinTech.
Second, the chapter shows that confirmation of users’ FinTech loan use expec-
tation is a significant determinant of satisfaction and a key factor as it also influences
the perceived usefulness of FinTech. Therefore, FinTech firms need to educate users
about the appropriate use of FinTech loan platforms so they can set the right kind of
expectations for its use because too high expectations could lead to disconfirmation,
and low perceived use will drag down the intention to continue using the FinTech
platform.

7.2 Limitations and Future Research Directions

This chapter seeks to understand users’ FinTech continuance intention for loan
services from a self-efficacy perspective. The study has made significant contribu-
tions but has a few limitations that provide avenues for future research.
First, this chapter limited the respondents to their experience with loan services
provided by FinTech firms. Hence, its findings should be applied with caution to
other FinTech-related services or products. Second, the study used cross-sectional
data. In the future, longitudinal research could be designed to test causal hypotheses
regarding financial self-efficacy, technological self-efficacy, and other critical factors
involved in FinTech use.
Finally, as our samples were from FinTech users in Nigeria, the results of this
chapter must be cautiously interpreted. Nonetheless, the findings have created more
insights into the behavior of FinTech users in different settings. Future studies
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 511

should consider national characteristics and cultural factors to explain the issue of
FinTech usage across other nations at the individual level.

Appendix

Constructs Items Adapted from

Financial self-efficacy FSE1 I am very capable of mak- Shiau et al.
The belief in one’s capability in achieving personal loan decisions (2020)
ing one’s ultimate financial goals FSE2 I am confident in my ability
(Forbes and Kara 2010) to make personal loan decisions
FSE3 I feel I am qualified for the
task of making personal loan
decisions
FSE4 Using available loan infor-
mation is well within the scope of
my abilities
FSE5 My past experiences
increase my confidence that I will
be able to make personal loan
decisions successfully
Technological self-efficacy (smartphone) TSE1 I can figure out how to use Hong et al.
The belief in one’s ability to operate the interface of a smartphone on (2014)
specific functions on a smartphone, such my own
as downloading and implementing soft- TSE3 I can figure out how to use
ware onto the smartphone (Hong et al. apps on my own
2014) TSE4 I can figure out how to use
the different functions provided
by smartphones on my own
Confirmation CON2 The service level provided Bhattacherjee
Users’ perception of the congruence by FinTech was better than I (2001), Shiau
between the expectation of FinTech use expected et al. (2020)
and its actual performance CON3 My experience obtaining a
(Bhattacherjee 2001) loan using FinTech was better
than I expected
CON4 Overall, most of my
expectations from using FinTech
were confirmed
Perceived usefulness PU1 Taking loans from FinTech Bhattacherjee
Users’ perception of the expected bene- improves my ability to obtain (2001), Shiau
fits of FinTech use (Bhattacherjee 2001) loans whenever I need and et al. (2020)
improves my personal finances
PU2 Taking loans from FinTech
increases my productivity in
managing personal finances
PU3 Taking loans from FinTech
enhances my effectiveness in
managing personal finances
PU4 Overall, loans from FinTech
are helpful in managing my
expenses
(continued)
512 S. Danilola et al.

Constructs Items Adapted from

Satisfaction I feel that my overall experience Shiau et al.
Users’ affect with (feelings about) prior of FinTech use is (2020)
FinTech use (Bhattacherjee 2001) SAT1 Very satisfied
SAT2 Very pleased
SAT3 Very contented
FinTech continuance intention FC1 I intend to continue taking Bhattacherjee
Users’ intention to continue taking loans loans and using FinTech rather (2001)
by FinTech (Bhattacherjee 2001) than discontinue its use
FC2 I intend to continue taking
loans using FinTech than using
any alternative means
FC4 I will continue taking loans
using FinTech
Notes: FSE Financial self-efficacy, TSE Technological self-efficacy, CON Confirmation, PU
Perceived usefulness, SAT Satisfaction, FC FinTech continuance intention

References

Akinwale YO, Kyari AK (2022) Factors influencing attitudes and intention to adopt financial
technology services among the end-users in Lagos State, Nigeria. Afr J Sci Technol Innov
Dev 14(1):272–279
Alalwan AA, Dwivedi YK, Rana NP, Williams MD (2016) Consumer adoption of mobile banking
in Jordan: examining the role of usefulness, ease of use, perceived risk and self-efficacy. J
Enterp Inf Manag 29(1):118–139
Asebedo S, Payne P (2019) Market volatility and financial satisfaction: the role of financial self-
efficacy. J Behav Financ 20(1):42–52
Bandura A (1986) The explanatory and predictive scope of self-efficacy theory. J Soc Clin Psychol
4(3):359–373
Bandura A (1991) Social cognitive theory of self-regulation. Organ Behav Hum Decis Process
50(2):248–287
Bhattacherjee A (2001) Understanding information systems continuance. MIS Q 25(3):351–370
Cassar G, Friedman H (2009) Does self-efficacy affect entrepreneurial investment? Strateg Entrep J
3(3):241–260
Cepeda Carrion G, Henseler J, Ringle CM, Roldán JL (2016) Prediction-oriented modeling in
business research by means of PLS path modeling. J Bus Res 69(10):4545–4551
Chin WW, Peterson RA, Brown PS (2008) Structural equation modelling in marketing: some
practical reminders. J Mark Theory Pract 16(4):287–298
Choi S (2018) What promotes smartphone-based mobile commerce? Mobile-specific and self-
service characteristics. Internet Res 28(1):105–122
Diamantopoulos A, Siguaw JA (2006) Formative vs reflective indicators in measure development:
does the choice of indicators matter? Br J Manag 13(4):263–282
Farrell L, Fry TR, Risse L (2016) The significance of financial self-efficacy in explaining women’s
personal finance behaviour. J Econ Psychol 54:85–99
Forbes J, Kara SM (2010) Confidence mediates how investment knowledge influences investing
self-efficacy. J Econ Psychol 31(3):435–443
Fornell C, Larcker DF (1981) Evaluating structural equation models with unobservable variables
and measurement error. J Mark Res 18(1):39–50
FinTech Loan Continuance Intention: How Far Can Self-Efficacies Go? 513

Franque FB, Oliveira T, Tam C (2021) Understanding the factors of mobile payment continuance
intention: empirical test in an African context. Heliyon 7(8):1–12
Gefen D, Rigdon EE, Straub D (2011) An update and extension to SEM guidelines for adminis-
trative and social science research. MIS Q 35(2):3–14
Hair JF, Hult GTM, Ringle C, Sarstedt M (2013) A primer on partial least squares structural
equation modelling (PLS-SEM). Sage, Los Angeles
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2016) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Hollingsworth CL, Randolph AB, Chong AYL (2017) An updated and expanded assess-
ment of PLS-SEM in information systems research. Ind Manag Data Syst 117(3):442–458
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Holden H, Rada R (2011) Understanding the influence of perceived usability and technology self-
efficacy on teachers’ technology acceptance. J Res Technol Educ 43(4):343–367
Hong J-C, Hwang M-Y, Tai K-H, Chen Y-L (2014) Using calibration to enhance students’ self-
confidence in English vocabulary learning relevant to their judgment of over-confidence and
predicted by smartphone self-efficacy and English learning anxiety. Comp Educ 72:313–322
Hope A (2019) Improving access to finance through fintech. Trade Law Centre (tralac). https://
www.tralac.org/documents/events/tralac/2743-tralac-brief-improving-access-to-finance-
through-fintech-march-2019/file.html. Accessed 25 Apr 2022
Hossain MA, Quaddus M (2012) Expectation–confirmation theory in information system research:
a review and analysis. Inf Syst Theory:441–469
Jarvis CB, Mackenzie SB, Podsakoff PM (2003) A critical review of construct indicators and
measurement model misspecification in marketing and consumer research. J Consum Res 30(2):
199–218
Joo SH, Grable JE (2004) An exploratory framework of the determinants of financial satisfaction. J
Fam Econ Iss 25(1):25–50
Khan GF, Sarstedt M, Shiau W-L, Hair JF, Ringle CM, Fritze MP (2019) Methodological research
on partial least squares structural equation modeling (PLS-SEM): an analysis based on social
network approaches. Internet Res 29(3):407–429
Kim Y, Choi J, Park Y-J, Yeon J (2016) The adoption of mobile payment services for Fintech. Int J
Appl Eng Res 11(2):1058–1061
Kock N (2015) Common method bias in PLS-SEM: a full collinearity assessment approach. Int J
e-Collab 11(4):1–10
Lim SH, Kim DJ, Hur Y, Park K (2018) An empirical study of the impacts of perceived security and
knowledge on continuous intention to use mobile fintech payment services. Int J Hum-Comp
Interact 35(10):886–898
Ma Q, Liu L (2005) The role of internet self-efficacy in the acceptance of web-based electronic
medical records. J Organ End User Comput 17(1):38–57
MacKenzie SB, Podsakoff PM (2012) Common method bias in marketing: causes, mechanisms,
and procedural remedies. J Retail 88(4):542–555
Mattila AS, Enz CA (2002) The role of emotions in service encounters. J Serv Res 4(4):268–277
Nascimento B, Oliveira T, Tam C (2018) Wearable technology: what explains continuance
intention in smartwatches? J Retail Consum Serv 43:157–169
Pan X (2020) Technology acceptance, technological self-efficacy, and attitude toward technology-
based self-directed learning: learning motivation as a mediator. Front Psychol 11:1–11
Podsakoff PM, Organ DW (1986) Self-reports in organisational research: problems and prospects. J
Manag 12(4):531–544
PWC (2016) Customers in the spotlight: how FinTech is reshaping banking. Global FinTech
Survey 2016 – report. PWC. https://www.pwc.com/gx/en/industries/financial-services/publica
tions/fintech-is-reshaping-banking.html. Accessed 7 Mar 2021
514 S. Danilola et al.

Ringle CM, Sarstedt M, Mitchell R, Gudergan SP (2020) Partial least squares structural equation
modeling in HRM research. Int J Hum Resour Manag 31(12):1617–1643
Ryu HS (2018) What makes users willing or hesitant to use Fintech? The moderating effect of user
type. Ind Manag Data Syst 118(3):541–569
Sarstedt M, Ringle CM, Hair JF (2017) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg A (eds) Handbook of market research. Springer,
Heidelberg
Schreiber JB, Nora A, Stage FK, Barlow EA, King J (2006) Reporting structural equation modeling
and confirmatory factor analysis results: a review. J Educ Res 99(6):323–338
Shiau WL, Sarstedt M, Hair JF (2019) Internet research using partial least squares structural
equation modeling (PLS-SEM). Internet Res 29(3):398–406
Shiau W-L, Yuan Y, Pu X, Ray S, Chen CC (2020) Understanding fintech continuance: perspec-
tives from self-efficacy and ECT-IS theories. Ind Manag Data Syst 120(9):1659–1689
Shmueli G, Ray S, Velasquez Estrada JM, Chatla SB (2016) The elephant in the room: predictive
performance of PLS models. J Bus Res 69(10):4552–4564
Susanto A, Chang Y, Ha Y (2016) Determinants of continuance intention to use the banking
services: an extension to the expectation-confirmation model. Ind Manag Data Syst 116(3):
508–525
Wang Z, Guan Z, Hou F, Li B, Zhou W (2019) What determines customers’ continuance intention
of FinTech? Evidence from YuEbao. Ind Manag Data Syst 119(8):1625–1637
Warsame MH, Ireri EM (2021) Fintechs’ future in Kenya: does social influence matter? J Afr Bus
23(4):1067–1087
Yuan S, Liu Y, Yao R, Liu J (2016) An investigation of users’ continuance intention towards
mobile banking in China. Inf Dev 32(1):20–34
Zhao X, Lynch JG, Chen Q (2010) Reconsidering Baron and Kenny: myths and truths about
mediation analysis. J Consum Res 37(2):197–206
Zhou W, Tsiga Z, Li B, Zheng S, Jiang S (2018) What influence users’ e-finance continuance
intention? The moderating role of trust. Ind Manag Data Syst 118(8):1647–1670
Does Coercive Pressure Matter
on the Practices and Performance
of the Public Procurement?

Gerald Zachary Paga Tinali

1 Introduction

Procurement is critical as more than 50% of the revenue in the private sector is spent
on procurement activities to enable the production of final products; meanwhile, in
the public sector, a large percentage of the government budget is being spent on
public procurement activities (Alsetoohy and Ayoun 2018; Dobrzykowski et al.
2012; OECD 2017; Ordanini and Rubera 2008; Presutti 2003; Van Weele 2018).
Procurement in the private and public sectors is like two sides of the coin with a
few similarities while being dominated by huge differences. Procurement in the
private sector has typically been characterized by the main objective of profit-
making and supply chain channels that are geared toward contributing to the return
on investment (Vachon and Klassen 2007). Meanwhile, procurement settings in the
public sector are geared to the issues such as market structure and equity and have
been centered on offering service to the general public who normally finance it
(Guarnieri and Gomes 2019; Johnson et al. 2003; Prier et al. 2016). Another notable
difference is that public procurement is mainly restricted to what legislation requires
leaving behind procurement in private sector with the more comprehensive option
with minimal limitations (Guarnieri and Gomes 2019).
Moreover, one of the primary difficulties confronting public bodies, with public
procurement happening in them, is the trade-off between accountability and effi-
ciency. In due course, governments have adopted rules and regulations as one of the
reforms to regulate the procurement process (Duncombe and Searcy 2007). In
Tanzania, institutional and legal framework interferences, inadequate knowledge

G. Z. P. Tinali (✉)
Department of General Management, University of Dar es Salaam Business School, University
of Dar es Salaam, Dar es Salaam, Tanzania
e-mail: [email protected]

of procurement among participants, and lack of information technology facilities are

among the challenges faced by the public procurement sector (Rasheli 2017). As per
the literature, low enforcement of the established procurement rules is among the
challenges to effective public procurement in the public sector (Mwakibinga and
Buvik 2013). These operational and regulatory challenges have motivated the
researcher to write this chapter focusing on public sector procurement.
In Tanzania, procurement constitutes about 70% of the government budget, thus
making the function a key focus for government institutions such as the Public
Procurement Regulatory Authority (PPRA). To improve operational efficiency, the
government has taken several measures, as in many countries. Established rules (also
referred to as coercive means) are prerequisites for buyers to improve efficiency and
accountability in the procurement process. Thus, the chapter investigated whether
these coercive means substantially impact public procurement success.
Additionally, the reviewed literature indicates that a lot has been done on
procurement-related studies in the private sector, especially the manufacturing
sector, with the focus on supply chain practices, e-procurement, management strat-
egies, and performance, whereas less attention has been paid to the impact of
coercive pressure on the performance of procurement in the public sector (Colin
et al. 2015; Gawankar et al. 2017; Makabira and Waiganjo 2014; Moon 2005;
Quesada et al. 2010). On top of that, considering agency theory with a focus on
the negative incentive scheme, which emphasizes the use of coercive means of
sanctions and monitoring, this chapter aims at examining whether coercive pressures
could have a significant impact on the performance of public procurement either
directly or as a moderator in relationship between practices and performance.

2 Literature Review

2.1 Theoretical Review

Scholars have been using a resource-based view (RBV) to explain the association
between procurement/purchasing practices and the performance of an organization.
Foerstl et al. (2016), grounded on RBV, found a significant impact of procurement
practices on operational performance. As procurement practices are intangible, this
makes the initiatives to find the interlinkage between processes and routines and
their effect on the firm’s general performance very complex (Ellram and Liu 2002;
Kerkfeld and Hartmann 2012). RBV tends to ignore the role of contextual variables
such as external environment uncertainty, and this however could be one of the
reasons most studies have uncertain conclusions on the relationship between prac-
tices and performance, as most of these studies ignored the role that external
environmental factors play as possible factors that could have an impact on an
organization’s performance. Therefore, this chapter addressed this theoretical gap
by incorporating coercive pressures in the model as a potential external factor that
could influence public procurement performance.
Does Coercive Pressure Matter on the Practices and Performance of. . . 517

Scholars have conceptualized coercive pressure as pressures that emerge from

either informal factors such as cultural forces or the firm’s operating environment or
can emerge from formal factors in terms of government laws and regulatory prac-
tices such as rules and codes of practice (DiMaggio and Powell 2000; Kshetri 2009;
Masocha and Fatoki 2018; Wu et al. 2013). Meanwhile, other scholars view coercive
pressure as the pressure related to rules, regulations, and punishment by government
regulators and other stakeholders (Phan and Baird 2015; Sarkis et al. 2010).
According to other scholars, coercive pressure comes from government regulations
and regulatory authorities in the public procurement sector. Regulations and regu-
latory authorities normally impose some fines, penalties, or threats of legal sanction
in case of noncompliance. This chapter adopted the agency theory to validate the role
of coercive pressures in order to induce performance in public sector procurement.
Researchers have used agency theory as a declaration for the unit of analysis
between agents and principals wherein in one part of the contract the principal
delegates work to the other part of the contract (the agent), expecting that the work
will be completed according to the principal’s expectations (Bendickson et al. 2016;
Eisenhardt 1989; Jensen and Meckling 1976). The main concern of the agency
theory is to resolve two problems within agency relationships. The first one is the
goal conflict problem between the two parts, and the other is the risk-sharing
problem as a result of different attitudes/preferences towards the risk between the
principal and the agent (Eisenhardt 1989). Some scholars regard the agency problem
as the problem of incentives, whereas others regard it as a problem resulting from
institutional structure (Panda and Leepsa 2017). Meanwhile, other scholars regard
the relationship between a principal and an agent as a contractual relationship
whereby the parties work for their own interests (Jensen and Meckling 1976;
Panda and Leepsa 2017).
In order to induce a certain level of desired effort in an agent, the principal always
uses the available information about the behavior of an agent to generate the agent’s
incentives; however, in some situations, the information cannot be easily obtained as
it involves some costs (Demougin and Fluet 2001). Due to an asymmetry of
information between a principal and an agent, the former cannot assess the latter’s
actions that coincide with those of the principal; in this situation, an agent can avail
an opportunity. This asymmetrical information has raised the issue of control in the
relationship between the agent and the principal; consequently, the principal could
opt for monitoring as an alternative way of motivating the agent (Demougin and
Fluet 2001). Monitoring refers to gathering all necessary information about the
agent’s actions; thus, it includes the direct supervision of the agent and the use of
output-related indicators (Demougin and Fluet 2001). In a situation whereby the
agent’s action can be less costly and visible to the principal, which is not always the
case, the principal could use a formal contract through which the agent will be paid
based on higher effort (Miller and Whitford 2007).
Adopting the tenets of the agency theory under public procurement settings, the
procurement officers/managers can assume the role of agents, and the government
assumes the role of the principal. For example, during procurement proceedings, an
agent can have information concerning particular bidders who might satisfy the
518 G. Z. P. Tinali

principal’s interests. However, in some situations, the agent may not release this
information to the principals, and the principals may end up with an inferior bidder
selected by the organization (McCue and Prier 2008).
The issue of information asymmetry under the principal–agency relationship
could be overcome by invoking the right incentives (Miller and Whitford 2007).
As per the principal agency theory, efficiency gain under public bureaucracy could
be realized by invoking a right incentive contract (Laffont and Martimort 2002;
Miller 2005; Miller and Whitford 2007; Mwakibinga and Buvik 2013). The concept
of the right incentive scheme for agents has not been widely applied in public
bureaucracies, the reason being that the issues of information asymmetry and risk
aversion altogether make it impossible for incentives to replace the monitoring of the
behavior of an agent (Miller 2005; Mwakibinga and Buvik 2013). Therefore, there is
another controlling mechanism regarded as the negative incentive scheme, which
greatly emphasizes the use of coercive means of monitoring and sanctions, which
has been defended by the literature on the applications of monitoring and sanction
mechanisms in inducing agent compliance (Miller 2005; Mwakibinga and Buvik
2013). Several studies have indicated the usefulness of monitoring actions in
lessening opportunistic but damaging behaviors to increase compliance (Eisenhardt
1989; Rokkan and Buvik 2003).
In public organizational settings, especially under public procurement, the issue
of rule compliance is one of the preconditions to which buyers are exposed
(Mwakibinga and Buvik 2013). However, studies indicate that enforcing procure-
ment practices that comply with the established framework has been challenging,
with some studies pinpointing the issues of monitoring and sanctions as ways of
improving compliance (Mwakibinga and Buvik 2013). The issue of monitoring has
previously received less theoretical attention, with more emphasis on monetary
incentives (Miller and Whitford 2007). As per Miller and Whitford (2007), the
right incentive scheme is proportional to the agency’s efficacy (ability to deliver
what is expected); thus, if the agency’s efficacy is very low, the right incentive
scheme will not help to generate efficiency. Most public organizations seldom rely
on pure incentive contracts; instead, they focus much on coercive means of moni-
toring and sanctions (Miller and Whitford 2007). In most cases, using bonuses as an
incentive to enhance effectiveness is always costly to the principal’s position; thus,
in controlling agency problems, principals opt for oversight and monitoring (Miller
and Whitford 2007).
One way of enforcing the behavior of procurement practitioners in the public
sector is through coercive means such as effective monitoring arrangements. In
Tanzania, different organizations have monitored procurement activities, including
the national audit office and the public procurement regulatory authority
(Mwakibinga and Buvik 2013). A study by Mwakibinga and Buvik (2013) indicated
that a significant and positive link exists between monitoring actions and the level of
compliance with the rules governing procurement in Tanzania. The effectiveness of
monitoring activities in the public procurement sector in Tanzania has been chal-
lenged by the lack of adherence to the rules requirement, with evidence indicating
that only a few procurement management units comply with the requirement, the
Does Coercive Pressure Matter on the Practices and Performance of. . . 519

main reason being the distance between the principal and the agent associated with a
lack of resources to enable implementation (Mwakibinga and Buvik 2013).
Additionally, applying sanctions (negative incentives) as compliance enforce-
ment is among the initiatives used by principals to make an agent behave as agreed to
(Mwakibinga and Buvik 2013). Sanctions as punitive measures are aimed at dis-
couraging undesirable behaviors. In the public procurement sector, among the
established procurement frameworks, they impose sanctions on those practitioners
who fail to comply. The sanctions can take the form of ﬁnes, debarments to public
procurement contracts, and blacklisting. In Tanzania’s public procurement sector,
there are established sanction mechanisms, such as the Public Procurement Act of
2011, which have some provisions on actions to be taken for those noncompliant
procurement practitioners (URT 2011).
Considering the Tanzanian experience, the existence of various coercive means
like regulators such as the Public Procurement Regulatory Authority (PPRA)
together with the various rules and regulations such as the Public Procurement Act
number 7 of 2011 and their accompanying regulations of 2013 act as negative
incentives to agents and inﬂuence higher public procurement performance (URT
2011, 2013).

2.2 Empirical Literature and Hypotheses Development

Studies in the private sector, speciﬁcally in manufacturing industries, indicate

significant positive relationships between practices and firm performance (Chen
et al. 2011; Chow et al. 2008; Ibrahim et al. 2010; Lenny Koh et al. 2007).
Additionally, several studies have shown a strong correlation between practices
and procurement performance (Alsetoohy and Ayoun 2018; Bag 2012; Makabira
and Waiganjo 2014; Quesada et al. 2010). A study using information gathered from
manufacturing companies in developing nations revealed a strong correlation
between procurement practices and performance (Quesada et al. 2010). According
to Sukati et al. (2012), Malaysian manufacturing industry performance is strongly
and significantly correlated with supply chain practices. Additionally, a study by
Alsetoohy and Ayoun (2018) discovered that procurement practices favorably
impact the procurement process performance.
In the public sector setting of the developed countries, Lingg et al. (2016) found
in their comparative qualitative study conducted on the effect of procurement
practices on the quality of medical services or services received differences in
terms of procurement process across countries. Meanwhile, some countries such as
Mexico were found to have deficiencies in terms of procurement regulations and
practices. The procurement practice in Mexico was found to be highly centralized
with price being the focus. However, this was seen as effective in fighting corruption
and increased transparency in procurement practices; this could compromise clinical
outcomes, especially for higher-risk medical devices, which was in contrast to the
condition in European countries. Furthermore, in Mexico’s health care procurement
520 G. Z. P. Tinali

practices, there is a legal requirement of an orthopedic specialist to be involved in

decision-making related to procurement; this is in contrast to the European countries.
A study by Callea et al. (2017) based on a survey study conducted in Italian hospitals
indicated that the regional health technological assessment (which is a centralized
practice) has a high chance of enabling purchasing of medical devices at lower costs,
whereas the hospital-based health technological assessment (which is a decentralized
practice) led to a costly purchase of medical devices. Duncombe and Searcy (2007)
based on a survey of New York State school districts developed a cost model. The
result indicated that procurement practices have a significant association with cost
savings.
Moreover, the available studies in developing countries indicated a positive
association between procurement practices and performance. In their study
conducted in Pakistan, Noor et al. (2013) referred to procurement practices as
different procurement methods, such as traditional and nontraditional. The study
based on archival review and case studies found procurement, especially process,
and methods, to impact directly the successful outcome of the projects.
A majority of these studies defined the term procurement practices differently and
also the context in which they were conducted and the methodology used. This
chapter strictly relied on the way Alsetoohy and Ayoun (2018) and Quesada et al.
(2010) conceptualized the construct procurement practices, the data were based only
on respondents who are public procurement practitioners in less developed coun-
tries, and it was purely quantitative research. Moreover, resource-based view had
been incorporated in previous studies to help explain the relationship between
practices and performance (Eisenhardt and Martin 2000; Foerstl et al. 2016; Teece
et al. 1997; Tinali 2021). Thus, based on these evidence, the chapter proposes a
hypothesis that examines the direct effect between procurement practices and pro-
curement performance.
H1: Procurement practices will be positively related to procurement performance.
On the other hand, applying sanctions (negative incentives) as compliance
enforcement is among the initiatives used by principals to make an agent behave
as required by the principal (Mwakibinga and Buvik 2013). Sanctions as punitive
measures are aimed at discouraging undesirable behaviors. Studies indicate that the
application of sanctions has attracted opposing views, with one group believing in
positive reinforcements such as incentives as an effective alternative to sanctions and
the other group supporting the issue of sanctions to enhance compliance
(Mwakibinga and Buvik 2013). Under the public procurement sector, among the
established procurement frameworks, imposition of sanctions on those practitioners
who fail to comply is included. The sanctions can take the form of fines, debarments
to public procurement contracts, and blacklisting. In Tanzania’s public procurement
sector, there are established sanction mechanisms such as the Public Procurement
Act of 2011 has some provisions on actions to be taken against noncompliant
procurement practitioners. A study by Mwakibinga and Buvik (2013) found a
significant positive relationship between sanctions and compliance with the appli-
cable procurement rules in public sector procurement.
Does Coercive Pressure Matter on the Practices and Performance of. . . 521

Even though studies argue that a firm’s operating environment outside of its
immediate environment influences the structure and operations of the firm, the
influence of the external environment in a firm’s supply chain has been largely
and significantly ignored (Donaldson 2001). According to research conducted by
Wiengarten et al. (2015), firms in environments with a high level of coercive
pressure may experience greater performance benefits. Taking Tanzania’s experi-
ence of the existence of various regulators such as the PPRA, together with the
existence of various rules and regulations such as the Public Procurement Act
number 7 of 2011 and their accompanied regulations of 2013, acts as negative
incentives to agents, influences agency compliance, and hence realizes higher public
procurement performance. With the help of the agency theory, this chapter decided
to incorporate the role of external environmental factors, such as regulatory pres-
sures, as potential factors that could impact public procurement performance. Thus,
the aforementioned hypothesis was developed based on these justifications.
H2: Coercive pressure will be positively related to procurement performance.
Despite studies indicating the positive impact between procurement practices and
procurement performance, several studies also indicate a conflicting result for the
two relationships (Zimmermann and Foerstl 2014). Moreover, little is known about
regulatory pressure as a moderator between procurement practices and procurement
performance, whereby several studies use regulatory pressure as a moderator on the
relationship between other variables rather than procurement practices (Tinali 2022;
Zhu and Sarkis 2007). A study by Zhu and Sarkis (2007) with data based on
manufacturing enterprises in China the study indicated that normative and coercive
pressures lead to better environmental performance (Zhu and Sarkis 2007). Further-
more, a literature review study on coercive pressure as a moderator indicated an
inconsistent result on the effect of organizational performance when coercive pres-
sure is present in an organization (Shamsuddin et al. 2013).
Public procurement, among the number of government functions, is also charac-
terized by principal agency relationships. The presence of regulations that guide the
conduct of the public procurement sector in Tanzania influences procurement
practices and public procurement performance. In order to test the moderating effect
of coercive pressure, the following hypothesis was developed based on these facts:
H3: Coercive pressure moderates the relationship between procurement practice
and procurement performance, so the relationship is stronger when coercive
pressures are high.

2.3 Research Model

Considering the above reviewed theoretical literature, empirical literature, and the
hypotheses formulated, the research model was formulated as shown in Fig. 1.
522 G. Z. P. Tinali

H2 (+)

Performance
Procurement
Coercive Pressure
H3 (+)
Procurement Practices

H1 (+)

Fig. 1 Research model developed based on the literature review: direct

effects and interaction effects

3 Methodology

This chapter believes that a social entity exists in reality outside of the social actors
and the chapter aimed at theory testing rather than theory development. Therefore,
the deductive approach suits the study (Saunders et al. 2009). Positivist epistemol-
ogy was opted because the researcher believed that truth is out there to be discovered
contrary to interpretivists who believe that truth should be developed based on social
interactions. Considering the nature of the research problems, the research model,
and the data source (collected from key informants from the procurement manage-
ment unit in government organizations using a well-structured questionnaire), this
chapter opted for a cross-sectional design (Creswell 2009; Robson 2002).
The population involved all 540 registered public procurement management units
(PMUs) in mainland Tanzania (PPRA 2018). As the study expected to use structural
equation modeling during data analysis, a sample size of 207 respondents was set for
this research (Hair et al. 2006; Schumacker and Lomax 2004). The nonprobabilistic
strategy of sampling based on subjective judgment was employed to select a sample.
Questionnaire was adapted as an instrument for the collection of intended data. The
choice of questionnaire was made due to its ease of administration, cost of admin-
istering, and reliability, and always when it is well constructed, it might contain
minimum errors in comparison with other instruments. In the process of identifying
potential problems related to scale and to ensure content validity and reliability, a
pilot study was conducted and the result was used to improve the final tool for data
collection. The unit of analysis was set at an organizational level. To provide more
balanced perspectives on variables, the study used the key informant approach by
collecting data from one respondent in each organization.
Based on criteria set on the decision of an appropriate epistemic relationship
(Jarvis et al. 2003), latent constructs were molded as reflective. The study measured
the procurement performance as a higher order construct (HOC) using three dimen-
sions, namely internal performance, supplier, and customer-related performances as
adopted from earlier research (Alsetoohy and Ayoun 2018; Mishra 2005; Quesada
et al. 2010; Tinali 2021, 2022). This chapter also measured procurement practices as
a HOC using the five dimensions proposed by Quesada et al. (2010) and later on
adapted by Alsetoohy and Ayoun (2018). The level of coercive pressure in public
Does Coercive Pressure Matter on the Practices and Performance of. . . 523

procurement was measured using the measurement items suggested by Kim et al.
(2015), Huang et al. (2016), and Zhu et al. (2013) and guidance from procurement
practitioners and experts. All items from all constructs were measured using a ﬁve-
point Likert scale that ranged from 1 (not at all) to 5 (to a great extent).

4 Results

The data collected were screened and validated. The screening process involved the
assessment of data entry accuracy and checking for missing data and suspicious
response pattern. After data screening, the next step was a data validation process,
which involved measurement model assessment, and it was followed by structural
model assessment (Hair et al. 2017). All the analyses were done with the aid of
SmartPLS Version 3 (Ringle et al. 2015).

4.1 Reﬂective Measurement Model Evaluation

As the model for the study was the reflective–reflective hierarchical component
model (HCM), it has been argued that in the reflective–reflective and formative–
reflective HCMs, the relation between the HOC and the lower order constructs
(LOCs) serves as the input for the estimation of the validity and reliability.
In Table 1, both LOC and HOC values for composite reliability are higher than
0.7, which provides an indication of the existence of internal consistency in the
underlying constructs. Moreover, Table 1 also indicates that convergent validity was
achieved as the values of the average variance extracted (AVE) for HOCs and LOCs
are greater than or equal to 0.5. Therefore, for the model developed in this chapter
each variable explains more than half of the variances of its indicator.
As the model involved higher order constructs, it is recommended that discrim-
inant validity between HOC and LOC is of no concern as we expect conceptual and
empirical redundancies (Hair et al. 2018; Sarstedt et al. 2019). However, it is further
recommended that discriminant validity needs to be established between the LOCs
in a particular HCM (Hair et al. 2018). Thus, as indicated in Table 2, the value of
0.687 is the highest HTMT value. For the model under this chapter, discriminant
validity of the variables was acceptable.

4.2 Sample Characteristics

Demographic information of key informants revealed that men constituted 64.7%

and women 35.3%. Furthermore, based on their work experience, almost 40% had
work experience in the same organization of less than 5 years, whereas the majority,
Table 1 Reliability and validity
524

Second-order Average variance extracted Composite reliability

construct CR (AVE) Construct Item Loadings (CR) AVE
Procurement prac- 0.819 0.5 Information gathering Infgath1 0.700 0.855 0.596
tices (INFGATH) Infgath2 0.738
(PRACT) Infgath3 0.851
Infgath4 0.792
Supplier contact Supcont1 0.712 0.847 0.581
(SUPCONT) Supcont2 0.815
Supcont3 0.748
Supcont4 0.771
Contracting, negotiation, and fulfill- Contnf2 0.806 0.899 0.640
ment Contnf3 0.847
(CONTNF) Contnf4 0.771
Contnf5 0.766
Contnf6 0.809
Requisition, negotiation, and fulfill- Requnf1 0.641 0.796 0.569
ment Requnf3 0.837
(REQUNF) Requnf4 0.771
Intelligent analysis (negotiation and Intanaly1 0.732 0.899 0.562
fulfillment) Intanaly3 0.804
(INTANALY) Intanaly4 0.759
Intanaly5 0.683
Intanaly6 0.721
Intanaly7 0.784
Intanaly8 0.756
Coercive pressure Coerpre1 0.647 0.813 0.521
(COERPRE) Coerpre2 0.793
G. Z. P. Tinali
Coerpre4 0.701
Coerpre5 0.740
Procurement perfor- 0.861 0.678 Internal performance Inteperf1 0.760 0.918 0.584
mance (INTEPERF) Inteperf2 0.833
(PERFO) Inteperf3 0.810
Inteperf4 0.660
Inteperf5 0.812
Inteperf6 0.684
Inteperf7 0.788
Inteperf8 0.751
Supplier-related performance Supperf2 0.883 0.869 0.691
(SUPPERF) Supperf3 0.854
Supperf4 0.750
Customer-related performance Custperf1 0.805 0.943 0.675
(CUSTPERF) Custperf2 0.769
Custperf3 0.807
Custperf4 0.794
Custperf5 0.836
Custperf6 0.877
Custperf7 0.873
Does Coercive Pressure Matter on the Practices and Performance of. . .

Custperf8 0.803
Notes: Items excluded: All items that were loaded below 0.5 and those items with loadings between 0.5 and 0.7 that negatively impact the average variance
extracted (AVE); Contnf1, Requnf2, Intanaly2, Coerpre3, Supperf1
525
526 G. Z. P. Tinali

Table 2 Discriminant validity on heterotrait–monotrait ratio (HTMT)

1 2 3 4 5 6 7 8 9
1. COERPRE
2. CONTNF 0.368
3. CUSTPERF 0.627 0.223
4. INFGATH 0.508 0.323 0.401
5. INTANALY 0.613 0.514 0.382 0.537
6. INTEPERF 0.578 0.304 0.687 0.300 0.417
7. REQUNF 0.440 0.426 0.241 0.417 0.564 0.283
8. SUPCONT 0.469 0.310 0.290 0.554 0.490 0.422 0.398
9. SUPPERF 0.443 0.270 0.569 0.298 0.409 0.550 0.151 0.323

about 60% of the key informants, had work experience of more than 5 years. The
respondent’s experience is an indication that they had enough knowledge and
possess adequate information on the subject of the study and thus were qualified
to be key informants. 64.3% were at the senior level of management in procurement-
related matters and 35.7% of the key informants were procurement and supplier
officers but with enough experience in public procurement. The final sample for the
study represents 85% of organizations from urban areas and 15% from rural areas.
This finding is logical taking into consideration the Tanzanian context whereby the
majority of government organizations are located in urban areas.

4.3 Model Estimation and Findings

The model in this chapter used a repeated indicator approach; that is, the HOC uses
all indicators of the LOC (Becker et al. 2012; Hair et al. 2018; Wetzels et al. 2009;
Wold 1982). Furthermore, the mode of measurement for repeated indicators in the
second-order latent variables was also measured reflectively (Henseler et al. 2009;
Wold 1982).
As path coefficient estimations in SEM is based on an ordinary least squares
regression of each endogenous latent variable on a given corresponding predecessor
construct, checking for collinearity issues is a requirement (Hair et al. 2017).
Collinearity was assessed by using the variance inflation factor (VIF). Table 3
demonstrates that all VIF values were less than the minimum needed threshold
value of 3. As a result, we conclude that the model is devoid of collinearity and
hence qualifies to be used for further analysis.
As shown in Table 3 as well as Fig. 2, all direct relationship estimations were
significant, thus supporting both the first and the second hypotheses stated in the
study. Coercive pressure had the most significant and strongest impact with public
procurement performance (β = 0.442; p < 0.01) compared to procurement practices
(β = 0.235; p < 0.01). These two latent variables together explain more than 35% of
procurement performance (R2 = 0.354). The value of effective size ( f2) is 0.062 for
Does Coercive Pressure Matter on the Practices and Performance of. . . 527

Table 3 Hypothesis testing for direct effects

95% conﬁdence interval
Hypothesis Relationships β t-value f2 (CI) VIF
H1 PRACT → PERFO 0.235 2.703* 0.062 [0.060;0.402] 1.377
H2 COERPRE → 0.442 5.506* 0.215 [0.272;0.583] 1.377
PERFO
Notes: β = standardized beta; effective size ( f2): 0.02 small, 0.15 medium, 0.35 large (Hair et al.
2019)
*p < 0.01

Fig. 2 PLS path model results for direct relationships

the ﬁrst hypothesis and 0.215 for the second hypothesis. As these values are far
above the smallest threshold, it indicates that though our path models as presented in
Table 3 are signiﬁcant they are also relevant. Furthermore, blindfolding result
528 G. Z. P. Tinali

indicates the value of Q2 for procurement performance to be 0.152. This Q2 value is

above the minimum required threshold value of 0, which is a clear justiﬁcation of the
presence of path model predictive relevance for each given endogenous variable
(Hair et al. 2017).

4.4 Moderation

The lack of a moderation effect in our model is justified by the fact that Table 4
shows the interaction effect to be insignificant. The relationship between the path
from COERPR*PRACT to PERFO is positive but statistically insignificant (β
= 0.024; p > 0.1). Table 4 and Fig. 3 demonstrate this. Additionally, the study
used the generated effective size to determine whether the moderators are relevant.
Table 4 shows the effective size of COERPR*PRACT on PERFO indicated a very
low effect ( f2 = 0.001), which is below the threshold, which further justifies that the
moderation effects under study are irrelevant and insignificant. As a significant
moderation effect was not achieved, there was no need to further do a post hoc
analysis of the moderation effect.

4.5 Importance–Performance Map Analysis (IPMA)

Importance–performance map analysis (IPMA) for both the construct level and
indicator level was created speciﬁcally to explore the issues that need improvement
in the public procurement sector. The result as shown in Figs. 4 and 5 is that for both
levels, the construct and items for coercive pressure (COERPR) are all important and
performing well.

5 Discussion of the Findings

This chapter has investigated the direct effects of coercive pressure and procurement
practices on public procurement performance and the moderation effect of coercive
pressure on the association between procurement practices and performance. The

Table 4 Moderation analysis report

Hypothesis Relationships β t–value f2 95% CI
H3 COERPR*PRACT → PERFO 0.024 0.229 ns 0.001 [-0.079;0.192]
Notes: β = standardized beta; effective size ( f2) for the interaction term whereby for f 2 = 0.005 or
higher denotes small effect, f 2 = 0.01 or higher denotes medium effect, and f 2 = 0.025 or higher
denotes strong effect (Kenny 2018)
ns not signiﬁcant
Does Coercive Pressure Matter on the Practices and Performance of. . . 529

Fig. 3 PLS path model results for the interaction effect

study expands our general understanding of procurement performance in public

organizations. The chapter aims at establishing theory applications and assessing
whether the results obtained support the existing evidence. Two theories have been
used under this chapter to explain the public procurement performance model. The
theories employed were the agency theory and the resource-based view. To supple-
ment the theories a number of empirical studies related to procurement management
performance were used. Both the theoretical and empirical evidence provided a good
base for the arguments on the elements of the public procurement sector discussed in
this chapter.
The results on practices and performance relationship was supported (β = 0.235,
p < 0.01 two tails). This implies that an increase in procurement practices partially
increases the performance of procurement management. This ﬁnding concurs with
530 G. Z. P. Tinali

90.000

COERPRE
85.000
Performance

80.000

75.000

PRACT

70.000

.200 .250 .300 .350 .400 .450

Importance

Fig. 4 Importance–performance map (construct level)

RBV and also supports the previous findings that proposed a similar conclusion
(Alsetoohy and Ayoun 2018; Bag 2012; Eyaa and Ntayi 2010; Makabira and
Waiganjo 2014; Quesada et al. 2010; Sukati et al. 2012), and more specifically
studies conducted in public sector procurement also supported the findings of this
study (Callea et al. 2017; Lingg et al. 2016; Noor et al. 2013; Tinali 2021).
Moreover, the relationship between coercive pressure and procurement perfor-
mance was positive and significant (β = 0.442, p < 0.01 two tails). The finding
implies that an increase in coercive pressure increases the performance of procure-
ment management. This finding is in support of agency theory, which supports the
use of coercive means in monitoring procurement so as to increase the performance
of procurement in the public sector. Finally in accordance with hypothesis 3 (H3) the
results on the relationship of the path from interaction term COERPR*PRACT to
PERFO is 0.024 and statistically insignificant (t = 0.229; p > 0.1). This implies that
a variation in coercive pressures as a moderator has nothing to do with the link
between practice and performance of public procurement; instead, coercive pressure
directly affects the performance of public procurement. Additionally, results from
IPMA indicate how important the variable coercive pressure is together with its
impact on public procurement performance.
Most procurement related studies tend to ignore the external environment’s
influence on organizational performance. Researchers have also criticized the
Does Coercive Pressure Matter on the Practices and Performance of. . . 531

Fig. 5 Importance–performance map (indicator level)

resource-based perspective for ignoring contextual factors such as external environ-

mental uncertainty that could affect the performance of the organization (Miller and
Shamsie 1996; Teece et al. 1997). Thus, the study has capitalized on this gap by
employing the agency theory to consider coercive pressure as an external environ-
ment uncertainty that could have an impact on public procurement performance.

6 Conclusion and Implications

This chapter is limited to government organizations and excludes private firms with
the main reason being that there are a lot of studies on procurement performance that
were strictly conducted by collecting information from manufacturing firms with
most of them being private firms. Furthermore, this chapter has been limited only to
some determinants of public procurement performance, such as procurement prac-
tices and coercive pressure. Therefore, the study recommends for further studies to
include other determinants such as organization age and procurement centralization.
The chapter used a reflective HOC and the analysis of the field data was accom-
plished using SmartPLS Version 3. Research involving partial least squares
532 G. Z. P. Tinali

structural equation modeling mostly tend to use the first order constructs only. Thus,
this chapter provides a significant methodological addition to partial least squares
studies, more specifically those involving hierarchical component models. The
chapter further highlights some key important elements to procurement performance
that need attention in the public sector. The positive and significant impacts of both
procurement practices and coercive pressure on performance suggest that procure-
ment practitioners aiming at higher procurement performance need to concentrate on
the use of coercive means. Therefore, practitioners can improve their poor perfor-
mance in terms of procurement management by invoking coercive pressures and
taking into consideration the improvement of their procurement practices.

References

Alsetoohy O, Ayoun B (2018) Intelligent agent technology: the relationships with hotel food
procurement practices and performance. J Hosp Tour Technol 9(1):106–120
Bag S (2012) World class procurement practices and its impact on firm performance: a selected case
study of an Indian manufacturing firm. J Supply Chain Manag Syst 1(3):27–39
Becker JM, Klein K, Wetzels M (2012) Hierarchical latent variable models in PLS-SEM: guidelines
for using reflective-formative type models. Long Range Plan 45(5–6):359–394
Bendickson J, Muldoon J, Liguori E, Davis PE (2016) Agency theory: the times, they are a-changin.
Manag Decis 54(1):174–193
Callea G, Armeni P, Marsilio M, Jommi C, Tarricone R (2017) The impact of HTA and procure-
ment practices on the selection and prices of medical devices. Soc Sci Med 174:89–95
Chen YJ, Su KW, Hsu MH, Hwang WJ, Wang JW (2011) The impact of aligning supply chain and
information system strategies on performance. Int J Bus Perform Manag 12(4):309–333
Chow WS, Madu CN, Kuei CH, Lu MH, Lin C, Tseng H (2008) Supply chain management in the
US and Taiwan: an empirical study. Omega 36(5):665–679
Colin M, Galindo R, Hernández O (2015) Information and communication technology as a key
strategy for efficient supply chain management in manufacturing SMEs. Procedia Comput Sci
55:833–842
Creswell JW (2009) Research design: qualitative, quantitative, and mixed methods approaches, 3rd
edn. Sage, California
Demougin D, Fluet C (2001) Monitoring versus incentives. Eur Econ Rev 45(9):1741–1764
DiMaggio PJ, Powell WW (2000) The iron cage revisited institutional isomorphism and collective
rationality in organizational fields. In: Baum JAC, Dobbin F (eds) Economics meets sociology
in strategic management (Advances in strategic management), vol 17. Emerald Group, Bingley,
pp 143–166
Dobrzykowski DD, Hong PC, Park SJ (2012) Building procurement capability for firm perfor-
mance: a service-dominant logic view. Benchmarking 19(4/5):567–584
Donaldson L (2001) The contingency theory of organizations. Sage, US
Duncombe W, Searcy C (2007) Can the use of recommended procurement practices save money?
Public Budg Financ 27(2):68–87
Eisenhardt KM (1989) Agency theory: an assessment and review. Acad Manag Rev 14(1):57–74
Eisenhardt KM, Martin AJ (2000) Dynamic capabilities: what are they? Strateg Manag J 21(10–11):
1105–1121
Ellram LM, Liu B (2002) The financial impact of supply management. Supply Chain Manag Rev
6(6):30–37
Eyaa S, Ntayi JM (2010) Procurement practices and supply chain performance of SMEs in
Kampala. Asian J Bus Manag 2(4):82–88
Does Coercive Pressure Matter on the Practices and Performance of. . . 533

Foerstl K, Franke H, Zimmermann F (2016) Mediation effects in the ‘purchasing and supply
management (SM) practice – performance link’: findings from a meta-analytical structural
equation model. J Purch Supply Manag 22:351–366
Gawankar SA, Kamble S, Raut R (2017) An investigation of the relationship between supply chain
management practices (SCMP) on supply chain performance measurement (SCPM) of Indian
retail chain using SEM. Benchmarking 24(1):257–295
Guarnieri P, Gomes RC (2019) Can public procurement be strategic? A future agenda proposition. J
Publ Procure 19(4):295–321
Hair JF, Black WC, Babin BJ, Anderson RE, Tatham RL (2006) Multivariate data analysis. Pearson
Education, NJ
Hair JF, Hult GT, Ringle CM, Sarstedt M (2017) A primer on partial least squares structural
equation modeling (PLS-SEM), 2nd edn. Sage, Thousand Oaks
Hair JF, Sarstedt M, Ringle CM, Gudergan SP (2018) Advanced issues in partial least squares
structural equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. Adv Int Mark 20:277–319
Huang YC, Yang ML, Wong YJ (2016) Institutional pressures, resources commitment, and returns
management. Supply Chain Manag 21(3):398–416
Ibrahim AR, Zolait AH, Sundram VP (2010) Supply chain management practices and firm
performance: an empirical study of the electronics industry in Malaysia. Int J Technol Diffus
1(3):48–55
Jarvis CB, MacKenzie SB, Podsakoff PM (2003) A critical review of construct indicators and
measurement model misspecification in marketing and consumer research. J Consum Res 30(2):
199–218
Jensen MC, Meckling WH (1976) Theory of the firm: managerial behaviour, agency costs and
ownership structure. J Financ Econ 3:305–360
Johnson PF, Leenders MR, McCue C (2003) A comparison of purchasing’s organizational roles and
responsibilities in the public and private sector. J Publ Procure 3(1):57–74
Kenny D (2018) Moderator variables. http://davidakenny.net/cm/moderation.htm. Accessed
4 May 2021
Kerkfeld D, Hartmann E (2012) Maximizing impact of investments into purchasing and supply
management. Int J Phys Distrib Logist Manag 42(5):464–489
Kim S, Kim G, French AM (2015) Relationships between need-pull/technology-push and infor-
mation security management and the moderating role of regulatory pressure. Inf Technol Manag
16(3):173–192
Kshetri N (2009) The development of market orientation: a consideration of institutional influence
in China. Asia Pac J Mark Logist 21(1):19–40
Laffont J, Martimort D (2002) The theory of incentives: the principal – agent model. Princeton
University Press, Princeton, NJ
Lenny Koh SC, Demirbag M, Bayraktar E, Tatoglu E, Zaim S (2007) The impact of supply chain
management practices on performance of SMEs. Ind Manag Data Syst 107(1):103–124
Lingg M, Wyss K, Durán-Arenas L (2016) Effects of procurement practices on quality of medical
device or service received: a qualitative study comparing countries. BMC Health Serv Res
16(1):1–13
Makabira DK, Waiganjo E (2014) Role of procurement practices on the performance of corporate
organizations in Kenya: a case study of Kenya national police service. Int J Acad Res Bus Soc
Sci 4(10):369–385
Masocha R, Fatoki O (2018) The impact of coercive pressures on sustainability practices of small
businesses in South Africa. Sustainability 10(9):3032
McCue C, Prier E (2008) Using agency theory to model cooperative public purchasing. J Publ
Procure 8(1):1–35
534 G. Z. P. Tinali

Miller GJ (2005) The political evolution of principal-agent models. Annu Rev Polit Sci 8:203–225
Miller D, Shamsie J (1996) The resource-based view of the firm in two environments: the
hollywood film studios from 1936 to 1965. Acad Manag J 39(3):519–543
Miller GJ, Whitford AB (2007) The principal’s moral hazard: constraints on the use of incentives in
hierarchy. J Public Adm Res Theory 17(2):213–233
Mishra AN (2005) Antecedents and consequences of internet use in procurement: an empirical
investigation of US manufacturing firms. Inf Syst Res 18(1):103–120
Moon MJ (2005) E-procurement management in state governments: diffusion of e-procurement
practices and its determinants. J Publ Procure 5(1):54–72
Mwakibinga FA, Buvik A (2013) An empirical analysis of coercive means of enforcing compliance
in public procurement. J Publ Procure 13(2):243–273
Noor MA, Khalfan MMA, Maqsood T (2013) The role of procurement practices in effective
implementation of infrastructure projects in Pakistan. Int J Manag Proj Bus 6(4):802–826
OECD (2017) Public Procurement. In: Government at a Glance 2017. pp 171–182. https://doi.org/
10.1787/gov_glance-2017-en
Ordanini A, Rubera G (2008) Strategic capabilities and internet resources in procurement: a
resource-based view of B-to-B buying process. Int J Oper Prod Manag 28(1):27–52
Panda B, Leepsa NM (2017) Agency theory: review of theory and evidence on problems and
perspectives. Ind J Corp Gover 10(1):74–95
Phan TN, Baird K (2015) The comprehensiveness of environmental management systems: the
influence of institutional pressures and the impact on environmental performance. J Environ
Manag 160:45–56
PPRA (2018) Annual performance evaluation report for financial year 2017/2018. Public Procure-
ment Regulatory Authorty, Dodoma
Presutti WD (2003) Supply management and e-procurement: creating value added in the supply
chain. Ind Mark Manag 32(3):219–226
Prier E, Schwerin E, McCue CP (2016) Implementation of sustainable public procurement practices
and policies: a sorting framework. J Publ Procure 16(3):312–346
Quesada G, González ME, Mueller J, Mueller R (2010) Impact of e-procurement on procurement
practices and performance. Benchmarking 17(4):516–538
Rasheli GA (2017) Action research in procurement management; evidence from selected lower
local government authorities in Tanzania. Action Res 15(4):373–385
Ringle CM, Wende S, Becker J-M (2015) SmartPLS 3. SmartPLS GmbH, Boenningstedt. http://
www.smartpls.com
Robson C (2002) Real world research: a resource for users of social research methods in applied
settings, 2nd edn. Blackwell, Oxford
Rokkan AI, Buvik A (2003) Inter-firm cooperation and the problem of free riding behavior: an
empirical study of voluntary retail chains. J Purch Supply Manag 9:247–256
Sarkis J, Gonzalez-Torre P, Adenso-Diaz B (2010) Stakeholder pressure and the adoption of
environmental practices: the mediating effect of training. J Oper Manag 28(2):163–176
Sarstedt M, Hair JF, Cheah JH, Becker JM, Ringle CM (2019) How to specify, estimate, and
validate higher-order constructs in PLS-SEM. Australas Mark J 27(3):197–211
Saunders M, Lewis P, Thornhill A (2009) Research methods for business students, 5th edn. Pearson
Education, London
Schumacker RE, Lomax RG (2004) A beginner’s guide to structural equation modeling, 2nd edn.
Lawrence Erlbaum Associates, NJ
Shamsuddin S, Abd-el. Moemenb M, Shoukryb AM, Attab M, Shalapxb M, Iqbalc SMJ,
Jahanzebc A, Saif-Ur-Rehman (2013) In the moderating effect of regulatory-pressure: impor-
tance of partnership and leadership in supply chain management practices. Life Sci 10:373–380
Sukati I, Hamid AB, Baharun R, Yusoff RM (2012) The study of supply chain management
strategy and practices on supply chain performance. Procedia Soc Behav Sci 40:225–233
Teece DJ, Pisano G, Shuen A (1997) Dynamic capabilities and strategic management. Strateg
Manag J 18(7):509–533
Does Coercive Pressure Matter on the Practices and Performance of. . . 535

Tinali GZP (2021) The mediation effect of procurement competence on the relationship between
practices and performance of the public sector procurement in Tanzania using higher-order
constructs in SmartPLS. ORSEA J 11(1):1–21
Tinali GZP (2022) Technology usage and public procurement performance in Tanzania. The
moderating role of regulatory pressure. Univ Dar es Salaam Libr J 17(1):18–37
URT (2011) Public Procurement Act No. 7 of 2011. Government Printers, Dar es Salaam
URT (2013) Public Procurement Regulations GN Number 446. Government Printer, Dar es Salaam
Vachon S, Klassen RD (2007) Supply chain management and environmental technologies: the role
of integration. Int J Prod Res 45(2):401–423
Van Weele AJ (2018) Purchasing and supply chain management, 7th edn. Cengage Learning
EMEA, UK
Wetzels M, Odekerken-schröder G, Van Oppen C (2009) Using PLS path modeling for assessing
hierarchical construct models: guidelines and empirical illustration. MIS Q 33(1):177–195
Wiengarten F, Bhakoo V, Gimenez C (2015) The impact of host country regulatory quality on the
value creation process in e-business supply chains. Int J Prod Res 53(16):4963–4978
Wold H (1982) Soft modeling: the basic design and some extensions. In: Joreskog KG, Wold H
(eds) Systems under indirect observation: causality, structure, prediction, vol 2. Elsevier,
Amsterdam, pp 1–54
Wu T, Daniel EM, Hinton M, Quintas P (2013) Isomorphic mechanisms in manufacturing supply
chains: a comparison of indigenous Chinese ﬁrms and foreign-owned MNCs. Supply Chain
Manag 18(2):161–177
Zhu Q, Sarkis J (2007) The moderating effects of institutional pressures on emergent green supply
chain practices and performance. Int J Prod Res 45(18–19):4333–4355
Zhu Q, Sarkis J, Hung LK (2013) Institutional-based antecedents and performance outcomes of
internal and external green supply chain management practices. J Purch Supply Manag 19(2):
106–117
Zimmermann F, Foerstl K (2014) A meta-analysis of the “purchasing and supply management
practice-performance link”. J Supply Chain Manag 50(3):37–54
Determinants of Behavioral Intention to Use
E-Procurement System in Developing
Countries: Suppliers’ Perception from
Tanzania

Deus N. Shatta

1 Introduction

Worldwide, there has been an incredible shift toward the adoption of e-procurement
system (Esimit and Kibet 2021; Ko et al. 2016; Masudin et al. 2021; Ndei and
Mutuku 2021; Ratnawati and Suryawan 2021; Waithaka and Kimani 2021b; Willy
and Paul 2021; Yuliawati et al. 2021). In the developing countries, Africa in
particular, the concept of e-procurement is just gaining popularity, especially in
the public sector (Afolabi et al. 2019; Alehegn 2019; Bainson 2018; Gambo et al.
2019; Jeptoo and Karanja 2017; Makoba et al. 2017; Mwangi et al. 2018; Pitso et al.
2018; Shatta et al. 2020b; Siwandeti et al. 2021a).
This is because e-procurement system is one of the most effective ways of
attaining sustainable procurement performance, which meets the needs of the present
suppliers and buyers without comprising the ability of future suppliers and buyers to
meet their own needs (Iles 2017; Waithaka and Kimani 2021b; Siwandeti et al.
2021b; Ndei and Mutuku 2021; Ratnawati and Suryawan 2021; Esimit and Kibet
2021; Willy and Paul 2021; Yuliawati et al. 2021; Masudin et al. 2021).

2 Background

Notwithstanding the beneﬁts and efﬁciency of e-procurement system in procurement

performance, this system is still not widely accepted in some suppliers in the
developing countries like Tanzania (URT 2018; Shilla 2019). This behavioral

D. N. Shatta (✉)
Department of Business and Entrepreneurship Studies, National Institute of Transport, Dar es
Salaam, Tanzania

intention to use e-procurement system has a signiﬁcant effect on procurement

performance of the system (Alehegn 2019; Siwandeti et al. 2021b).
Likewise, despite the increasing literature on the adoption of new technologies
(Elias 2021; Shatta and Shayo 2021; San et al. 2017; Dwivedi et al. 2019; Alias
2018; Taluka 2016; Shatta et al. 2020a; Waithaka and Kimani 2021b; Ndei and
Mutuku 2021; Ratnawati and Suryawan 2021; Esimit and Kibet 2021; Willy and
Paul 2021; Yuliawati et al. 2021; Masudin et al. 2021), less attention has been paid
on the mediation effect of behavioral intention to use e-procurement on procurement
performance, and many scholars have revealed inconsistent findings regarding the
determinants of e-procurement adoption (Alehegn 2019; Soong et al. 2019, 2020,
2021; Tutu et al. 2019;Waithaka and Kimani 2021a; Watuleke 2017).
For example, Waithaka and Kimani (2021a) and Alehegn (2019) cited e-security,
staffing, user acceptance, supplier readiness, perceived benefit, information technol-
ogy (IT) infrastructure, employee training, and top management support as the
determinants influencing the adoption of e-procurement. On the other hand, other
scholars (i.e., Elias 2021; San et al. 2017; Suliantoro et al. 2015) cited performance
expectancy, effort expectancy, social influence, and facilitating condition, suggested
in the Unified Theory of Acceptance and Use of Technology (UTAUT) model as the
determinants of the adoption of new technologies. Furthermore, studies (i.e.,
Dwivedi et al. 2019; Elias 2021; San et al. 2017; Shatta and Shayo 2021) consider
the effects of performance expectancy, social influence, and effort expectancy as
instrumental on the adoption of new technologies, while Taluka (2016),
Martawidjaja et al. (2021), and Shin et al. (2016) revealed contrastive findings on
the effects of effort expectancy on behavioral intention to use new technologies.
Similarly, Elias (2021) revealed that e-procurement usage had an insignificant
effect on the contractors’ performance ( p-value > 0.05). These findings are in
contrast with the findings of prior studies regarding the effect of the adoption of
new technologies on performance (Waithaka and Kimani 2021b; Ndei and Mutuku
2021; Ratnawati and Suryawan 2021; Esimit and Kibet 2021; Willy and Paul 2021;
Yuliawati et al. 2021; Masudin et al. 2021; Yalew and Solomon 2019).
In addition, Dwivedi et al. (2019) highlighted the mediation effect of attitude on
usage behavior of new technologies but did not consider the mediation effect of
behavioral intention to use new technologies on performance.
Therefore, the current chapter aimed at showing the mediation effects of behav-
ioral intention to use Tanzania National e-Procurement System (TANePS) on pro-
curement performance as the foremost determinant of extending the original
UTAUT by Venkatesh et al. (2003) toward the adoption of new technologies in
e-procurement context.
Determinants of Behavioral Intention to Use E-Procurement System. . . 539

3 Model Development

This chapter adopted the original UTAUT. The original UTAUT was established by
Venkatesh et al. (2003) after revising a Theory of Reasoned Action (TRA) and a
Theory of Planned Behavior (TPB), a Combined Theory of Planned Behavior/
Technology Acceptance Model (TPB/TAM), a Model of PC Utilization (MPCU),
Diffusion of Innovation Theory (DIT), Social Cognitive Theory (SCT), Motivational
Model (MM), and Technology Acceptance Model (TAM).
Venkatesh et al. (2003) used data from four organizations with three points of
measurements. The study’s findings were as follows: The eight models explained
between 17% and 53% of the variance in users’ intention to use information
technologies (ITs). Nevertheless, all the eight models are bested by the UTAUT
using the same data because it explained about 70% of variance in behavioral
intention (Dwivedi et al. 2019). The power to explain 70% of variance in behavioral
intention was the major reason to adopt this model in the current study. According to
Venkatesh et al. (2003), two direct elements of technology use, “intentions to use”
and “facilitating conditions”, were postulated.
The intention to use was affected by performance expectancy, effort expectancy,
social influence, and facilitating conditions, which are four key constructs that are
direct determinants of usage intention. Gender, age, experience, and voluntariness of
use act as moderators of the impact of the four key constructs on usage and
behavioral intention (Venkatesh et al. 2003). However, this chapter adopted only
the direct element of technology use, intentions to use, and facilitating conditions
was ignored because it has no direct effect on behavior intention. Figure 1 shows the
four direct elements and the moderators of the original UTAUT.
Furthermore, several prior studies have tested the effect of social influence on the
adoption of new technologies. For example, a study by Elias (2021) examined the
factors determining e-procurement usage and its influence on contractors’ perfor-
mance. Specifically, this study assessed the effect of social influence on

Performance Expectancy (PE)

Effort Expectancy (EE) Behavior

Intention Use
Behavior
Social Influence (SI)

Facilitating Condition (FC)

Voluntariness of
Gender Age Experience
Use

Fig. 1 Uniﬁed theory of acceptance and use of technology

540 D. N. Shatta

e-procurement usage as suggested in the Uniﬁed Theory of Acceptance and Use of

Technology (UTAUT) model by Venkatesh et al. (2003). The researchers used
stratified random sampling and a sample of 250 respondents. Questionnaires were
used to collect data, and the structural equation modeling (SEM) method was
employed for data analysis and for the assessment of the research model of the
study. The findings revealed that social influence was found to have a significant
relationship with e-procurement usage ( p-value < 0.05).
Other scholars such as Elias (2021), San et al. (2017), Dwivedi et al. (2019), and
Taluka (2016) justified the effects of social influence on different contexts of the
adoption of new technologies. However, the consistent results of the effect of social
influence on the adoption of new technologies with variations in sample size and the
contexts of the prior studies call for further validation of these findings. The current
study proposed the direct and indirect effects of social influence on behavioral
intention to use Tanzania National e-Procurement System and procurement perfor-
mance, respectively, a domain that has not been tested. In the context of this study,
social influence is defined as the degree of individual’s perception on those impor-
tant others (e.g., family and friends) who believe they should use TANePS. UTAUT
model by Venkatesh et al. (2003) validated the positive effect of social influence on
behavioral intention; there are a lot of empirical studies that have confirmed the
effect of social influence on the adoption of new technologies. For the justification of
the findings of the existing empirical studies and the UTAUT model suggestion with
regard to the effect of social influence on behavioral intention, the current chapter
proposed that social influence would positively and directly influence behavioral
intention to use TANePS.
H1: Social influence (SI) positively and directly affects behavioral intention to use
Tanzania National e-Procurement System.
In connection to that, the effect of performance expectancy on the adoption of
new technologies has been tested by many prior researchers. For example, Elias
(2021) examined the factors determining e-procurement usage and its influence on
contractors’ performance using the Unified Theory of Acceptance and Use of
Technology (UTAUT) model. The performance expectancy was hypothesized to
influence the e-procurement usage. Elias (2021) used stratified random sampling and
a sample of 250 respondents, and structural equation modeling (SEM) method was
employed for the analysis of data and evaluation of the research model.
The findings revealed that performance expectancy had a significant effect on
e-procurement usage ( p-value < 0.05). These findings not only justify the effect of
performance expectancy on e-procurement usage, but also call upon further studies
to validate the significant effect of performance expectancy on behavioral intention
to use e-procurement.
In addition, San et al. (2017) sought to incorporate the Unified Theory of
Acceptance and Use of Technology (UTAUT) model by Venkatesh et al. (2003)
with perceived risk theory (security risk and privacy risk) to explore its impact on the
intention to use mobile government services. Performance expectancy was used to
predict its influence on the intention to use mobile government services. San et al.
Determinants of Behavioral Intention to Use E-Procurement System. . . 541

(2017) adopted age, gender, and education level as moderator variables to provide
in-depth understanding of citizens’ preference for mobile government services. The
study used questionnaires to collect data from a sample of 265 respondents and
partial least squares structural equation modeling (PLS-SEM) method to analyze
the data.
The findings indicated that performance expectancy can be used as a predictor of
the adoption of mobile government services ( p-value < 0.05). These findings not
only confirm the application of performance expectancy in the mobile government
context, but also show a need for future researchers to consider the effect of
performance expectancy on behavioral intention to any other new technologies’
adoption (e.g., e-procurement adoption).
In the current chapter, performance expectancy is defined as a degree to which
using the technology will provide benefits to suppliers in performing certain activ-
ities and is theorized to affect behavioral intention to use TANePS. According to the
UTAUT model by Venkatesh et al. (2003), performance expectancy has been
verified to have a strong and positive effect on behavioral intention. However,
prior empirical studies have justified the effect of performance expectancy on the
adoption of new technologies (Shatta and Shayo 2021; Elias 2021; San et al. 2017).
For the purpose of validation of the findings of the existing empirical studies and the
UTAUT model’s suggestion concerning the effect of performance expectancy on
behavioral intention, the current chapter proposed that performance expectancy
would positively and directly influence behavioral intention to use TANePS.
H2: Performance expectancy (PE) positively and directly affects behavioral inten-
tion to use Tanzania National e-Procurement System.
Likewise, the effect of effort expectancy on the adoption of new technologies has
been confirmed by numerous prior studies. For example, a study by Dwivedi et al.
(2019) explained the acceptance and use of information system (IS) and information
technology (IT) innovations through testing of the effects of the effort expectancy on
the adoption of new technologies.
Dwivedi et al. (2019) used the UTAUT model and specifically predicted the
direct effect of the effort expectancy on behavioral intentions and usage of IT. A
combination of meta-analysis and structural equation modeling (MASEM) tech-
niques was used. The meta-analysis based on 1,600 observations on 21 relationships
coded from 162 prior studies on IS or IT acceptance and use of technologies. The
findings showed that effort expectancy has a direct effect and a statistically signif-
icant effect on the adoption of new technologies. These findings not only justify the
direct effect of effort expectancy on behavioral intentions and usage of IS or IT but
also call for further study to consider the indirect effect of the effort expectancy on
performance context (i.e., procurement performance).
Furthermore, Elias (2021) examined the factors determining e-procurement usage
and its influence on contractors’ performance. Specifically, the study assessed the
effect of effort expectancy on e-procurement usage using the Unified Theory of
Acceptance and Use of Technology (UTAUT) model. The effort expectancy was
hypothesized to influence e-procurement usage. The study used stratified random
542 D. N. Shatta

sampling to sample respondents, a sample of 250 respondents to collect data, and the
structural equation modeling (SEM) to assess the research model.
The findings revealed that the effort expectancy had a significant effect on
e-procurement usage ( p-value < 0.05). These findings confirm not only the influ-
ence of effort expectancy on the adoption of new technologies but also call for
further studies to validate the effect of effort expectancy on behavioral intention to
use e-procurement system and its procurement performance. In the current study,
effort expectancy is defined as a degree of ease associated with the supplier’s use of
TANePS. On the other hand, the UTAUT model by Venkatesh et al. (2003)
confirmed the positive effect of effort expectancy on behavioral intention to use
new technologies. In addition, more empirical studies have validated the effect of
effort expectancy on the adoption of new technologies including e-procurement
system. For the justification of findings of the existing empirical studies and the
UTAUT model’ proposal, the current chapter proposed that effort expectancy would
positively and directly influence behavioral intention to use TANePS.
H3: Effort expectancy (EE) positively and directly affects behavioral intention to use
Tanzania National e-Procurement System.
In the context of this chapter, behavioral intention is defined as an individual’s
readiness to use an Internet-based public procurement system (TANePS). According
to the UTAUT model by Venkatesh et al. (2003), behavioral intention has a strong
and positive effect on the usage behavior of new technologies. However, less
empirical studies justified the effect of e-procurement usage on performance (Elias
2021). For the validation of the findings of the existing empirical studies and the
UTAUT model’s suggestion with regard to the effect of the adoption of new
technologies on performance, the current chapter proposed that behavioral intention
would positively and directly influence procurement performance.
H4: Behavioral intention to use Tanzania National e-Procurement System (BIUT)
positively and directly affects procurement performance (PP).
Above all, the understanding of mediation effects of behavioral intention to use
new technologies on performance is essential because the mediation effect always
aims at extending the original theories and models. For example, the original
UTAUT model by Venkatesh et al. (2003) has been modified by prior studies
through the mediation effect (Venkatesh et al. 2016; Dwivedi et al. 2019).
Dwivedi et al. (2019) highlighted the mediation effect of attitude on usage
behavior of new technologies but did not consider the mediation effect of behavioral
intention to use new technologies on performance. Therefore, this chapter aimed at
showing the mediation effects of behavioral intention to use TANePS as a foremost
determinant of empirical knowledge contribution, a source of explaining compre-
hensively the social problem under a study area and a determinant of extending the
original UTAUT by Venkatesh et al. (2003) on the adoption of new technologies in
the context of e-procurement system.
The indirect effects of social influence, performance expectancy, and effort
expectancy on procurement performance through behavioral intention to use
Determinants of Behavioral Intention to Use E-Procurement System. . . 543

Independent Constructs Mediator Construct Dependent Construct

Social
Influence
(SI) H1+

Behavioural Procurement
H2+ H4+
Performance Intention to Performance
Expectancy Use TANePS (PP)
(PE)
(BIUT)

H3+

Effort
Expectancy
(EE)

Fig. 2 Conceptual model. Notes: Existing relationships from the original UTAUT
model. Theoretical gap (relationship that does not exist in UTAUT)

Tanzania National e-Procurement System are missing in prior theoretical and empir-
ical literature. According to the UTAUT model by Venkatesh et al. (2003), behav-
ioral intention mediates the effects of social influence, performance expectancy, and
effort expectancy on the usage behavior of new technologies. Empirical literature
has confirmed the effects of social influence, performance expectancy, and effort
expectancy on the adoption of new technologies. In addition, empirical literature has
validated the effects of new technologies on performance. For the justification of the
findings of the existing empirical studies and the UTAUT model proposal, the
current chapter proposed that social influence, performance expectancy, and effort
expectancy would positively and indirectly influence procurement performance
through behavioral intention to use TANePS.
H1×4: Behavioral intention to use Tanzania National e-Procurement System medi-
ates the positive and indirect effects of social influence on procurement perfor-
mance (H1×H4).
H2×4: Behavioral intention to use Tanzania National e-Procurement System medi-
ates the indirect effects of performance expectancy on procurement performance
(H2×H4).
H3×4: Behavioral intention to use Tanzania National e-Procurement System medi-
ates the indirect effects of effort expectancy on procurement performance
(H3×H4).
Figure 2 shows the conceptual model of this chapter with the hypothesized
relationships between the independent constructs, mediator construct, and the depen-
dent construct.
544 D. N. Shatta

Based on reﬂective measurement model criteria, the mathematical model

x = lY + e was adopted in this chapter to show the relationship between a latent
variable and its observed indicators shown in Fig. 2, where x is the observed
indicator variable, Y is the latent variable, the loading l is a regression coefficient
quantifying the strength of the relationship between x and Y, and e represents the
random measurement error (Sarstedt et al. 2022).
The original UTAUT model by Venkatesh et al. (2003) and the modified
UTAUTs by Chen et al. (2011) and Dwivedi et al. (2019) are clarifying only the
effect of behavioral intention and facilitating condition on the behavioral intention.
However, the linkage between behavioral intention and performance of new tech-
nology has less been considered in the existing theories concerning the adoption of
new technologies.
To fill this theoretical gap, the current chapter theorized that the mediation effect
of behavioral intention to use Tanzania National e-Procurement System on procure-
ment performance would be a new theoretical contribution because it has not been
comprehended in the existing theories.

4 Methodology

Based on the philosophical underpinning of this chapter, explanatory cross-sectional

survey research design was adopted because data were collected once from a
targeted population by studying a sample of that population (Creswell and Plano
2018). In addition, this chapter employed a survey method in gathering data because
it allows quantitative data to be collected and analyzed quantitatively using descrip-
tive and inferential statistics.
In order to meet the hypotheses of this chapter, 3169 suppliers registered in
TANePS were the target population, and among these, 730 suppliers were registered
after being trained and 2439 suppliers were registered in TANePS without being
trained (Shilla 2019; URT 2018). The sample size was obtained using Yamane
formula: n = N / (1 + N (e)2) (Magigi 2015) where n = the required sample, N =
target population, e = level of precision assuming a 95% conﬁdence level and a
precision of ±5%. Therefore, given N = 3169 (Shilla 2019), then n was computed to
355 (approximated).
In addition, based on the conceptual research model of this chapter, the rule of
thumb suggested by Hair et al. (2019) for applying PLS-SEM and SmartPLS
Version 3 software in data analysis was used to justify the minimum number of
suppliers required. Questionnaires with closed-ended questions were assigned num-
bers in simplifying and making the process of quantitative data analysis more
accurate. The collected quantitative data for respondents’ proﬁle were analyzed
using descriptive statistics with the help of IBM SPSS Statistics software Version
21. Partial least squares structural equation modeling (PLS-SEM) with the help of
SmartPLS Version 3 software was used for inferential statistical analysis for testing
the hypotheses.
Determinants of Behavioral Intention to Use E-Procurement System. . . 545

The evaluation of the measurement model and the structural model procedures
were conducted by using the established criteria by Hair et al. (2019). Four stages
were involved in assessing the reflective measurement models: examining the
indicator loadings (reliabilities), assessing the internal consistent reliability,
assessing the convergent validity, and assessing the discriminant validity. Likewise,
collinearity for structural model constructs was examined.
After checking for collinearity, the key criteria for assessing the structural model
in PLS-SEM were the following: the significance of the path coefficients, the level of
the R2 values, the f 2 effect size, and the predictive relevance (Q2 and the q2 effect
size) (Hair et al. 2019; Becker et al. 2018). The results for the evaluation of the
measurement and the structural models were good and met all criteria established by
Hair et al. (2019).

5 Results

About 60% of the expected sample size was obtained in the actual data collection;
among these, 37% were suppliers who registered in TANePS after being trained and
about 23% were suppliers who registered in the system without being trained. This
respondents’ rate was higher than the rate reported in some of the previous studies,
which amounted to only 24% (Siwandeti et al. 2021a, b).
The findings concerning education level against the time of registration of
respondents in TANePS revealed that majority of the respondents (44.3%) had a
bachelor degree. On the other hand, the majority of suppliers who were registered
without being trained had low level of education compared to those who were
registered in the system after being trained.
These results imply that education level had an influence on the decision for
registration in the system after being trained. Table 1 presents the findings on the
education level of respondents against the time of registration of respondents.
In addition, the results of chi-squared tests revealed very strong association
between the education of the respondents and the time of registration in TANePS.
Table 2 shows the results of chi-squared tests.
After running bootstrapping in SmartPLS Version 3, all hypothesized relation-
ships were supported ( p-value < 0.05). These results indicated that the conceptual
research model of the current chapter can be used in decision-making because all
theorized associations appeared to exist in real life. Figure 3 shows the statistical
significance of the hypothesized relationships.
Additionally, Table 3 shows the hypotheses tested, effects, p-values, and the
remark in each hypothesis.
Furthermore, two groups (suppliers who registered in TANePS after being trained
and suppliers who registered in TANePS without being trained) were compared with
regard to their perception on the determinants of behavioral intention to use TANePS
and their effect on the procurement performance. The chapter hypothesized that the
groups had no significant difference. The results are presented in Table 4, which
546 D. N. Shatta

Table 1 Education of respondents × time of registration in TANePS

After being Without being
trained trained Total
Level of education of Standard seven 0 5 5
respondents Secondary 0 15 15
level
Certiﬁcate 0 5 5
level
Diploma level 10 18 28
Bachelor’s 64 30 94
degree
Master’s 55 9 64
degree
PhD degree 1 0 1
Total 130 82 212

Table 2 Chi-squared tests’ results of education × time of registration

Degree of freedom Asymptotic signiﬁcance
Value (df) (two-sided)
Pearson’s chi-square 66.170a 6 0.000
Likelihood ratio 76.722 6 0.000
Linear-by-linear 62.793 1 0.000
association
Number of valid cases 212
a
6 cells (42.9%) have expected count less than 5. The minimum expected count is 0.39

SI_1 BIUT_1 BIUT_2 BIUT_3 BIUT_4 BIUT_5 PP_1

0.000
SI_2 PP_2
0.000
0.000 0.000 0.000
0.000 0.000 0.000 0.000
SI_3 PP_3
0.000 0.000
0.001
Social Influence 0.000
SI_4 PP_4
(SI) 0.000
0.000
0.000
PE_1 PP_5
0.000
0.001
0.000 0.000
PE_2 Behavioral Procurement PP_6
0.000
Intention to Use Performance (PP) 0.000
0.000
PE_3 TANePS (BIUT) PP_7
0.000

PE_4 Performance PP_8

Expectancy (PE)
EE_1

0.000
EE_2
0.000
0.000
EE_3
0.000

EE_4 Effort Expectancy

(EE)

Fig. 3 Statistical signiﬁcance of the hypothesized relationships

Determinants of Behavioral Intention to Use E-Procurement System. . . 547

Table 3 Hypotheses, tested effects, p-values, and the remark in each hypothesis
p-
Hypothesis Effect value Remark
Behavioral intention to use TANePS (BIUT) → Pro- H4 Direct 0.000 Supported
curement performance (PP)
Effort expectancy (EE) → Behavioral intention to use H3 Direct 0.000 Supported
TANePS (BIUT)
Effort expectancy (EE) → Procurement performance H3×H4 Indirect 0.000 Supported
(PP)
Performance expectancy (PE) → Behavioral intention H2 Direct 0.001 Supported
to use TANePS (BIUT)
Performance expectancy (PE) → Procurement perfor- H2×H4 Indirect 0.001 Supported
mance (PP)
Social inﬂuence (SI) → Behavioral intention to use H1 Direct 0.001 Supported
TANePS (BIUT)
Social inﬂuence (SI) → Procurement performance (PP) H1×H4 Indirect 0.002 Supported

Table 4 PLS–multigroup analysis results

Absolute difference of p-
Hypotheses total effects value Remarks
Behavioral intention to use TANePS (BIUT) → 0.027 0.356 Rejected
Procurement performance (PP)
Effort expectancy (EE) → Behavioral intention to 0.143 0.195 Rejected
use TANePS (BIUT)
Effort expectancy (EE) → Procurement perfor- 0.113 0.177 Rejected
mance (PP)
Performance expectancy (PE) → Behavioral inten- 0.193 0.917 Rejected
tion to use TANePS (BIUT)
Performance expectancy (PE) → Procurement per- 0.135 0.907 Rejected
formance (PP)
Social inﬂuence (SI) → Behavioral intention to use 0.137 0.207 Rejected
TANePS (BIUT)
Social inﬂuence (SI) → Procurement performance 0.108 0.213 Rejected
(PP)

shows that all hypotheses were rejected and therefore the two groups had no
difference in their perception.
In addition, the importance–performance map analysis was performed. Based on
the position of the constructs’ social inﬂuence, performance expectancy, and effort
expectancy in the second quadrant, it is apparent that these constructs are not as
important but have higher performance for the existence of the research model. On
the other hand, the position of behavioral intention in the ﬁrst quadrant is more
important and has higher performance for the decision-making about the applicabil-
ity of the research model. The results are shown in Fig. 4, which imply that the model
is useful in decision-making.
548 D. N. Shatta

Importance-Performance Map
100
90
80
70
60
Procurement Performance (PP) 50
40
30
20
10
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Total Effects

Behavioral Intention to Use TANePS (BIUT) Effort Expectancy (EE) Performance Expectancy (PE)
Social Influence (SI)

Fig. 4 Importance–performance map (TA, TANePS Adoption) (constructs’ unstandardized

effects)

6 Additional Analyses and Discussion

The effect of social influence on the behavioral intention to use TANePS was
theorized in this chapter. The findings of this hypothesis confirmed the positive
path coefficient, which was statistically significant ( p-value < 0.05) implying that
the relationship of social influence and behavioral intention to use TANePS exists in
real life. In addition, the findings revealed a positive path coefficient, which means
that an increase in one standard deviation of social influence leads to an increase in
the rate of behavioral intention to use TANePS; hence, this determinant has a direct
influence on the suppliers’ decisions to use the new procurement system (TANePS).
On the other hand, these findings are similar to the findings of prior studies, which
tested the effect of social influence on the adoption of new technologies (Elias 2021;
San et al. 2017; Dwivedi et al. 2019; Taluka 2016). These prior studies justified the
effects of social influence on different contexts of the adoption of new technologies
including the adoption of e-procurement system.
The chapter also predicted that performance expectancy would have a positive
effect on behavioral intention to use TANePS. The findings revealed a positive path
coefficient, which means that an increase in one standard deviation of performance
expectancy leads to an increase in the rate of behavioral intention to use TANePS.
Hence, this determinant has a direct influence on the suppliers’ decisions to use the
new procurement system (TANePS). In addition, the findings revealed a significant
effect of performance expectancy on behavioral intention to use TANePS ( p-value
< 0.05), which implies that the effect of performance expectancy on behavioral
intention to use TANePS exists in real life.
These findings are consistent with the findings of prior studies, which tested the
effects of performance expectancy on the adoption of new technologies including the
adoption of e-procurement (Shatta and Shayo 2021; Elias 2021; San et al. 2017;
Determinants of Behavioral Intention to Use E-Procurement System. . . 549

Dwivedi et al. 2019). On the other hand, the findings of the current chapter are
contrary to the findings of other prior studies, which hypothesized the effect of
performance expectancy on the adoption of new technologies (Taluka 2016).
Moreover, the effect of effort expectancy on behavioral intention to use TANePS
was predicted in this chapter. The results showed a positive path coefficient, which
implies that an increase in one standard deviation of effort expectancy entails an
increase in the rate of behavioral intention to use TANePS. Hence, this determinant
has a direct influence on the suppliers’ decisions to use TANePS. In addition, the
findings revealed a significant effect of effort expectancy on behavioral intention to
use TANePS ( p-value < 0.05), which means that the effect of effort expectancy on
behavioral intention to use TANePS exists in real life.
These findings are similar to the findings of past studies. For example, a study by
Elias (2021) revealed that effort expectancy had a significant effect on the usage of
e-procurement ( p-value < 0.05). On the other hand, a study by San et al. (2017)
indicated that effort expectancy can be used as a predictor of the adoption of mobile
government services ( p-value < 0.05). However, Taluka (2016) revealed that the
effect of effort expectancy on behavioral to use mobile payment services was weak,
contrary to the findings of the current study and those of other prior studies (Dwivedi
et al. 2019; Elias 2021; San et al. 2017).
In addition, the effect of behavioral intention to use TANePS on procurement
performance was predicted in this chapter. The results showed a positive path
coefficient, which means that an increase in one standard deviation of behavioral
intention to use TANePS entails an increase in the rate of procurement performance.
Hence, the behavioral intention of suppliers to use TANePS has a direct influence on
the success of procurement performance of the system. In addition, the findings
revealed a significant effect of behavioral intention to use TANePS on procurement
performance ( p-value < 0.05), which means that the behavioral intention of sup-
pliers to use TANePS on procurement performance exists in real life.
These findings are similar with the findings of prior studies (Waithaka and
Kimani 2021b; Ndei and Mutuku 2021; Ratnawati and Suryawan 2021; Esimit
and Kibet 2021; Willy and Paul 2021; Yuliawati et al. 2021; Masudin et al. 2021),
which indicated that the adoption of new technologies has a positive significant
effect on performance.
However, these results are in contrast with the findings of a study by Elias (2021)
who revealed that the usage of e-procurement has an insignificant effect on the
contractors’ performance ( p-value > 0.05). Generally, prior studies have shown that
the adoption of e-procurement system leads to improved procurement performance
because it facilitates electronic documentation of the bidding process, thus enhanc-
ing accountability and transparency, reducing transaction cost, improving process
efficiency, increasing contract compliance, and reducing cycle times and inventory
costs (Masudin et al. 2021).
On the other hand, Yuliawati et al. (2021) observed that one of the factors
influencing the performance of procurement personnel is the application of
e-procurement because it does not only prioritize websites and online services, but
it also prioritizes the infrastructure of information and communication technology
550 D. N. Shatta

Independent Constructs Mediator Construct Dependent Construct

Social
Influence
(SI)

Behaviour
Procurement
Effort Intention to
Performance
Expectancy Use TANePS
(PP)
(EE) (BIUT)

Performance
Expectancy
(PE)

Fig. 5 Validated conceptual model. Notes: Supported hypotheses that exist in

UTAUT model. Supported hypothesis as theoretical contribution

and human resource capacity. Thus, the application of e-procurement leads to

improved procurement performance (Bainson 2018; Waithaka and Kimani 2021b;
Ndei and Mutuku 2021).
The indirect and positive effects of social influence, performance expectancy, and
effort expectancy on procurement performance through behavioral intention to use
TANePS were predicted. The findings revealed positive path coefficients, which
imply that an increase in one standard deviation of social influence, performance
expectancy, and effort expectancy leads to an increase in the rate of procurement
performance. Hence, social influence, performance expectancy, and effort expec-
tancy have a positive and an indirect influence on the success of procurement
performance of the system. In addition, the findings revealed a statistically positive
significant effect of social influence, performance expectancy, and effort expectancy
on procurement performance ( p-value < 0.05), which means that positive and
indirect effects of social influence, performance expectancy, and effort expectancy
on procurement performance exist in real life.
However, these findings are in contrast with the findings of prior studies because
they are regarded as a source of contribution to empirical knowledge, explaining
comprehensively the social problem under the study and extending the original
UTAUT model by Venkatesh et al. (2003) toward the adoption of new technologies
in the context of e-procurement adoption system and its performance. Figure 5
shows the final validated conceptual model of this chapter with positive and indirect
effects of social influence, performance expectancy, and effort expectancy on pro-
curement performance through behavioral intention to use Tanzania National
Determinants of Behavioral Intention to Use E-Procurement System. . . 551

SI_1 BIUT_1 BIUT_2 BIUT_3 BIUT_4 BIUT_5 PP_1

0.000
SI_2 PP_2
0.000
0.000 0.000 0.000
0.000 0.000 0.000 0.000
SI_3 PP_3
0.000 0.000
0.002 0.477
0.000
SI_4 Social Influence PP_4
(SI) 0.000
0.000
0.000
PE_1 PP_5
0.000
0.001 0.000
0.000 0.000
PE_2 Behavioral Procurement PP_6
0.000 Intention to Use Performance (PP) 0.000
0.000 TANePS (BIUT)
PE_3 0.000 PP_7
0.000 0.774
PE_4 Performance PP_8
Expectancy (PE)
EE_1
0.000
EE_2
0.000
0.000
EE_3
0.000

EE_4 Effort Expectancy

(EE)

Fig. 6 The results of mediation effects

e-Procurement System, a substance that is missing in prior theoretical and empirical

literature.
Additionally, the types of mediation effects that could exist in the validated
conceptual model were tested. The situation of partial mediation occurs when both
the indirect and direct effects are significant, while full mediation occurs when the
direct effect is not significant and the indirect effect is significant (Hair et al. 2019).
Based on this fact, further analysis was performed to test the kind of mediation
effects that exist in the validated conceptual model. Direct effects of social influence,
performance expectancy, and effort expectancy on procurement performance were
introduced in the validated conceptual model, and bootstrapping process was
performed.
After bootstrapping process, it was noted that the direct effects of social influence
and effort expectancy on procurement performance were statistically insignificant,
while their indirect effects were statistically significant, which means full mediation
exist (social influence and effort expectancy are fully mediated by behavioral
intention to use TANePS). On the other hand, the direct and indirect effects of
performance expectancy on procurement performance were both statistically signif-
icant, which means partial mediation exists (performance expectancy is partially
mediated by behavioral intention to use TANePS). Following these results, it is
confirmed that partial and full mediation effects exist in the validated conceptual
model. Figure 6 shows the results of testing mediation effects.
552 D. N. Shatta

7 Conclusion and Implications for Theory and Practice

Since all relationships of the validated model have positive path coefficients and
were statistically significant ( p-value < 0.05), it is therefore concluded that social
influence, performance expectancy, and effort expectancy are determinants of sup-
pliers’ behavioral intention to use TANePS.
On the other hand, behavioral intention of suppliers to use TANePS is a deter-
minant of the success of procurement performance of the system. Based on the
validated model of this chapter shown in Fig. 5, the linkage of behavioral intention to
use TANePS and procurement performance was missing in UTAUT model. This
chapter has validated the existence of the effect of behavioral intention to use
TANePS on procurement performance. These results imply that the new linkage
between behavioral intention to use TANePS and procurement performance is a
theoretical contribution to UTAUT model in the adoption of e-procurement.
However, UTAUT model has contributed to the validation of the hypotheses
predicted in the proposed model in Fig. 2. The validated model has practical
implications because all relationships hypothesized in this chapter were supported.
This means that the success of procurement performance of the system depends
directly on the suppliers’ behavioral intention to use TANePS, and it also depends
indirectly on the social influence, performance expectancy, and effort expectancy.
For that reason, the training of suppliers on the determinants of behavioral
intention to use TANePS is vital.
Furthermore, the results of chi-squared tests in Table 2 showed that education
level of respondents and the time of registration had a strong correlation ( p-value
< 0.05). Likewise, the results of PLS–multigroup analysis in Table 4 indicated that
the perception of groups of suppliers on the determinants of behavioral intention to
use TANePS was similar ( p-value > 0.05), and the results of importance–perfor-
mance of the constructs in Fig. 4 indicated that all constructs are important in the
validated model and can be used in decision-making.
Based on these findings, the governments in the developing countries should keep
on training suppliers pertaining to the determinants of procurement performance not
only for the success of expansion and reform of e-procurement adoption but also for
achieving value for money because more suppliers will be motivated to register in
the new procurement system.
This chapter ignored the facilitating condition and was limited to some constructs
(social influence, performance expectancy, effort expectancy, and behavioral inten-
tion) from the UTAUT model. The constructs included in this chapter were able to
explain only 56.9% of the variance in the procurement performance compared to the
UTAUT model that explained about 70% of variance in behavioral intention
(Dwivedi et al. 2019).
It is therefore recommended that further research may include the facilitating
condition to improve the variance in the procurement performance.
Determinants of Behavioral Intention to Use E-Procurement System. . . 553

References

Afolabi A, Ibem E, Aduwo E, Tunji-Olayeni P, Oluwunmi O, Ayo-Vaughan E (2019) Gauging

parameters for e-procurement acquisition in construction businesses in Nigeria. Int J Constr
Manag 22(3):426–435
Alehegn B (2019) Factors affecting adoption of electronic procurement in Ethio-Telecom Head
Office, Addis Ababa. Master Thesis, College of Business and Economics, Bahir Dar University,
Ethiopia
Alias SE (2018) Using the UTAUT model to analyze e-procurement adoption in Malaysian
construction industry. Insight J 1(1):90–99
Bainson H (2018) Improving procurement performance in the public sector with the implementa-
tion of e-procurement: a study of selected metropolitan and municipal assemblies in the Ashanti
Region of Ghana. Master Thesis, Kwame Nkrumah University of Science and Technology,
Ghana
Becker JM, Ringle CM, Sarstedt M (2018) Estimating moderating effects in PLS-SEM and PLSc-
SEM: interaction term generation data treatment. J Appl Struct Equation Model 2(2):1–21
Chen LS-L, Kuan CJ, Lee Y-H, Huang H-L (2011) Applicability of the UTAUT model in playing
online game through mobile phones: moderating effects of user experience. In: First Interna-
tional Technology Management Conference, San Jose, CA, pp 625–629
Creswell JW, Plano CVL (2018) Designing and conducting mixed methods research. Sage,
Thousand Oaks, CA
Dwivedi YK, Rana NP, Jeyaraj A, Clement M, Williams MD (2019) Re-examining the Unified
Theory of Acceptance and Use of Technology (UTAUT): towards a revised theoretical model.
Inf Syst Front 21(3):719–734
Elias SB (2021) The adoption of e-procurement on performance impact: Malaysian contractors’
perspectives. PhD Thesis, Asia University, Malaysia
Esimit FE, Kibet Y (2021) Influence of e-selection on performance of evaluation committee in
county government of Turkana. Int Res J Bus Strat Manag 2(1):120–129
Gambo MM, Dodo M, Yusuf H (2019) Assessment of stakeholders’ perception of risk determinants
associated with the adoption of e-procurement in the Nigerian construction industry. In: West
Africa Built Environment Research (WABER) Conference Proceedings, Accra, Ghana, pp
188–199
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Iles J (2017) How is e-procurement related to the success of U.S. Cities’ Sustainable Purchasing
Policies? Arizona State University, ASU School of Sustainability, pp 1–16. https://keep.lib.asu.
edu/_flysystem/fedora/c269/JIles-Final_Report.pdf. Accessed 5 May 2021
Jeptoo N, Karanja K (2017) Effect of governance structure on e-procurement adoption by state
corporations in Kenya. Int Acad J Procure Supply Chain Manag 2(3):76–91
Ko W-H, Jeong SR, Shin K-H, Lee CY (2016) A study on intention of use e-procurement systems
for public institutions user. Int J u- e-Serv Sci Technol 9(10):435–444
Magigi W (2015) Research proposal development and report writing, pathway for success in higher
learning institutions. Safi Publishers, Moshi
Makoba N, Nyamagere G, Eliufoo H (2017) E-procurement risks and mitigation: the case for
Tanzania construction companies. Int J Constr Eng Manag 6(4):180–186
Martawidjaja H, Hendriana D, Budiarto E, Baskoro G, Soetomo MAA, Nasution H (2021) Pro-
curement framework analysis and evaluation of e-procurement implementation adoption using
UTAUT model. In: Strategic and Program Management - Proceedings of the Conference on
Management and Engineering in Industry, 3(3):31–36
Masudin I, Aprilia GD, Nugraha A, Restuputri DP (2021) Impact of e-procurement adoption on
company performance: evidence from Indonesian manufacturing industry. Logistics 5(1):1–16
554 D. N. Shatta

Mwangi GP, Kiarie D, Kiai R (2018) Integrated financial management information system,
procurement performance and customer satisfaction in the county Government of Nyeri. Int J
Supply Chain Manag 3(1):127–145
Ndei FM, Mutuku M (2021) E-procurement and performance of non-governmental organizations in
Kenya. Empirical evidence from Pathfinder International, Kenya. J Bus Manag Sci 9(2):71–80
Pitso N, Kabanda S, Kapepo M (2018) Exploring e-procurement adoption in the context of a
developing country: the case of Lesotho, e-infrastructure and e-services for developing
countries. In: Odumuyiwa V, Adegboyega O, Uwadia C (eds) Lecture Notes of the Institute
for Computer Sciences, Social: E-Infrastructure and E-Services for Developing Countries: 9th
International Conference Proceedings, Africomm 2017, Lagos, Nigeria, December 11–12.
Springer International Publishing
Ratnawati NLTT, Suryawan MY (2021) E-procurement implementation as reflection of good
governance in north Lombok regency. J Human Soc Sci Publ Admin Manag 1(1):8–14
San ANC, Yee CJ, Moorthy K, Lee AFT (2017) Intention to use m-government services: age,
gender and education really matter? Int J eBus eGovern Stud 9(2):1–31
Sarstedt M, Ringle CM, Hair JF (2022) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg AE (eds) Handbook of market research. Springer, Cham,
pp 587–632
Shatta DN, Shayo FA, Layaa JN (2020a) Influence and linkage of buyers’-suppliers’ attitude
towards e-procurement adoption in developing countries: Tanzania context. Int J Appl Res
Manag Econ 3(1):52–63
Shatta DN, Shayo FA, Layaa JN, Mchopa DA (2020b) Stakeholders’ perception on critical success
factors influencing electronic procurement adoption in developing counties: experience from
Tanzania. Am Scientif Res J Eng Technol Sci 71(1):61–86
Shatta DN, Shayo FA (2021) The influence of performance expectancy on e-procurement adoption
model in developing countries: Tanzanian’s perception. J Eng Technol Indus Applic Manaus
7(29):4–12
Shilla P (2019) Highlights of Tanzania national e-procurement system. In: 12th East African Public
Procurement Forum, Arusha, Tanzania
Shin K-H, Lee CY, Ko W-H (2016) A study on the intention of public institution managers and
procurement business bidding managers to use the e-procurement system. Int J Secur Applic
10(10):57–66
Siwandeti M, Sanga C, Panga F (2021a) Technological factors influencing vendors’ participation in
public e-Procurement system in Ilala, Tanzania. East Afr J Soc Appl Sci 3(1):91–102
Siwandeti M, Sanga C, Panga F (2021b) Perceived benefits of participation in public electronic
procurement: a comparative analysis of vendors in Ilala District, Tanzania. J Co-oper Bus Stud
6(1):165–176
Soong KK, Ahmed EM, Tan KS (2019) Factors influencing Malaysian small and medium enter-
prises adoption of electronic government procurement. J Publ Procure 20(1):38–61
Soong KK, Ahmed EM, Tan KS (2020) Factors affecting Malaysia’s SMEs in using public
electronic procurement. J Inf Knowl Manag 19(2):1–22
Soong KK, Ahmed EM, Tan KS (2021) Social influences’ effects on Malaysia’s SMEs’ public
electronic procurement usage. Int J Electr Govern Res 17(1):68–82
Suliantoro H, Ghozali I, Wibowo MA (2015) E-procurement adoption in government institution:
predicting social values effect on intention and usage behavior of e-procurement. Int J Bus Soc
16(2):167–184
Taluka EA (2016) Understanding determinants influencing consumer behavioral intention to use
mobile payment services in rural Tanzania: a case of Pwani region. Master dissertation,
University of Dar es Salaam, Tanzania
Tutu SO, Kissi E, Osei-Tutu E, Desmond A (2019) Evaluating critical determinants for the adoption
of e-procurement in Ghana. Int J Procure Manag 12(1):1–14
United Republic of Tanzania URT (2018) Public Procurement Regulatory Authority Annual
performance evaluation report for 2017/2018 financial year
Determinants of Behavioral Intention to Use E-Procurement System. . . 555

Venkatesh V, Morris MG, Davis GB, Davis FD (2003) Users’ acceptance of information technol-
ogy: toward a unified view. MIS Q 27(3):287–294
Venkatesh V, Thong JYL, Xu X (2016) Unified theory of acceptance and use of technology: a
synthesis and the road ahead. J Assoc Inf Syst 17(5):328–376
Waithaka RK, Kimani JG (2021a) Determinants of adoption of e-procurement practices: a critique
of literature review. Glob J Purchas Procure Manag 1(1):22–31
Waithaka RK, Kimani JG (2021b) Effect of e-procurement practices on supply chain performance.
Glob J Purchas Procure Manag 1(1):32–42
Watuleke J (2017) E-procurement: evolution and adoption. A review of literature. IJRDO-J Educ
Res 2(5):50–62
Willy TK, Paul M (2021) Integrated financial management information system implementation on
procurement performance of state-owned enterprises in Kenya. Int Res J Publish 2(2):298–314
Yalew S, Solomon B (2019) The effect of procurement planning on public procurement perfor-
mance in Amahara Region. Master Thesis, College of Business and Economics, Bahir Dar
University, Ethiopia
Yuliawati NM, Icih I, Kurniawan A (2021) The effect of e-procurement system implementation,
competence and compensation on employee performance of goods/services procurement.
Account Res J Sutaatmadja 5(1):103–111
Receptivity of Eastern and Southern
African English-Speaking Countries’
Executives to Use, Diffuse and Adopt
Humanitarian Logistics Digital Business
Ecosystem

Benjamin Ohene Kwapong Baffoe and Wenping Luo

1 Introduction

The global rise of the African continent and the growing dependence on advanced
digital technology are encouraging its humanitarian and economic self-sustenance
(African Union Commission 2019). Sustaining its humanitarian and economic self-
sustenance, the African Union (AU) implemented sustainable and disaster risk
prevention programs aiming at minimising the economic impact and loss of
human lives by its “Programme of Action for the Implementation of the Sendai
Framework 2015-2030 in Africa” tailored towards the prevention of loss of human
lives, property and businesses serving economies (African Union Commission
2019). For example, it was estimated that an average loss of US$3.6 billion of
gross domestic product (GDP) in the West African region in 2015 was due to the
impact of the Ebola virus outbreak, which disrupted many logistics and supply
chains and loss of lives not only in West Africa but also in other parts of Africa
(UNPD 2015). Such overwhelming natural and man-made disaster impacts, affect

B. O. K. Baffoe (✉)
Institute of Logistics Science and Engineering, Shanghai Maritime University, Pudong District,
Shanghai, PR China
W. Luo
School of Economics & Management, Shanghai Maritime University, Pudong District,
Shanghai, PR China
e-mail: [email protected]

government and regional bodies ermersely that businesses, non-governmental orga-

nisations (NGOs), and philanthropist collectively join hands as a multi-stakeholder
to address the impact (Baffoe et al. 2022).
Multi-stakeholder collaborations with governments serve as essential partners in
disaster mitigation, response and prevention. Such collaborative support comes in
many forms; for example, financial support, expert technical know-how, transpor-
tation and logistics service, volunteerism, and others. Despite the support obtained
from these multi-stakeholder team during such cases, the complexity of disasters
response and associated logistical challenges faced by the them prompts for a more
mutual disaster support and response. As sometimes their capacities to respond is
overwhelmingly compromised (Jahre et al. 2016; Kovács and Spens 2007; Lai et al.
2018; Nurmala et al. 2018; Tofighi et al. 2016; Trunick 2005; Van Wassenhove
2006; Zokaee et al. 2016). Lack of expertise, lead time variability, resource and
capacity waste due to overhaul, inaccessibility of transport routes and infrastruc-
tures, and other challenges continue to prompt managers in the humanitarian and
business logistics fields to employ innovative ways in tackling them. Among such
innovative ways is the employment of technology.
Technological employment continues to change the dynamics of addressing such
issues in the logistics and humanitarian sector (Cichosz 2018). It has become
inevitable in the twenty-first century that humanitarian organisations and business
logistics companies and consumers’ interdependency on digital technology cannot
be done away with in addressing their challenges or needs (Cichosz 2018;
Capgemini Consulting 2019). Owing to the benefits obtained from using digital
technology in optimising their operation to be agile, effective, efficient, and lean,
using digital technology in their operations plays an essential role in saving lives and
maximising their profits (Adner 2006; de Camargo Fiorini et al. 2018; Dubey et al.
2018, 2019a, b; Govindan et al. 2018). Becoming apparent, collaboration under an
ecosystem among actors and users is facilitated by such technologies. Despite the
advantages of digital technology usage, it is apparent that it also poses disadvantages
or risks to users, such as platform dominance, safety and security (Lenkenhoff et al.
2018). Amidst the drawbacks, a managed and coordinated digital ecosystem plat-
form of actors serves the objectives of its users as now been encouraged to address
logistical and humanitarian challenges in Africa.
The African Union (AU) addressing such humanitarian challenges from hinder-
ing the continent growth has set up an African Humanitarian Agency (AfHA)
(Africa Union 2019), an agency mandated to see to the self-reliance and sustainable
response on humanitarian issues in the continent by 2025. Indicated in the imple-
mentation plan is the need to facilitate a good collaboration among different stake-
holders, creating a financial pool, staffing, and addressing its logistical challenges.
Nonetheless, concerns are raised on its achievement; thus, harnessing human
resource expertise, digital ecosystem platform employment by actors and others.
Addressing such concerns, some scholars have presented likely solutions that may
mitigate such challenges (Beamon and Balcik 2008; Kabra et al. 2017; Kovács and
Spens 2009; Baffoe and Luo 2020; Nurmala et al. 2017, 2018; Oloruntoba and Gray
2009; Thomas and Fritz 2006; Van Wassenhove 2006). Adding to another
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 559

collaborative platform by World Food Programme (WFP), Baffoe and Luo (2020,
2021) proposed a Humanitarian Logistics Digital Business Ecosystem (HLDBE), a
digital ecosystem platform of humanitarian and business logistics actors harnessing
their resources and expertise to meet and sustain their operations.
While digital technology utilisation and the collaboration of humanitarian orga-
nisations and business logistics companies under a digital business ecosystem like
HLDBE in English-speaking Eastern and Southern African (ESA) countries of the
African Union (AU) are encouraged for regional growth and development, the
propensity of prospective users (thus, senior-level managers) to use, diffuse and
adopt digital business ecosystem creates much controversy and interests (SADC
2012; EAC 1999–2000). Addressing such concerns, this study utilises technology
diffusion and innovation theories to examine the propensity of executives working in
the humanitarian organisations and business logistics companies to use, diffuse and
adopt HLDBE as another future operation tool in the ESA countries. However,
works of literature have not examined what factors would inﬂuence prospective
humanitarian organisations and business logistics companies’ decision-makers’
behaviour to use, diffuse and adopt HLDBE in current ESA environment using
PLS-SEM multivariate analysis.

2 Deﬁnition and Some Relevant Theories and Framework

2.1 Humanitarian Logistics Digital Business Ecosystem

According to Baffoe and Luo’s (2020) study, the HLDBE platform serves as another
essential tool for humanitarian and business logistics collaborative operations. Under
the HLDBE platform, humanitarian organisations, business logistics companies and
government agencies serve as users of the digital ecosystem. For the definition of
HLDBE, see Baffoe and Luo (2020). According to Heskett (1977), the success or
failure of a business can be identified from the employment of logistics. These
successes and failures are sometimes reflected from the organisation/companies’
strategic, tactical and operational point of view, which either help them satisfy
customer’s needs, make profit, obtain market share, stay competitive, or even save
lives as is the case of disaster relief response. According to Handfield et al. (2013),
logistics managers are routinely faced with meeting customer services, adaptation to
network thinking (thus collaborative thinking) among actors in the network, sus-
tainability issues, inability to estimate logistics costs and lack of transportation
infrastructure aiding their operations, and others are pressing issues prompting the
investment of new technologies, human resources and reassessment of logistics
strategy. In the advent of globalisation, the advancement of technologies, and high
demand of customer cost reduction and preferences, managers continue to look for
innovative solutions that would help them address the challenges they face via an
agile, lean, efficient, and cost-effective ways. Concurrently, Baffoe and Luo (2020)
proposed and stressed on the need for a flexible, collaborative and technology-driven
560 B. O. K. Baffoe and W. Luo

system that seeks to address issues of this nature as indicated in the HLDBE digital
ecosystem platform. Supportably, notable scholars have stressed on the need for a
collaborative effort of diverse users via harnessing their expertise and resources
under a digital ecosystem platform (Jahre et al. 2016; Kovács and Spens 2007; Lai
et al. 2018; Nurmala et al. 2018; Toﬁghi et al. 2016; Trunick 2005; Van Wassenhove
2006; Zokaee et al. 2016). Although the HLDBE is yet to materialise as proposed by
Baffoe and Luo (2020), examining the perceived interest of its usage by the
prospective users via the technology of diffusion theories is essential to its imple-
mentation, especially in the African continent.

2.2 Supporting Technology of Diffusion and Adoption

Theories

Technology diffusion and adoption theories aid in studying and predicting the
perceived interest and spread of technology innovation in a social system. Such
theories may be grouped as diffusion theories, personality theories, organisation
structure theories, decision-making theories, and user acceptance theories as posited
by Hillmer (2009). Notable among such theories are diffusion of innovation (DOI),
technology–Organisation–Environment (TOE) framework, theory of reasoned
action (TRA), theory of planned behaviour (TPB) and others (Ajzen 1991; Ajzen
and Fishbein 1973; Bower and Christensen 1995, p. 1942; Compeau and Higgins
1995; Daft and Lengel 1983; Davis 1989; Hillmer 2009; Moore 1995; Rogers 1962;
Schumpeter 1942; Tornatzky and Fleischer 1990; Venkatesh et al. 2003). All are
acting as theories used in examining the diffusion and adoption of technology
innovation applied from the practical and academic perspectives (Hillmer 2009).
Noted by Hillmer, economists, psychologist, marketers and other users usually apply
them in their work ﬁeld for ascertaining human behaviour and intention of diffusing
and adopting a phenomenon. The DOI is applied commonly in information technol-
ogy industries, and the Uniﬁed Theory of Acceptance and Use of Technology
(UTAUT) is usually applied for academic purposes (Hillmer 2009; MacVaugh and
Schiavone 2010). Although all of these theories have several advantages and
disadvantages, we employ DOI, TOE, institutional theory (Inst.T) to examine the
senior-level manager’s perceived interest to diffuse and adopt the HLDBE in their
operations in English-speaking Eastern and Southern African countries. Below is the
explanation of the supporting theories for this study.

2.2.1 Diffusion of Innovation Theory

Diffusion of innovation (DOI) theory is a widely used theory for technology

innovation diffusion and adoption. Proposed by Professor Rogers, DOI theory
employment aids decision-makers to understand the advantages and disadvantages
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 561

an innovation will bring to their operations (Rogers 1962). Predominately, applied in

the information technology fields, they are also applied in other areas such as
tobacco control for health issues, education, material flow cost accounting, fashion
design piracy, enterprise resource planning (ERP) system in an NGO and others,
making it an excellent practical theory for examining the perceived interest of users
on an innovation (Appel et al. 2018; Brewer 2016; Miranda et al. 2016; Valente et al.
2015). Adding to its benefits and usage, DOI aids innovation to be spread via a
decision-making process and its channel of communication in a social system
(Rogers 2003). According to Rogers, relative advantage, compatibility, complexity,
trialability and observability are attributes used to examine the diffusion of an
innovation. DOI’ attributes have been tested as significant for the diffusion of an
innovation. Owing to the relevance of DOI attributes, we employed perceived
relative advantage (PRA), perceived compatibility (PC) and perceived trialability
(PT) as were significantly observed in the study of Kendall et al. (2001), additionally
with perceived security and safety concerns (PSSC) variables to examine the
HLDBE in the context of this research’s scope.

2.2.2 Technology–Organisation–Environment Framework

Another diffusion and adoption framework was developed by DePietro et al. (1990).
Posited out of the DOI, the framework is noted for aiding stakeholders to access the
internal and external capabilities of their organisations/companies before their deci-
sion to diffusion/adoption or rejection of an innovation. Noted for the TOE’s
expansion based on the DOI theory, the addition of external attributes informs
managers’ decision on the propensity to use, diffuse and adopt a technology inno-
vation (Gangwar et al. 2015). Additionally, Gangwar et al. (2015) indicated that
managers understanding the organisations’ resources, structure and capabilities
coupled with the environment they operate in well inform their decision on diffusing
and adopting an innovation, thus helping decision-makers to perceive the advantages
and disadvantages of innovation such as the HLDBE before its implementation and
adoption. Due to it practical implication (Gangwar et al. 2015; Wang et al. 2007,
2010; Zhu 2004), the TOE was used to assess the perceived interest of senior-level
managers’ interest to use, diffuse and adopt the HLDBE from the internal (thus,
infrastructure and expertise [IE], donor/top-level management support [D_TLMS],
organisational size [Org_size]) and external (thus, normative pressure [NP] and law
and governance [LG]) point of views.

2.2.3 Institutional Theory

Institutional theory plays an essential role in examining innovation diffusion and

adoption from the economic, political and societal perspectives. A theory was
developed by DiMaggio and Powell (1983). According to DiMaggio and Powell
562 B. O. K. Baffoe and W. Luo

(1983), their theory helps managers to understand the structure and culture of an
organisation from either the micro- or macro-perspectives. As applied in the logistics
ﬁeld as found in Dubey et al. (2019b) and Tsai and Tang (2012), serving as an
essential element, the organisational culture and structure (SCS) dimension was used
to ascertain the internal phenomenon of an institution prior to its extension outside
the organisation’s domain. Signiﬁcantly tested and applied, its organisational culture
and structure (OCS) dimension is used for assessing senior-level managers’ per-
ceived interest to use, diffuse and adopt the HLDBE in the Eastern and Southern
African English-speaking countries.
The objectives of this study were to examine the propensity of senior executives
working in the humanitarian organisations and business logistics companies to use,
diffuse and adopt the HLDBE as a future operational tool when implemented; to
understand, from the opinions of users, the salient latent variables that inform their
behaviour to use, diffuse and adopt the HLDBE; and to identify essential indicators
that should get attention. Understanding from the decision-makers’ perspective, if
organisation size is a moderating factor, their perceived interest to use, diffuse and
adopt the HLDBE may have a non-linear effect.

3 Methodology

3.1 Study Design

The study was designed using a cross-sectional quantitative survey data with
instruments developed from the technology of diffusion theories (DOI, TOE and
Inst.T). Furthermore, the works of Lee (2015) and Tweel (2012) provided a guide for
instrument development.

3.2 Research Instrument

SurveyMonkey is an online web-based survey platform that is reliable and secure.

Developing, administering and collecting survey response were done through
SurveyMonkey on the Internet. Thus, a web-based survey was conducted. The
content validity, face validity, questionnaire instrument reliability and validity
were established and assessed using works of literature; experts in the humanitarian
logistics and business logistics ﬁelds. Also, a pre- and post-test assessment were
conducted to ensure that questions utilised were simple, free from respondent
burden, understandable, coherent and valid in terms of content before the question-
naire were administered.
The reliability of questionnaire items was based on the works of literature, all
measuring Diff_&_Adopt_HLDBE (Cameron et al. 2006; Lee 2015; Ouchi 1980).
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 563

3.3 Sampling, Measures and Data Collection

Eastern and Southern African population samples were obtained from the Global
Database and SurveyMonkey audience (Global Database 2019; SurveyMonkey
Audience 2020). The senior-level managers of a humanitarian organisation and
business logistics company with logistics focus were included in the analysis.
Obtaining the needed sample size from the population sample size of 2218, a
conﬁdence level of 95% and a margin of error 5% were set. Needed sample size for
the study was 328. After administering the survey, a total return rate was 3.69%
(82 of 2218), a non-response rate, and bounce mail rate was 96.30% (2136 of 2218).
Our observation of the low response rate was due to the present Coronavirus disease
(COVID-19) pandemic, changes in the employment of senior-level managers
resulting in inaccessibility of survey sent to their former e-mail addresses, and
retirement. Nonetheless, due to the low response rate, more data were obtained
using SurveyMonkey audience (thus, obtaining a total sample of 113 responses),
making it a total of 195 responses suitable for the analysis of the study. The survey
was administered from November 2019 to May 2020.

3.4 Data Screening and Pre-analysis

Microsoft Excel, IBM SPSS Version 24 and SmartPLS Version 3.2.9 software were
used for outliers, missing values and normality test data screening.
Accounting for multiple outlier check, influential outliers were identified from the
box plot analysis; further analysis was done with the Mahalanobis distance and
Studentised deleted residuals to augment the box plot results. Two outliers were
found and removed (Aguinis et al. 2013; Osborne 2013). Addressing the influential
outlier was hampered because all respondents were anonymous, posing a challenge
to probe further on the reason for the outliers. More so, missing values pattern
observed was non-monotone. The missing values were substituted with a median of
nearby point in IBM SPSS Version 24. Furthermore, normality issues were adjusted
for with SmartPLS bootstrapping (Hair et al. 2013).

3.5 Sample Assessment

Following the rule of thumb of partial least squares structural equation modelling
(PLS-SEM), sample size calculation can be obtained from a ten times rule using the
maximum number of the pointed arrow towards a speciﬁc latent variable (Hair et al.
2014). Nonetheless, Hair et al. (2014) reiterated the use of G*Power analysis for an
objective sample size calculation. Adhering to Hair et al.’s guide, we employed
α = 0.05, power (1 - β) = 0.85, f 2 = 0.15, and the number of predictors = 8; a
564 B. O. K. Baffoe and W. Luo

required sample size of 120 was needed, making our total sample size of 135 suitable
for our study (Faul et al. 2009). The measurements used were of the Likert scale. See
appendices for measurements used for the questionnaires. More so, a simple random
sampling method was used for respondents’ sampling.

3.6 Reason for Using PLS-SEM

PLS-SEM is a multivariate statistical tool for testing variance-based analysis and

serving as another useful tool for examining phenomenon with small sample size
(Hair et al. 2011, 2019; Wong 2019). The purpose of utilising PLS-SEM is that it can
perform both reﬂective and formative measurement approaches with objective
analysis in a non-parametric view. Furthermore, it is noted for its check for moder-
ation and quadratic effects in the analysis of the obtained data.

4 Data Analysis

4.1 Common-Method Variance Bias Test

Adherence to Podsakoff’s guidelines, Herman’s single factor test was conducted to

examine any common-method bias in the instruments variable (Podsakoff et al.
2003). Using IBM SPSS Version 24, a principal component analysis (PCA) with
oblimin rotation aided in identifying any sign of single factor for the analysis. The
generated output extracted eight different factors with 41 items of measurement
constructs and eight Eigenvalues rotations (ﬁrst Eigen rotation variance of 34.5%
and 71.9% for the eighth extracted cumulative eigenvalue rotation). The Kaiser–
Meyer–Olkin (KMO) score was 0.883, signiﬁcant Bartlett’s test score was 0.000,
and acceptable communalities scores for items were 0.491–0.846. All indicating that
there was no common-method bias observed.
Attributions of respondents are illustrated in Table 1. Concerning other senior-
level position for job level, respondents held positions such as Executive Manager
(2), Board Manager (1), Consultant in NGO industry (1), Executive Manager of
International Business (1), Team Leader (1), Head of the Department (1), Supervisor
(3), Analyst (1), Owner (2), Owner of International Transport Organisation (1) and
Senior Manager of International Business (1), with two respondents indicating as
Driver (1) and Buyer (1). Also, respondents indicated their company/organisation
type as private company providing services to local South African NGO market (1),
international transport logistics company (1), vehicle telematics (1) and construction/
transport company (1).
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 565

Table 1 Demographic characteristics of the respondents

Category Frequency Percent
Country
Botswana 1 0.7
Kenya 5 3.7
Malawi 1 0.7
South Africa 121 89.6
South Sudan 1 0.7
Zambia 5 3.7
Zimbabwe 1 0.7

Job level
President/Chief Executive Ofﬁcer 21 15.6
Senior Vice President/Vice President 4 3
C-Suite Level Executives 1 0.7
Executive Director/Director 18 13.3
Deputy/Associate/Assistant Director 4 3
Manager 49 36.3
Deputy/Assistant Manager 21 15.6
Other Senior-Level Position 17 12.6

Organisation/company type
Federal/Central Government International Organisation (IO) 9 6.7
Government Organisation (GO) 20 14.8
Religious Non-Government Organisation (RNGO) 6 4.4
International Non-Government Organisation (INGO) 13 9.6
National/Local Non-Government Organisation (N/L-NGO) 17 12.6
Logistics/Supply Chain Company (L/SC-C) 32 23.7
Transport/Shipping Company (T/S-C) 34 25.2
Others 4 3

Organisation/company area of focus

Food relief supply 24 17.8
Health relief supply 25 18.5
Livelihood relief supply 13 9.6
Shelter relief supply 7 5.2
Water and sanitation relief supply 19 14.1
Transport and logistics services 69 51.1
Transport and logistics consultation service 36 26.7

Organisation size
1–50 employees 35 25.9
51–100 employees 15 11.1
101–500 employees 17 12.6
501–1000 employees 18 13.3
(continued)
566 B. O. K. Baffoe and W. Luo

Table 1 (continued)
Category Frequency Percent
1001–5000 employees 23 17
5001–10,000 employees 11 8.1
More than 10,000 employees 16 11.9

4.2 Heterogeneity Assessment

Preventing business from an aggregated analysed result, accounting for heterogene-

ity testing is essential. Following the guidelines (Hair et al. 2016; Matthews et al.
2016), a finite mixture partial least squares (FIMIX-PLS) analysis was conducted.
With a maximum number of eight arrows pointing (indicator arrow to PRA con-
struct) a specific indicator or construct in the model, a minimum R2 of 0.25, and
α = 0.05, a calculated minimum sample size of 54 will be needed for performing the
FIMIX-PLS. In addition, a 2.5 (135 sample size/54 = 2.5) theoretical upper bound
for obtaining the required segment for the unobserved heterogeneity analysis was
recorded. Obtaining the exact segment fit indices, a FIMIX-PLS analysis was done.
See Table 2 for fit indices results. Adherence to fit indices checks’ rule of thumb
given by Hair et al. (2016), a good fit is achieved from the combinations AIC3
+ CAIC; AIC4 + BIC, using the minimum indices score in the total segment results
and using normed entropy (EN) as a check. Adherence to Sarstedt et al. (2022),
AIC3 and AIC4 direct to a three-segment fit indicating heterogeneity; nonetheless,
based on our small sample size, it was not suitable to proceed with the analyse for
heterogeneity. In this regard, we acknowledge heterogeneity issue exist, but due to
the sample size limitation, we were not able to do so (Matthews et al. 2016; Sarstedt
et al. 2022). We strongly advice scholars with a large sample size to proceed with the
heterogeneity assessment when it is observed.
Nonetheless, in adherence to the rule of thumb, we proceeded to analyse the
aggregated result as the FIMIX results indicate that the data are free from
heterogeneity bias.

4.3 Formative Model Assessment

As the rule of thumb for PLS-SEM multivariate analysis, a two-stage analysis is

predominately done. Thus, the assessment of measurement model was in Stage 1 and
structural model assessment in Stage 2 (Hair et al. 2016, 2019). Measurement model
analyses are done using either the formative measurement approach (exploratory
approach) or the reflective measurement approach (confirmative approach), or both
the approaches together.
This study focuses on the formative model approach to explore the perceived
interest of senior-level managers to diffuse and adopt HLDBE for their operations.
Adherence to the modus operandi, a redundancy analysis, variance inflation factor
(VIF), significance and relevance of the indicators need to be addressed before
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 567

Table 2 FIMIX-PLS segment check for heterogeneity

Fit indices Segment 1 Segment 2 Segment 3
Akaike information criterion (AIC) 422.194 383.238 348.803
Modified AIC with factor 3 (AIC3) 435.194 410.238 389.803
Modified AIC with factor 4 (AIC4) 448.194 437.238 430.803
Bayesian information criterion (BIC) 460.058 461.880 468.222
Consistent AIC (CAIC) 473.058 488.880 509.222
Hannan–Quinn criterion (HQ) 437.581 415.196 397.332
Minimum description length with factor 5 (MDL5) 715.516 992.446 1273.898
Log likelihood (LnL) -198.097 -164.619 -133.402
Normed entropy statistic (EN) 0.000 0.759 0.680
Non-fuzzy index (NFI) 0.000 0.779 0.665
Normalised entropy criterion (NEC) 0.000 32.824 43.554
Notes: Fit indices results obtained using the PLS algorithm settings (initial weights 1, 3000
maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings were used, path
weighting scheme), FIMIX segmentation settings (10,000 maximum iteration, 10 repetitions, 7 stop
criterion measuring segments 1, 2, and 3). Bolded figures are the minimum indices score

D_TLMS1_1
PADT1_1
0.436 0.478
D_TLMS2_1 0.482 0.858 0.737 0.574 PADT2_1

0.195 0.112
PT1_1
D_TLMS3_1 D_TLMS Dlff_&_Adopt_HLDBE

Fig. 1 Redundancy analysis for D_TLMS and Diff_&_Adopt_HLDBE

moving to report the structural model measurement (Hair et al. 2013). Following the
guidelines for formative measurement, redundancy analysis was performed on
D_TLMS and Diff_&_Adopt_HLDBE as the primary constructs with the results
showcased in Fig. 1. To prevent the model from any collinearity issues, any VIF
score > 5 poses as a problem. Our study indicated that all indicators used had no
collinearity issues (see Table 3 for details). Further analysis indicated that there were
some indicators not significant as found in Table 4. To address this issue, the outer
weights and outer loadings were used to eliminate unessential indicators that may
have an impact on the study and also to retain indicators (thus, indicators with >0.50
outer loading score) that have an absolute contribution to the study. As such,
indicators IE1_1, IE2_1, IE3_1 and OC1_1 were removed to ensure that all indica-
tors are significant and have a contribution to the study. Note and caution should be
taken that for formative measurement assessment only outer weights should be
reported, but with a non-significant weight, outer loading can be used to assist in
determining whether the affected indicator or item has an absolute or a relative
importance to the study. Then, another VIF and significant testing were done for the
new model after eliminating the indicators to ensure that there were no collinearity
issues in the new assessment (see Tables 5 and 6 for details). Meeting all the
requirements for measurement model on a formative measurement approach without
568 B. O. K. Baffoe and W. Luo

Fig. 2 The results of PLS algorithm model

any collinearity issues and obtaining the significance, we proceed to assess the
structural model. See Fig. 1 for more details.
To verify the structural model for PLS-SEM, Hair et al. (2013)’s guidelines were
followed; thus, we checked for any collinearity issues before testing for the model’s
path prediction, significance and hypothesis. The structural model possesses no
collinearity issues, as found in Table 7. We then checked for the model’s predictive
accuracy by measuring for the model’s predictive power (R2) and its predictive
accuracy (Q2) (Hair et al. 2011, 2019; Henseler et al. 2009). Following the guide-
post, the in-sample predictive power (R2) measured as weak (0.25), moderate (0.50)
and substantial (0.75), while the predictive accuracy’s 0 indicated as small, 0.25 as
medium and 0.50 as large (Hair et al. 2019; Stone 1974).
Using a post-hoc power analysis (N = 135, α = 0.05, f 2 = 0.15, number of
predictors = 8), we recorded a 89.9% statistical confidence of detecting relevant
effect with a substantial explanatory power of 74.9% (R2 = 0.749); we assessed the
significance and hypothesis support for the questions raised for the model, any
confidence intervals bias corrected signs with the same sign was classified as
significant and relevant in support of the hypothesis analysed. Table 8 illustrates
the results. In addition, accounting for the non-linearity, a quadratic effect was
modelled to ascertain whether senior-level managers see the perceived interest to
diffuse and adopt the HLDBE in their operation as a linear effect. The significance is
illustrated in Table 8. More so, using organisation size as a moderating factor for
HLDBE and D_TLMS, we assessed its significant effect as can also be found in
Table 8 (Becker et al. 2018).
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 569

Table 3 Collinearity assessment for formative indicators

Formative constructs Formative indicators VIF Collinearity if VIF > 5?
D_TLMS D_TLMS1_1 2.182 No
D_TLMS2_1 2.607 No
D_TLMS3_1 2.115 No
IE IE1_1 1.549 No
IE2_1 2.040 No
IE3_1 1.945 No
IE4_1 2.060 No
IE5_1 2.130 No
IE6_1 2.016 No
IE7_1 2.219 No
LG LG1_1 2.045 No
LG2_1 2.259 No
LG3_1 2.366 No
NP NP1_1 1.383 No
NP2_1 2.087 No
NP3_1 2.012 No
OCS OC_1 1.000 No
OS_1 1.000 No
PC PC1_1 3.247 No
PC2_1 4.054 No
PC3_1 2.621 No
PRA PRA1_1 4.570 No
PRA2_1 4.888 No
PRA3_1 3.397 No
PRA4_1 3.577 No
PRA5_1 4.554 No
PRA6_1 4.515 No
PRA7_1 4.540 No
PRA8_1 4.503 No
PSSC PSSC1_1 3.039 No
PSSC2_1 3.039 No
Diff_&_Adopt_HLDBE PT1_1 1.045 No
PADT1_1 1.968 No
PADT2_1 1.996 No
Org_Size 1.000 No
QE (D_TLMS) 1.000 No
D_TLMS × Org_Size 1.000 No
Notes: Results derived from the updated PLS algorithm testing settings (initial weights 1, 3000
maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings were used, path
weighting scheme). The bold part indicates VIF score <5 with no collinearity issues
Table 4 Signiﬁcant results for formative measurement indicators
570

Conﬁdence Signiﬁcant if When outer weights are not

intervals bias- confidence intervals significant, retain the outer
Formative indicators’ Outer Outer t- corrected bias-corrected signs weights if outer loadings >
Formative constructs path weights loadings value 2.5% 97.5% are same 0.50
D_TLMS D_TLMS1_1 → 0.385 0.887 5.068 0.231 0.530 Significant Retain
D_TLMS
D_TLMS2_1 → 0.500 0.937 5.540 0.327 0.684 Non-significant Do not retain
D_TLMS
D_TLMS3_1 → 0.230 0.824 3.232 0.082 0.362 Significant Retain
D_TLMS
IE IE1_1 → IE 0.006 0.336 0.023 -0.492 0.447 Non-significant Do not retain
IE2_1 → IE -0.147 0.466 0.560 -0.679 0.339 Non-significant Do not retain
IE3_1 → IE -0.194 0.483 0.880 -0.638 0.231 Non-significant Do not retain
IE4_1 → IE 0.180 0.649 0.820 -0.266 0.590 Non-significant Do not retain
IE5_1 → IE 0.377 0.802 2.002 0.034 0.785 Significant Retain
IE6_1 → IE 0.194 0.746 0.817 -0.240 0.696 Non-significant Do not retain
IE7_1 → IE 0.648 0.920 3.158 0.284 1.039 Significant Retain
LG LG1_1 → LG 0.668 0.940 2.536 0.157 1.185 Significant Retain
LG2_1 → LG -0.096 0.692 0.363 -0.645 0.402 Non-significant Do not retain
LG3_1 → LG 0.500 0.879 2.004 -0.004 0.978 Non-significant Do not retain
NP NP1_1 → NP 0.275 0.707 3.854 0.136 0.416 Significant Retain
NP2_1 → NP 0.656 0.950 7.017 0.452 0.822 Significant Retain
NP3_1 → NP 0.225 0.810 2.163 0.027 0.434 Significant Retain
OCS OC_1 ← OCS -0.657 -0.648 1.036 -1.014 -0.574 Non-significant Do not retain
OS_1 ← OCS 0.762 0.754 1.205 -0.700 1.004 Non-significant Do not retain
PC PC1_1 → PC 0.171 0.868 1.015 -0.174 0.500 Non-significant Do not retain
PC2_1 → PC 0.456 0.948 2.357 0.088 0.855 Significant Retain
B. O. K. Baffoe and W. Luo
PC3_1 → PC 0.451 0.927 3.706 0.206 0.688 Significant Retain
PRA PRA1_1 → PRA -0.079 0.819 0.289 -0.644 0.427 Non-significant Do not retain
PRA2_1 → PRA 0.078 0.857 0.310 -0.372 0.606 Non-significant Do not retain
PRA3_1 → PRA -0.094 0.765 0.501 -0.473 0.268 Non-significant Do not retain
PRA4_1 → PRA 0.355 0.906 1.510 -0.051 0.872 Non-significant Do not retain
PRA5_1 → PRA 0.030 0.858 0.123 -0.528 0.443 Non-significant Do not retain
PRA6_1 → PRA 0.225 0.880 0.971 -0.229 0.683 Non-significant Do not retain
PRA7_1 → PRA 0.178 0.903 0.860 -0.255 0.553 Non-significant Do not retain
PRA8_1 → PRA 0.389 0.936 1.793 -0.029 0.824 Non-significant Do not retain
PSSC PSSC1_1→ PSSC 0.863 0.996 2.927 0.184 1.355 Significant Retain
PSSC2_1 → PSSC 0.162 0.869 0.484 -0.468 0.843 Significant Retain
Diff_&_Adopt_HLDBE PT1_1 → 0.116 0.313 2.299 0.015 0.215 Significant Retain
DIff_&_Adopt_HLDBE
PADT1_1 → 0.489 0.903 4.894 0.291 0.681 Significant Retain
DIff_&_Adopt_HLDBE
PADT2_1 → 0.562 0.929 5.640 0.353 0.742 Significant Retain
DIff_&_Adopt_HLDBE
Org_Size ← Org_Size 1.000 1.000 0.000 1.000 1.000 Significant Retain
a
QE (D_TLMS) → QE 1.000 1.000 1.000 1.000 Significant Retain
(D_TLMS)
a
D_TLMS × Org_Size 1.000 1.000 1.000 1.000 Significant Retain
→ D_TLMS ×
Org_Size
Receptivity of Eastern and Southern African English-Speaking Countries’. . .

Notes: Results from the PLS algorithm testing settings (initial weights 1, 3000 maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings
were used, path weighting scheme); for bootstrapping testing setting complexity, percentile bootstrap confidence interval method, parallel processing, 10,000
samples, fixed seed, significance of 0.05 and a two-tailed test type were used. The bold part indicates same confidence intervals bias-corrected signs and a
significant path
a
Presence of quadratic effect indicator and moderator effect. Because it only serves as a quadratic effect, it does not have a standard deviation that is used to
calculate its t-statistics value
571
572 B. O. K. Baffoe and W. Luo

Table 5 New collinearity assessment for formative indicators

Formative constructs Formative indicators VIF Collinearity if VIF > 5?
D_TLMS D_TLMS1_1 2.182 No
D_TLMS2_1 2.607 No
D_TLMS3_1 2.115 No
IE IE4_1 1.780 No
IE5_1 1.890 No
IE6_1 2.011 No
IE7_1 2.057 No
LG LG1_1 2.045 No
LG2_1 2.259 No
LG3_1 2.366 No
NP NP1_1 1.383 No
NP2_1 2.087 No
NP3_1 2.012 No
OCS OS_1 1.000 No
PC PC1_1 3.247 No
PC2_1 4.054 No
PC3_1 2.621 No
PRA PRA1_1 4.570 No
PRA2_1 4.888 No
PRA3_1 3.397 No
PRA4_1 3.577 No
PRA5_1 4.554 No
PRA6_1 4.515 No
PRA7_1 4.540 No
PRA8_1 4.503 No
PSSC PSSC1_1 3.039 No
PSSC2_1 3.039 No
Diff_&_Adopt_HLDBE PT1_1 1.045 No
PADT1_1 1.968 No
PADT2_1 1.996 No
Org_Size 1.000 No
QE (D_TLMS) 1.000 No
D_TLMS × Org_Size 1.000 No
Notes: Results derived from the updated PLS algorithm testing settings (initial weights 1, 3000
maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings were used, path
weighting scheme). The bold part indicates VIF score <5 with no collinearity issues

After fulfilling the assessment requirements for PLS-SEM for this study (thus,
checking for the measurement and structural model assessment), an important–
performance measurement analysis (IPMA) was done to identify the importance of
variables that decision-makers need to focus on to aid them in their decisions of
diffusion and adoption. Following the rule of thumb outlined by Ringle and Sarstedt
(2016) and Hair et al. (2019), the IPMA results are displayed in Table 9.
Table 6 New significant results for formative measurement indicators
Confidence Significant if When outer weights are not
intervals bias- confidence intervals significant, retain the outer
Formative indicators Outer Outer t- corrected bias-corrected signs weights if outer loadings >
Formative constructs path weights loadings value 2.5% 97.5% are same 0.50
D_TLMS D_TLMS1_1 → 0.397 0.891 5.330 0.252 0.545 Significant Retain
D_TLMS
D_TLMS2_1 → 0.489 0.934 5.430 0.310 0.662 Significant Retain
D_TLMS
D_TLMS3_1 → 0.230 0.824 3.266 0.089 0.365 Significant Retain
D_TLMS
IE IE4_1 → IE 0.094 0.662 0.404 -0.406 0.491 Non-significant Retain
IE5_1 → IE 0.297 0.818 1.533 -0.102 0.665 Non-significant Retain
IE6_1 → IE 0.180 0.759 0.726 -0.281 0.699 Non-significant Retain
IE7_1 → IE 0.594 0.940 2.853 0.157 0.970 Significant Retain
LG LG1_1 → LG 0.666 0.939 2.524 0.153 1.183 Significant Retain
LG2_1 → LG -0.099 0.691 0.374 -0.650 0.398 Non-significant Retain
LG3_1 → LG 0.504 0.880 2.023 -0.001 0.982 Significant Retain
NP NP1_1 → NP 0.276 0.708 3.889 0.139 0.419 Significant Retain
NP2_1 → NP 0.654 0.949 7.019 0.450 0.820 Significant Retain
NP3_1 → NP 0.227 0.811 2.185 0.029 0.434 Significant Retain
OCS OS_1 ← OCS 1.000 1.000 0.000 1.000 1.000 Significant Retain
Receptivity of Eastern and Southern African English-Speaking Countries’. . .

PC PC1_1 → PC 0.171 0.868 1.010 -0.174 0.500 Non-signiﬁcant Retain

PC2_1 → PC 0.453 0.948 2.333 0.084 0.852 Significant Retain
PC3_1 → PC 0.454 0.928 3.721 0.210 0.692 Significant Retain
PRA PRA1_1 → PRA -0.075 0.821 0.275 -0.640 0.435 Non-significant Retain
PRA2_1 → PRA 0.076 0.858 0.305 -0.373 0.602 Non-significant Retain
PRA3_1 → PRA -0.095 0.766 0.501 -0.474 0.266 Non-significant Retain
573

(continued)
Table 6 (continued)
574

Conﬁdence Signiﬁcant if When outer weights are not

intervals bias- confidence intervals significant, retain the outer
Formative indicators Outer Outer t- corrected bias-corrected signs weights if outer loadings >
Formative constructs path weights loadings value 2.5% 97.5% are same 0.50
PRA4_1 → PRA 0.353 0.906 1.505 -0.053 0.868 Non-significant Retain
PRA5_1 → PRA 0.033 0.859 0.137 -0.520 0.448 Non-significant Retain
PRA6_1 → PRA 0.227 0.881 0.983 -0.228 0.683 Non-significant Retain
PRA7_1 → PRA 0.177 0.903 0.854 -0.260 0.551 Non-significant Retain
PRA8_1 → PRA 0.384 0.936 1.772 -0.035 0.819 Non-significant Retain
PSSC PSSC1_1→ PSSC 0.856 0.995 2.904 0.174 1.348 Significant Retain
PSSC2_1 → PSSC 0.169 0.871 0.509 -0.461 0.853 Significant Retain
Diff_&_Adopt_HLDBE PT1_1 → 0.116 0.314 2.314 0.015 0.216 Significant Retain
DIff_&_Adopt_HLDBE
PADT1_1 → 0.487 0.902 4.887 0.287 0.678 Significant Retain
DIff_&_Adopt_HLDBE
PADT2_1 → 0.564 0.929 5.688 0.360 0.744 Significant Retain
DIff_&_Adopt_HLDBE
Org_Size ← Org_Size 1.000 1.000 0.000 1.000 1.000 Significant Retain
a
QE (D_TLMS) → QE 1.000 1.000 1.000 1.000 Significant Retain
(D_TLMS)
a
D_TLMS × Org_Size 1.000 1.000 1.000 1.000 Significant Retain
→ D_TLMS ×
Org_Size
Notes: Results from the PLS algorithm testing settings (initial weights 1, 3000 maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings
were used, path weighting scheme); bootstrapping testing setting using (most important (faster) complexity, percentile bootstrap confidence interval method,
parallel processing, 10,000 samples, fixed seed, significance of 0.05 and a two-tailed test type. The bold part indicates same confidence intervals bias-corrected
signs and a significant path
a
Presence of quadratic effect indicator and moderator effect. Because it only serves as a quadratic effect, it does not have a standard deviation that is used to
calculate its t-statistics value
B. O. K. Baffoe and W. Luo
Table 7 Collinearity assessment for structure model constructs (collinearity if VIF > 5?)
QE D_TLMS ×
Formative constructs DIff_&_Adopt_HLDBE D_TLMS E LG NP OCS Org_Size PC PRA PSSC (D_TLMS) Org_Size
DIff_&_Adopt_HLDBE
D_TLMS 1.367
IE 2.050
LG 2.009
NP 1.663
OCS 1.082
Org_Size 1.019
PC 1.833
PRA 1.709
PSSC 1.791
QE (D_TLMS) 1.392
D_TLMS × Org_Size 1.052
Notes: Results derived from the updated PLS algorithm testing settings (initial weights 1, 3000 maximum iterations, 7 stop criterion, results standardised, no
Receptivity of Eastern and Southern African English-Speaking Countries’. . .

Lohoeller settings were used, path weighting scheme)

575
576 B. O. K. Baffoe and W. Luo

Table 8 Signiﬁcance testing results for structural path model coefﬁcients

Confidence Significant if confidence
intervals bias- intervals bias-corrected
Formative constructs Path corrected signs are same and p-
path coefficient t-value 2.5% 97.5% value <0.05
D_TLMS → 0.791 17.900 0.680 0.862 Significant
DIff_&_Adopt_HLDBE
IE → D_TLMS -0.074 1.110 -0.249 0.024 Non-significant
LG → D_TLMS 0.115 1.993 0.014 0.238 Significant
NP → D_TLMS 0.501 5.626 0.345 0.674 Significant
OCS → D_TLMS 0.100 2.305 0.028 0.198 Significant
Org_Size → 0.032 0.720 -0.056 0.116 Non-significant
DIff_&_Adopt_HLDBE
PC → D_TLMS 0.286 3.174 0.152 0.523 Significant
PRA → D_TLMS 0.185 2.743 0.048 0.302 Significant
PSSC → D_TLMS 0.044 0.862 -0.068 0.139 Non-significant
QE (D_TLMS) → -0.078 2.195 -0.156 -0.017 Significant
DIff_&_Adopt_HLDBE
D_TLMS × Org_Size -0.004 0.078 -0.097 0.094 Non-significant
→
DIff_&_Adopt_HLDBE
Notes: Results from the PLS algorithm testing settings (initial weights 1, 3000 maximum iterations,
7 stop criterion, results standardised, no Lohoeller settings were used, path weighting scheme);
bootstrapping testing setting using (most important (faster) complexity, percentile bootstrap confi-
dence interval method, parallel processing, 10,000 samples, fixed seed, significance of 0.05 and a
two-tailed test type. The bold part indicates same confidence intervals bias-corrected signs and a
significant path

Table 9 Index and total effects for Diff_&_Adopt_HLDBE

Formative Importance–performance map Construct performances for
constructs [DIff_&_Adopt_HLDBE] (constructs) [DIff_&_Adopt_HLDBE]
D_TLMS 0.791 71.377
IE -0.058 84.575
LG 0.091 81.294
NP 0.397 64.327
OCS 0.079 40.441
Org_Size 0.032 42.647
PC 0.226 68.355
PRA 0.147 73.608
PSSC 0.035 88.079
Notes: Results were derived from testing using initial weights 1, 3000 maximum iterations, 7 stop
criterion, results standardised, no Lohoeller settings were used, path weighting scheme. The bold
part indicates the highest values for the importance and performance values
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 577

According to the instructions of Hair et al. (2019), the purpose of PLS-SEM

assessment is to predict target constructs, thereby eliminating the notion of checking
for a goodness-of-ﬁt test for the model. Alternatively, a good predictive measure-
ment could be checked by using the root mean square error (RMSE) and mean
absolute error (MAE) as found in the PLSpredict function in SmartPLS; it was
assessing the out-sample predictive measurement (Shmueli et al. 2019). Our study
recorded a high predictive accuracy power for all the target constructs’ formative
factors (see Table 10 for detailed results).

5 Discussions

Several studies have been conducted on humanitarian organisation and business

logistics collaborations. Additionally, technology diffusion and adoption theories
(DOI and TOE) with Inst.T have also been used to ascertain senior-level managers’
perceived interest to diffuse, adopt and utilise an innovation. However, few studies
have applied these theories to examine the propensity of executives from Eastern and
Southern African English-speaking countries to use, diffuse and adopt the Human-
itarian Logistics Digital Business Ecosystem (HLDBE) as proposed by Baffoe and
Luo (2020) for their future operations, using objective cross-sectional quantitative
study data with technology diffusion and adoption theories and PLS-SEM analysis.

5.1 Perceived Relative Advantage

Perceived relative advantage (PRA) is the benefits derived from an innovation over
previously used technology/innovations (Rogers 2003). PRA of an innovation is
thought to inform executives in the humanitarian organisation or business logistics
company to reduce costs, eliminate waste, improve service quality, be competitive,
and others via the advantage they derived from an innovation in their challenging
and dynamic environments (Al-Mamary et al. 2015; Arshad et al. 2018;
Bhattacharya et al. 2019; Tarofder et al. 2017; Tashkandi and Al-Jabri 2015). In
general, an executive choice of diffusing, and adopting an innovation such as the
HLDBE, is based on the value it brings to the organisation/company. In this present
study, PRA was recorded as a significant variable that influences senior-level
management’s decision to use, diffuse and adopt the HLDBE in their future opera-
tions. This clearly indicates that managers’ employment of digital technology to
improve their operations in Africa is taken shape with governmental–private invest-
ment in ecosystem platforms as is the case of the HLDBE for their operations serving
as positive, sustainable tools that help to address their challenges in the Eastern and
Southern African regions (Ndung’u and Signé 2020). Thus, a collaborative ecosys-
tem with digital technologies such as Big Data analytics, cloud computing, transport
routing/inventory scheduling tools and others under a shared resource and digital
578

Table 10 PLSpredict result summary for model’s key endogenous constructs

Formative indicators Q2predict PLS-SEM_RMSE PLS-SEM_MAE LM_RMSE LM_MAE Predictive power
PADT1_1 0.467 0.845 0.671 0.970 0.757 High
PADT2_1 0.495 0.692 0.540 0.797 0.615 High
PT1_1 0.070 1.332 1.173 1.531 1.276 High
D_TLMS1_1 0.520 0.721 0.559 0.803 0.631 High
D_TLMS2_1 0.612 0.680 0.526 0.738 0.548 High
D_TLMS3_1 0.491 0.861 0.658 0.922 0.709 High
Notes: Results were derived from testing using initial weights 1, 3000 maximum iterations, 7 stop criterion, results standardised, no Lohoeller settings were
used, path weighting scheme, 10 number of repetitions, 10 number of folds and ﬁxed seed. LM = Linear regression Model
B. O. K. Baffoe and W. Luo
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 579

expertise platform (HLDBE) are things the region looks for in revamping their
organisation’s businesses and economies. Additionally, the growing increase in
investment, high younger generations, development prospect and the willingness
to be self-reliant also motivate such action. The results also support other research on
its importance for decision-makers (African Union Commission 2019; Kim and Chai
2017; Ndung’u and Signé 2020; Vagnani and Volpe 2017; Baffoe et al. 2023).

5.2 Perceived Compatibility

Perceived compatibility (PC) is a factor identifying an innovation that is consistent to

user’s organisations’ previous values (Rogers 2003), an inclination to how
favourable the innovation will affect the operations, culture, missions and vision in
ascertaining their set objectives. The findings from this study also support that PC is
an essential influencing factor to senior-level managers’ perceived interest to diffuse
and adopt the HLDBE. As the African continent seeks to be self-reliance on
addressing its disaster issues, organisations and companies’ alignment of its objec-
tives affirms that a collaborative venture using digital technology that meets their
various objectives is welcomed as indicated in these findings (Africa Union 2019;
African Union Commission 2019). Also, affirming Rogers and others’ studies,
decision-makers’ assertion of an innovation compatible with its objectives is essen-
tial when deciding to diffuse and adopt an innovation.

5.3 Infrastructure and Expertise

Tangible and intangible resources are a supporting tool for technology innovations.
Cable networks, Internet broadband and satellite connections, mobile phone plat-
forms and social media platforms all serving as supporting tangle resources of
organisations and firms rely on to network with others and expertise as intangible
resources. Despite latent variable for infrastructure and expertise recording a
non-significant effect on executives of humanitarian organisations and business
logistics companies’ decision to use, diffuse and adopt HLDBE, its indicator vari-
able for ICT and data science expertise to man the HLDBE was significant. This
indicates that Africa’s investment in digital technology needs more efforts to
improve the development and collaborative efforts of industries found in
low-income economies under a digital system with supporting infrastructures to
aid in achieving the sustainable development goals for the region (Eastern and
Southern Africa) and continent (Songwe 2019). As useful resources can be harness
from collaborative entities in a digital platform, transactions will be fast-tracked
quickly, thereby encouraging humanitarian and business logistics entities to meet the
services of consumers and beneficiaries.
580 B. O. K. Baffoe and W. Luo

5.4 Normative Pressure

Normative pressures are industrial pressures that inﬂuence decision-makers on the

likelihood of utilising innovation. As noted by DiMaggio and Powell (1983), the
influence of industry associations or organisations one belongs may pressure one to
comply with their norms. Supporting this view, the results affirm that executives use,
diffusing and adoption of the HLDBE may fall in compliance with the various
associations they belong to. Influencing associations such as Shippers Council of
Eastern Africa (SCEA n.d.), the Chartered Institute of Logistics and Transport (CILT
n.d.), the South African Association of Freight Forwarders (SAAFF n.d.), Truckers
Association of South Africa (TASA n.d.), the South African National NGO Coali-
tion (SANGOCO n.d.) and others. Normative pressure, an influencing factor, also
supports other studies such as Awa et al. (2017).

5.5 Laws and Governance

Domestic, national and international laws governing humanitarian organisations and

business logistics companies are an influencing factor for an innovation employ-
ment. Laws and governance of humanitarian, institutional and corporate sector are
important dimensions that influence users’ opinions in employing a new technology
innovation such as the HLDBE (Beamon and Balcik 2008; UN and World Economic
Forum 2008; Nurmala et al. 2017; OCHA 2012). The analysis of our results
indicates a significant relation of LG to executive’s propensity to use, diffuse and
adopt the HLDBE in ESA. The different domestic and international laws on corpo-
rate and humanitarian activities of various countries with different culture, language
and others when aligned can fully foster much understanding and collaboration of
hybrid humanitarian and business logistics entities in the ESA. The perceived
interest to employ the HLDBE by donor/top-level management of humanitarian-
business logistics sectors requires consented laws, principles and governance of the
African Union and its regional blocks in harnessing the ecosystem platform inno-
vations. More so, more research is needed to foster a good dynamic, collaborative
digital ecosystem for Africa and its various regional bodies. Additionally, this
supports Ndung’u and Signé study (2020) on the need to improve legal institutions
and regulations to foster investment growth for new innovations such as the HLDBE
by national and international bodies in Africa. Noted from the result, if AU’s 2025
and 2065 goals of self-reliance in humanitarian and business logistics issues are to be
achieved such laws and policies should be looked at (Africa Union 2019; African
Union Commission 2019). Harmonising laws and governance is an essential factor
for HLDBE implementation and African humanitarian and business logistics com-
panies’ development (AUC 2019).
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 581

5.6 Organisational Culture and Structure

Organisational culture and structure are factors that aid the operational functions of
organisational performance and development (Dubey et al. 2019a). Integration of
different actors maybe achieved with conﬂicting merging structure under a collab-
orative system. This study’s results indicated that OCS may inﬂuence the executives
to use, diffuse and adopt the HLDBE for their future operations. Considering that a
unique structure of the proposed HLDBE for Africa Union maybe needed to
integrated countries in the continent, business and humanitarian leaders in various
sectors continue to work with the AU in moving towards a single agenda using
digital ecosystem platform to aid the continents to tap into new ventures, simply
business operations and information sharing; to encourage collaborations and indus-
trial development; and to address humanitarian issues by lives’ saving and sustaining
economic development for the coming generations of Africans as captioned in the
Agendas 2030 and 2063 (Africa Union 2019; African Union Commission 2019; AU
Economic Affairs 2019).

5.7 Perceived Safety and Security Concerns

Smooth running digital ecosystem platform with value-driven functions is encour-

aged when perceived safety and security concerns are addressed (Nurmala et al.
2017). The events of digitisation encouraging more networking of users under
platforms, cybersecurity and data safety and security are stressing issues experts
and industrial organisations (humanitarian and business logistics companies) are
concerned about in the African continent as the diffusion and adoption of digital
technology continue to increase (Ndung’u and Signé 2020). We also notice that
PSSC had a non-signiﬁcant effect on senior-level management’s decision to use,
diffuse and adopt the HLDBE as their future operation tool. Nonetheless, the
indicator for PSSC1_1 recorded a signiﬁcant effect, indicating that both donors
and top-level managements who dubbed as prospective users of the HLDBE are
concerns about transparency in the digital ecosystem. This is not new as the ESA
region and the African continent keep working on building institutions to improve
transparency and minimising and eliminating corruption.

5.8 Donor/Top-Level Management Support

Donor/top-level management support is a crucial factor in innovation usage, diffu-

sion and adoption (DePietro et al. 1990). Reiterated by Lian et al. (2014), senior
management’s propensity to employ an innovation like HLDBE is based on the
individual experiences and internal and external factors surrounding their
582 B. O. K. Baffoe and W. Luo

organisations and companies. Allocation of resources, improving and recruiting of

new human resources, benefits, risks and other factors inform their decision to use,
diffuse and adopt an innovation. The result of our study indicates that indeed
executives’ propensity to employ the HLDBE was significant for their future
operational goals. Translating to the ESA region and African continent, catching
up with development in the digital economy era, there is a need for collaboration,
capacity sharing, and more influential institutions that can foster the growth and
sustainability of logistics business and humanitarian logistics organisations in the
region and continent. Managers’ importance and influence in innovation use, diffu-
sion, and adoption are supported in works of Al-Mamary et al. (2015) and
Bhattacharya et al. (2019).

5.9 Moderator and Quadratic Effects

Accounting for quadratic effect, this study had a signiﬁcant effect, indicating that the
use, diffusion and adoption of the HLDBE in the future by executives of humani-
tarian organisations and business logistics companies would be non-linear in the
future. This may be because when the organisations and companies in the ESA
region get to the technological advancement and integration in the HLDBE platform,
members may set up their digital ecosystem systems, especially with business
logistics companies as compared to humanitarian logistics organisations. In addition,
we observed that organisation size was not a signiﬁcant moderator for the users. As
the African Union, private, donor and government entities work towards digital
economic era with an affordable and integrated system, the size of an organisation
may not matter much with respect to the use, diffusion and adoption of an innovation
like the HLDBE.

6 Implication of Study

The present research has signiﬁcant implication for Eastern and Southern African
English-speaking countries and the entire African continent as more concerted effort
is made in integrating organisation and companies to meet the humanitarian, busi-
ness, human resource and economic growth needs. The results’ support from the
executives of humanitarian organisation and business logistics companies indicates
that Africa’s planned Agenda 2030 and 2063 is taking shape as more support has
been given. Clearly, policymaker’s understandings of the important factors of users
of the HLDBE will help in implementing policies to spearhead the development and
growth to meet the continent agenda.
More so, diffusion of innovation (DOI) theory, technology–organisation–envi-
ronment (TOE) framework and institutional theory also serve as exploratory theories
to predict the use, diffusion and adoption of the HLDBE in English-speaking Eastern
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 583

and Southern African countries by the prospective HLDBE users. The application of
DOI, TOE and Inst.T with PLS-SEM serves as a beginning path for researchers to
probe more into the use, diffusion and adoption of the HLDBE in the African
continent before its implementation. Theoretically, diffusion of innovation theory
and technology–organisation–environment framework variables having a signiﬁcant
effect in this study helps to contribute to the humanitarian logistics, business
logistics management, digital business ecosystem and collaboration system works
of literature.
Digital ecosystem developers can learn from the important opinions of donor/top-
level management for developing a dynamic suitable HLDBE for their needs.

7 Conclusions

Our study examined the controversy and interest of senior-level managers working
in the humanitarian organisations and business logistics companies in English-
speaking Eastern and Southern African countries and their propensity to use, diffuse
and adopt the HLDBE, a digital ecosystem proposed by Baffoe and Luo (2020). We
observed that DOI, TOE and Inst.T offer a plausible theoretical framework to predict
the perceived interest of executives in the humanitarian organisation and business
logistics companies to use, diffuse and adopt the HLDBE for their future operational
use. This study indicates that the propensity of senior-level managements to support
the use, diffusion and adoption of the HLDBE is significantly supported. This
behavioural action obtained from the analysed data would help bridge more collab-
orative and mutual benefits efforts among stakeholders, thereby supporting African
Union’s objectives of integrating all economies and in meeting the challenges it
faces using more influential institutions, sustainable instruments and digital ecosys-
tem technology to build economies and save lives. Thus, it is a self-reliance vision.
We encourage more research to be conducted on the HLDBE in the African
continent or other continents.
Due to small sample size and resources, generalising the study serves as a
limitation for this research. In addition, we only used English-speaking Eastern
and Southern African countries, thereby not given a holistic regional picture.
Future study should focus on obtaining more sample size, conducting research in
a complete regional block in any of the African continent or the entire continent to
ascertain a good objective view for executives’ perspective for HLDBE use, diffu-
sion and adoption for their future operations as another sustainable tool. We can
perform heterogeneity assessment and use advanced PLS-SEM test assessment.

Acknowledgments We duly acknowledge Assistant Professor Seth K. A. Baffoe, Mrs. Jessica

Baffoe, Mr. Joseph E.K. Baffoe and family, Mrs. Yan Yan Baffoe and all Organisational Executives
who responded to this study.
584 B. O. K. Baffoe and W. Luo

References

Adner R (2006) Match your innovation strategy to your innovation ecosystem. Harv Bus Rev 84(4):
98–107
Africa Union (2019) Press Release. https://au.int/sites/default/files/pressreleases/36381-pr-consul
tative_meeting_with_independent_experts_on_the_operationalisation_of_african_humanitar
ian_agency_opens_.pdf. Accessed Jun 2020
African Union Commission (2019) Africa Common Position to the 2019 Global Platform for
Disaster Risk Reduction. https://www.preventionweb.net/files/67051_
africacommonpositiontothe2019global.pdf. Accessed Jun 2020
Aguinis H, Gottfredson RK, Joo H (2013) Best-practice recommendations for defining, identifying,
and handling outliers. Organ Res Methods 16(2):270–301
Ajzen I (1991) The theory of planned behavior. Organ Behav Hum Decis Process 50(2):179–211
Ajzen I, Fishbein M (1973) Attitudinal and normative variables as predictors of specific behavior. J
Pers Soc Psychol 27(1):41–57
Al-Mamary YH, Shamsuddin A, Aziati N (2015) Investigating the key factors influencing on
management information systems adoption among telecommunication companies in Yemen:
the conceptual framework development. Int J Energy Inf Commun 16(1):59–68
Appel G, Libai B, Muller E (2018) On the monetary impact of fashion design piracy. Int J Res Mark
35(4):591–610
Arshad Y, Chin WP, Yahaya SN, Nizam NZ, Masrom NR, Ibrahim SNS (2018) Small and medium
enterprises’ adoption for e-commerce in Malaysia Tourism state. Int J Acad Res Bus Soc Sci
8(10):1457–1557
AU Economic Affairs (2019) 11th African Private Sector Forum. https://au.int/en/documents/201
91108/11th-african-private-sector-forum-declaration. Accessed Jun 2020
AUC African Union Commission (2019) African Regional Integration Report: towards an inte-
grated, prosperous and peaceful Africa. Voices of RECs, Addis Ababa (Ethiopia)
Awa HO, Ojiabo OU, Orokor LE (2017) Integrated technology-organization-environment (T-O-E)
taxonomies for technology adoption. J Enterp Inf Manag 30(6):893–921
Baffoe BOK, Luo W (2021) South African executives propensity to use, diffuse, and adopt the
humanitarian logistics digital business ecosystem (HLDBE). SAGE Open 11(3):1–18
Baffoe BOK, Luo W (2020) Humanitarian relief sustainability: a framework of humanitarian
logistics digital business ecosystem. Transport Res Procedia 48:363–387
Baffoe BOK, Wenping L, Phillips JI (2022) Humanitarian-business organisation collaboration in
Africa: a review perspective. Int J Econ Comm Manag X(5):278–292
Baffoe BOK, Luo W, Pan Q, Zhou S, Wu M, Atimu L, Darko P, Opoku-Mensah E (2023)
Assessing the factors for humanitarian logistics digital business ecosystem (HLDBE) using a
novel integrated correlation coefficient and standard deviation – combined compromise solution
(CCSD-CoCoSo) method. Decis Sci Lett 12(1):117–136
Beamon BM, Balcik B (2008) Performance measurement in humanitarian relief chains. Int J Public
Sect Manag 21(1):4–25
Becker J-M, Ringle CM, Sarstedt M (2018) Estimating Moderating Effects in PLS-SEM and PLSc-
SEM: Interaction Term Generation*Data Treatment. J Appl Struct Eq Model 2(2):1–21
Bhattacharya M, Wamba SF, Kamdjoug JRK (2019) Exploring the determinants of ERP adoption
intention: the case of ERP-enabled emergency service. Int J Technol Diffus 10(4):58–76
Bower JL, Christensen CM (1995) Disruptive technologies: catching the wave. Harv Bus Rev
73(1):43–53
Brewer ML (2016) Facilitating the dissemination of interprofessional education and practice using
an innovative conference approach to engage stakeholders. J Interprofessional Educ Pract 2:33–
38
Cameron KS, Quinn RE, DeGraff J, Thakor AV (2006) Competing values leadership: creating
value in organizations. Edward Elgar, UK
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 585

Capgemini Consulting (2019) Technological innovation for humanitarian aid and assistance.
https://www.europarl.europa.eu/RegData/etudes/STUD/2019/634411/EPRS_STU(2019)
634411_EN.pdf. Accessed 3 Mar 2020
Cichosz M (2018) Digitalization and competitiveness in the logistics service industry. e-mentor
77(5):73–82
CILT (n.d.) The Chartered Institute of Logistics and Transport. http://ciltsa.org.za/. Accessed
Jun 2020
Compeau DR, Higgins CA (1995) Application of social cognitive theory to training for computer
skills. Inf Syst Res 6(2):118–143
Daft RL, Lengel RH (1983) Information richness. A new approach to managerial behavior and
organization design – Technical Report. Defense Technical Information Center, Department of
Management Texas A&M University
Davis FD (1989) Perceived usefulness, perceived ease of use, and user acceptance of information
technology. MIS Q 13(3):319–340
de Camargo Fiorini P, Roman Pais Seles BM, Chiappetta Jabbour CJ, Barberio Mariano E, de
Sousa Jabbour ABL (2018) Management theory and big data literature: from a review to a
research agenda. Int J Inf Manag 43:112–129
DePietro R, Wiarda E, Fleischer M (1990) The context for change: organization, technology and
environment. In: Tornatzky LG, Fleischer M (eds) The process of technological innovation.
Lexington Books, Lexington, pp 151–175
DiMaggio PJ, Powell WW (1983) The iron cage revisited: institutional isomorphism and collective
rationality in organizational ﬁelds. Am Sociol Rev 48(2):147–160
Dubey R, Gunasekaran A, Childe SJ, Luo Z, Wamba SF, Roubaud D, Foropon C (2018) Examining
the role of big data and predictive analytics on collaborative performance in context to
sustainable consumption and production behaviour. J Clean Prod 196:1508–1521
Dubey R, Gunasekaran A, Childe SJ, Blome C, Papadopoulos T (2019a) Big data and predictive
analytics and manufacturing performance: integrating institutional theory, resource-based view
and big data culture. Br J Manag 30(2):341–361
Dubey R, Gunasekaran A, Childe SJ, Papadopoulos T, Luo Z, Wamba SF, Roubaud D (2019b) Can
big data and predictive analytics improve social and environmental sustainability? Technol
Forecast Soc Chang 144:534–545
EAC (1999–2020) EAC partner states. https://www.eac.int/eac-partner-states. Accessed Aug 2019
Faul F, Erdfelder E, Buchner A, Lang AG (2009) Statistical power analyses using G*Power 3.1:
tests for correlation and regression analyses. Behav Res Methods 41:1149–1160
Gangwar H, Date H, Ramaswamy R (2015) Understanding determinants of cloud computing
adoption using an integrated TAM-TOE model. J Enterp Inf Manag 28(1):107–130
Global Database (2019) Global Database. www.globaldatabase.com. Accessed 5 May 2020
Govindan K, Cheng TCE, Mishra N, Shukla N (2018) Big data analytics and application for
logistics and supply chain management. Transport Res E Logist Transport Rev 114:343–349
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2014) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Hult GTM, Ringle CM, Sarstedt M (2013) A primer on partial least squares structural
equation modeling (PLS-SEM). Sage, Thousand Oaks, CA
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Sarstedt M, Matthews LM, Ringle CM (2016) Identifying and treating unobserved
heterogeneity with FIMIX-PLS: Part I – Method. Eur Bus Rev 28(1):63–76
Handﬁeld R, Straube F, Pfohl H-C, Wieland A (2013) Trends and strategies in logistics and supply
chain management. Embracing global logistics complexity to drive market advantage. DVV
Media Group GmbH, Hamburg
586 B. O. K. Baffoe and W. Luo

Henseler J, Ringle CM, Sinkovics RR (2009) The use of partial least squares path modeling in
international marketing. In: Sinkovics RR, Ghauri PN (eds) New challenges to international
marketing (Advances in international marketing), vol 20. Emerald Group, Bingley, pp 277–319
Heskett JL (1977) Logistics-essential to strategy. Harv Bus Rev 55(6):85–96
Hillmer U (2009) Technology acceptance in mechatronics: the influences of identity on technology
acceptance. Gabler, Wiesbaden
Jahre M, Pazirandeh A, Van Wassenhove L (2016) Defining logistics preparedness: a framework
and research agenda. J Human Logist Supply Chain Manag 6(3):372–398
Kabra G, Ramesh A, Akhtar P, Dash MK (2017) Understanding behavioural intention to use
information technology: insights from humanitarian practitioners. Telematics Inform 34(7):
1250–1261
Kendall JD, Tung LL, Chua KH, Ng CHD, Tan SM (2001) Receptivity of Singapore’s SMEs to
electronic commerce adoption. J Strateg Inf Syst 10(3):223–242
Kim M, Chai S (2017) The impact of supplier innovativeness, information sharing and strategic
sourcing on improving supply chain agility: global supply chain perspective. Int J Prod Econ
187:42–52
Kovács G, Spens KM (2007) Humanitarian logistics in disaster relief operations. Int J Phys Distrib
Logist Manag 37(2):99–114
Kovács G, Spens KM (2009) Identifying challenges in humanitarian logistics. Int J Phys Distrib
Logist Manag 39(6):506–528
Lai Y, Sun H, Ren J (2018) Understanding the determinants of big data analytics (BDA) adoption in
logistics and supply chain management: an empirical investigation. Int J Logist Manag 29(2):
676–703
Lee TH (2015) Regression analysis of cloud computing adoption for U.S. hospitals. Doctoral thesis,
Walden University
Lenkenhoff K, Wilkens U, Zheng M, Süße T, Kuhlenkötter B, Ming X (2018) Key challenges of
digital business ecosystem development and how to cope with them. Procedia CIRP 73:167–
172
Lian J-W, Yen DC, Wang Y-T (2014) An exploratory study to understand the critical factors
affecting the decision to adopt cloud computing in Taiwan hospital. Int J Inf Manag 34(1):28–36
MacVaugh J, Schiavone F (2010) Limits to the diffusion of innovation: a literature review and
integrative model. Eur J Innov Manag 13(2):197–221
Matthews LM, Sarstedt M, Hair JF, Ringle CM (2016) Identifying and treating unobserved
heterogeneity with FIMIX-PLS: Part II – A case study. Eur Bus Rev 28(2):208–224
Miranda MQ, Farias JS, de Araújo Schwartz C, de Almeida JPL (2016) Technology adoption in
diffusion of innovations perspective: introduction of an ERP system in a non-profit organiza-
tion. Revista de Administração e Inovação 13(1):48–57
Moore JF (1993) Predators and prey: a new ecology of competition. Harv Bus Rev 71(3):75–86
Ndung’u N, Signé L (2020) The Fourth Industrial Revolution and digitization will transform Africa
into a global powerhouse. https://www.brookings.edu/research/the-fourth-industrial-revolution-
and-digitization-will-transform-africa-into-a-global-powerhouse/. Accessed Jun 2020
Nurmala N, de Leeuw S, Dullaert W (2017) Humanitarian–business partnerships in managing
humanitarian logistics. Supply Chain Manag 22(1):82–94
Nurmala N, de Vries J, de Leeuw S (2018) Cross-sector humanitarian–business partnerships in
managing humanitarian logistics: an empirical verification. Int J Prod Res 56(21):6842–6858
OCHA (2012) OCHA on message: humanitarian principles from what are humanitarian principles?
https://www.unocha.org/sites/dms/Documents/OOMhumanitarianprinciples_eng_June12.pdf.
Accessed Mar 2020
Oloruntoba R, Gray R (2009) Customer service in emergency relief chains. Int J Phys Distrib Logist
Manag 39(6):486–505
Osborne JW (2013) Best practices in data cleaning: a complete guide to everything you need to do
before and after collecting your data. Sage, US
Ouchi WG (1980) Markets, bureaucracies, and clans. Adm Sci Q 25(1):129–141
Receptivity of Eastern and Southern African English-Speaking Countries’. . . 587

Podsakoff PM, MacKenzie SB, Lee J-Y, Podsakoff NP (2003) Common method biases in behav-
ioral research: a critical review of the literature and recommended remedies. J Appl Psychol
88(5):879–903
Ringle CM, Sarstedt M (2016) Gain more insight from your PLS-SEM results. Ind Manag Data Syst
116(9):1865–1886
Rogers EM (1962) Diffusion of innovations. Free Press of Glencoe, New York
Rogers EM (2003) Diffusion of innovations. Simon & Schuster, Delran, NJ
SAAFF (n.d.) The South African Association of Freight Forwarders. https://saaff.org.za/. Accessed
Aug 2019
SADC (2012) Member States. https://www.sadc.int/member-states/. Accessed Aug 2019
SANGOCO (n.d.) The South African National NGO Coalition. http://www.sangoco.org.za/.
Accessed Aug 2019
Sarstedt M, Radomir L, Moisescu OI, Ringle CM (2022) Latent class analysis in PLS-SEM: a
review and recommendations for future applications. J Bus Res 138:398–407
SCEA (n.d.) Shippers Council of Eastern Africa https://www.shipperscouncilea.org/. Accessed
Aug 2019
Schumpeter J (1942) Creative destruction. Capitalism, socialism and democracy. 825:82–85
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Songwe V (2019) A digital Africa. https://www.imf.org/external/pubs/ft/fandd/2019/06/pdf/digital-
africa-songwe.pdf. Accessed May 2020
Stone M (1974) Cross-validatory choice and assessment of statistical predictions. J R Stat Soc Ser B
Methodol 36(2):111–147
SurveyMonkey Audience (2020) SurveyMonkey Audience. www.surveymonkey.com/mp/
audience
Tarofder AK, Azam SMF, Jalal AN (2017) Operational or strategic benefits: empirical investigation
of internet adoption in supply chain management. Manag Res Rev 40(1):28–52
TASA (n.d.) Truckers Association of South Africa. https://www.truckersassociation.co.za/.
Accessed Aug 2019
Tashkandi AN, Al-Jabri IM (2015) Cloud computing adoption by higher education institutions in
Saudi Arabia: an exploratory study. Clust Comput 18(4):1527–1537
Thomas A, Fritz L (2006) Disaster relief. Harv Bus Rev 84:114–122
Tofighi S, Torabi SA, Mansouri SA (2016) Humanitarian logistics network design under mixed
uncertainty. Eur J Oper Res 250(1):239–250
Tornatzky LG, Fleischer M (1990) The processes of technological innovation. Lexington
Books, US
Trunick PA (2005) Special report: Delivering relief to tsunami victims. Logist Today 46(2):1–3
Tsai W-C, Tang L-L (2012) A model of the adoption of radio frequency identification technology:
the case of logistics service firms. J Eng Technol Manag 29(1):131–151
Tweel A (2012) Examining the relationship between technological, organizational, and environ-
mental factors and cloud computing adoption. PhD Dissertation, Northcentral University
UN & World Economic Forum (2008) World Economic Forum (WEF)—OCHA Guiding Princi-
ples for Public-Private Collaboration in Humanitarian Action. https://
interagencystandingcommittee.org/system/files/legacy_files/World%20Economic%20Forum%
20-%20OCHA%20Guiding%20Principles%20for%20Public-Private%20Collaboration%20in
%20Humanitarian%20Action.pdf. Accessed Mar 2020
UNPD (2015) Socio-economic impact of Ebola virus disease in West African countries. https://
www.undp.org/content/dam/rba/docs/Reports/ebola-west-africa.pdf. Accessed 5 May 2020
Vagnani G, Volpe L (2017) Innovation attributes and managers’ decisions about the adoption of
innovations in organizations: a meta-analytical review. Int J Innov Stud 1(2):107–133
Valente TW, Dyal SR, Chu K-H, Wipfli H, Fujimoto K (2015) Diffusion of innovations theory
applied to global tobacco control treaty ratification. Soc Sci Med 145:89–97
588 B. O. K. Baffoe and W. Luo

Van Wassenhove LN (2006) Humanitarian aid logistics: supply chain management in high gear. J
Oper Res Soc 57(5):475–489
Venkatesh V, Morris MG, Davis GB, Davis FD (2003) User acceptance of information technology:
toward a uniﬁed view. MIS Q 27(3):425–478
Wang MW, Lee O-K, Lim KH (2007) Knowledge management systems diffusion in Chinese
enterprises: a multi-stage approach with the technology-organization-environment
framework. In: 11th Paciﬁc Asia Conference on Information Systems: Managing Diversity in
Digital Enterprises (PACIS 2007), Auckland, New Zealand, 4–6 July
Wang Y-M, Wang Y-S, Yang Y-F (2010) Understanding the determinants of RFID adoption in the
manufacturing industry. Technol Forecast Soc Chang 77(5):803–815
Wong KK-K (2019) Mastering partial least squares structural equation modeling (PLS-Sem) with
Smartpls in 38 hours. iUniverse, Bloomington, IN
Zhu K (2004) The complementarity of information technology infrastructure and e-commerce
capability: a resource-based assessment of their business value. J Manag Inf Syst 21(1):167–202
Zokaee S, Bozorgi-Amiri A, Sadjadi SJ (2016) A robust optimization model for humanitarian relief
chain design under uncertainty. Appl Math Model 40(17–18):7996–8016
A PLS-SEM Model on Auditing and GDPR:
The Mediating Role of Internal Audit
in Business Management

Stratos Moschidis, Evrikleia Chatzipetrou, George Drogalas,

and Paraskevi Zisiopoulou

1 Introduction

Corporate governance is a critical success factor for businesses, since it describes the
way they are managed and controlled. The audit is an essential key to corporate
governance (Rahman et al. 2019). During the audit, the auditor inspects and exam-
ines accounts and financial statements, in order to assess the company’s activity,
using scientifically proven procedures (Asbahr and Ruhnke 2019). The audit can be
both external and internal, with the internal control system receiving special atten-
tion (Caplan et al. 2018). The General Data Protection Regulation (GDPR) was
approved by the European Parliament in 2016 and has been in effect in all EU
Member States without exception since 2018 (Eur Lex 2016). The Regulation’s goal
is to harmonize the various privacy laws in force in the EU and to strengthen the
protection of personal or sensitive data (De Hert et al. 2018). Personal data are
defined as “any information relating to an identified or identifiable natural person”

S. Moschidis (✉)
Hellenic Statistical Authority, Piraeus, Greece
e-mail: [email protected]
E. Chatzipetrou
Department of Accounting and Information Systems, International Hellenic University, Sindos,
Greece
e-mail: [email protected]
G. Drogalas
Department of Business Administration, University of Macedonia, Egnatias, Thessaloniki,
Greece
e-mail: [email protected]
P. Zisiopoulou
University of Macedonia, Thessaloniki, Greece
e-mail: [email protected]

(Art. 4, GDPR—Eur Lex 2016). It includes data and information that can be used to
directly or indirectly identify an individual (e.g., e-mail address, bank account
details, photographs, medical information, Internet protocol [IP] addresses, etc.).
Special categories of data, such as genetic or biometric characteristics, are also
classified as sensitive. The audit activity has direct contact with elements protected
by the GDPR and must thus be used with caution. Since this crucially important new
parameter has been only introduced relatively recently as a result of the regulation,
and since there is consequently a shortage of substantial bibliographic coverage of
the topic, it is essential to investigate the perspectives of businesses on this matter. It
is, therefore, essential to examine the role of auditing in order to fill the research gap
surrounding the auditing processes in the GDPR implementation framework since
enterprises risk extremely severe penalties and fines if they fail to comply with the
regulations, which can amount to 20 million euros or 4% of their annual revenue.

2 Methodology: Literature Review and Research

Hypotheses
2.1 Internal Audit and Management Support

Management support is an important factor in the effective exercise of internal audit.

The work of Alzeban and Gwilliam (2014) justified the above proposal, by
portraying the company’s senior/top management as the component responsible
for providing internal control executives with the freedom and perspective that
will lead to the perceived effectiveness of internal audit. In addition, George et al.
(2015) demonstrated, in a relevant study, the types of responsibilities that signifi-
cantly contribute to the effectiveness of internal control.
H1: Management support affects positively and statistically significantly internal
audit.

2.2 Internal Audit and Risk Management

The Department of Internal Audit and the Risk Management shares the responsibil-
ity for risk and its management, in accordance with Brody and Lowe (2000). Proper
risk management is critical for businesses, as it can ensure the effective operation of
risk-based controls, while providing effective advice through internal audit (Frigo
and Anderson 2011).
H2: Internal audit affects positively and statistically signiﬁcantly risk management.
A PLS-SEM Model on Auditing and GDPR: The Mediating Role of Internal. . . 591

2.3 Internal Audit and GDPR

According to the GDPR, the definition of the auditor is redefined in terms of both
purposes and means of processing, while it must prove its compliance at any time
(Zaeem and Barber 2020). Internal audit, as an advisory body, ensures the proper
operation of the risk and audit procedures associated with GDPR compliance.
Internal auditing can also help businesses by raising awareness about data security
issues, assessing potential risks, identifying gaps, and testing restorative processes
(Lois et al. 2020).
H3: GDPR compliance is positively influenced by internal audit.

2.4 Management Support and GDPR

A company’s compliance with GDPR is critical, because failure to do so may result

in severe consequences or sanctions. Brodin (2019) proposed a framework approved
by senior management executives to assist small- and medium-sized enterprises
(SMEs) with GDPR compliance. Tamburri (2020) also proposed how managers
and data officers in companies can evaluate and use the software in accordance with
the general data protection (Labadie and Legner 2019) regulation. Furthermore,
Labadie and Legner (2019) defined a capability model that assists management
professionals in complying with the GDPR. The following hypothesis can be
developed based on the foregoing.
H4: Management support affects positively and statistically significantly GDPR
compliance.

2.5 The Mediating Role of Internal Audit in the Relationship

Between Management and GDPR

Management support is a signiﬁcant driver for the perceived success of the internal
audit function, according to the study by Alzeban and Gwilliam (2014), while
comparable studies have looked at the duties that contribute to effective internal
audit (Alqudah et al. 2019; Salehi 2016). Furthermore, GDPR compliance is one of
the top goals of the internal audit program. The foregoing connection of internal
audit with management support and the GDPR leads to the next explanatory
hypothesis.
H5: Internal audit serves as a mediator between management support and GDPR
implementation.
592 S. Moschidis et al.

3 Analysis

In order to investigate the above hypotheses, the PLS-SEM method, also known as
path modeling (Hair et al. 2011), was adopted. The software SmartPLS Version 3.0
(Sarstedt and Cheah 2019) was used as a tool for conducting the analysis. The
questionnaire consists of Likert scales (1 = strongly disagree, 5 = strongly agree).
According to the hypotheses (H1–H5), the following model (Fig. 1) was created with
SmartPLS Version 3 software (Ringle et al. 2015).
The proposed model was also monitored for minimum sample size requirement
according to inverse square root (ISR) method (Kock and Hadaya 2018). According
to ISR’s calculations, the sample that was used (120 enterprises) satisfies the
minimum sample size at the significance level of 5% where n > (0.248/pmin)2
= (2.486/0.304)2 = 66.7 is the minimum sample number and pmin is the minimum
value of path coefficient of inner model that is 0.304 in our case.

e1
0.839 (0.000)
e3 0.852 (0.000) 0.315
b10
0.803 (0.000)

e4 0.847 (0.000) Risk management

b11 0.849 (0.000)

0.825 (0.000) 0.562 (0.000) c1
e5

0.869 (0.000)
0.920 (0.000)
e6
b2 c2
0.744 (0.000) 0.937 (0.000)
0.474 (0.000) 0.225
0.758 (0.000)
b3 0.890 (0.000)
0.791 (0.000)
Management c4
Internal Audit
b4 support
0.788 (0.000)
0.304 (0.001) 0.382 (0.000)

f1 0.899 (0.000) 0.349

0.844 (0.000)
f3 0.890 (0.000)
GDPR

Fig. 1 PLS-SEM reflective–reflective model with SmartPLS Version 3.0 (Ringle et al. 2015).
Notes: Path coefficients and p-values for inner model, outer weights/loadings and p-values for outer
model, and R2 values for endogenous constructs
A PLS-SEM Model on Auditing and GDPR: The Mediating Role of Internal. . . 593

3.1 Assessment of Measurement Model

The assessment of the measurement model or the outer model is being performed
through the investigation into loadings, Cronbach’s α, rho_A, composite reliability
(CR), and average variance extracted (AVE) indicators (Table 1). All loadings
exceed the threshold of 0.708, AVE is higher than the critical value of 0.5, and all
construct reliabilities have values that exceed 0.7 (Sarstedt and Cheah 2019).
As part of the assessment of the measurement model, discriminant validity is also
being investigated through heterotrait–monotrait ratio of correlations (HTMT) cri-
terion. According to Henseler et al. (2015), the values in Table 2 are signiﬁcantly
lower than 0.85, thus supporting discriminant validity.

3.2 Assessment of Structural Model

The assessment of structural model includes collinearity, path coefﬁcient signiﬁ-

cance, explanatory (in-sample) power, and predictive (out-of-sample) relevance.
First, we assess the structural model for collinearity issues by examining the variance
inflation factor (VIF) (Hair et al. 2019; Sarstedt et al. 2021). Considering
Diamantopoulos and Siguaw (2006) and Hair et al. (2019)’s studies, we conclude
that variance inflation factor (VIF) values are below the threshold value of 3.3
(Table 3); hence, it is not critical for our model (Sarstedt et al. 2021).
Significance and relevance of path coefficients: The results of bootstrapping with
500 samples indicate that all model relationships are significant at 95% confidence
level.
In order to assess in-sample model fit, we considered R2. The analysis reveals that
values of R2 of the endogenous constructs are considered having something between
weak and moderate (0.349, 0.225, 0.315) in-sample explanatory power (Hair et al.
2019). In order to assess the out-of-sample predictive power, we used PLSpredict
(Shmueli et al. 2019) algorithm that is implemented in SmartPLS Version 3.0
(Ringle et al. 2015).
In order to evaluate the predictive power/relevance of our model, we used the
PLSpredict (Hair et al. 2019, 2020; Shmueli et al. 2019) procedure keeping cross-
validation and repetition default values k = 10 (k = folds) and r = 10 (r =
repetitions), while k selection still satisfied the lowest acceptable number of
instances (data rows) (Table 4).
We used RMSE to compare PLS-SEM and LM for predicting the relevance of
key constructs as errors are distributed normally as it can be seen on the Error
Histograms. Considering risk management as a key endogenous construct, the
minority of manifest variables (MV) satisfies PLS_RMSE < LM_RMSE, so we
conclude that our model has small predictive power (Sarstedt et al. 2021) (Table 5).
Table 1 Loadings, composite reliability, convergent validity, and discriminant validity
594

Latent Indicator/ Cronbach’s CR (composite AVE (average

variable (LV) item Description Loadings α rho_A reliability) variance extracted)
Management b10 The management supports the career development 0.825 0.886 0.895 0.912 0.635
support of its employees
b11 The management focuses on controlling processes 0.869
b2 The management provides human resources 0.744
b3 The management focuses on efficiency 0.758
b4 The management supports its employees 0.791
b9 The management focuses on optimizing internal 0.788
operations
Internal audit c1 Internal audit provides advice 0.920 0.903 0.905 0.940 0.838
c2 Internal audit enhances efficiency 0.937
c4 Internal audit ensures data security 0.890
Risk e1 Risk management helps to avoid uncertainty 0.839 0.895 0.906 0.922 0.702
management about events
e3 Risk management contributes to the assessment 0.852
and recognition of the risk itself
e4 Risk management can prevent insecure 0.803
information
e5 Risk management can help to reassure the out- 0.847
come of events
e6 Risk management can contribute to the develop- 0.849
ment of the solvency of the firm
GDPR f1 GDPR safeguards privacy and personal data 0.899 0.860 0.952 0.910 0.771
f3 The adoption of GDPR is very important for the 0.844
quality of the business
f4 Adopting the GDPR is particularly beneficial for 0.890
retaining important information
S. Moschidis et al.

Notes: Some items are seemed to be missing (e.g., after e1 goes to e3, then f1 goes to f3, etc.). This was done since the remainder of the used elements had
loadings below the critical value of 0.708 and, as a result, they were eliminated from the model.
A PLS-SEM Model on Auditing and GDPR: The Mediating Role of Internal. . . 595

Table 2 Discriminant validity assessment

HTMT criterion
Management Risk
support GDPR Internal audit management
Management 0.746
support [0.661; 0.818]
GDPR 0.496
[0.327; 0.674]
Internal audit 0.505 0.553
[0.251; 0.705] [0.411; 0.690]
Risk management 0.531 0.723 0.613
[0.374; 0.731] [0.584; 0.811] [0.426;
0.777]
Notes: All numbers in brackets represent 95% bias-corrected and accelerated (BCa) conﬁdence
(Shmueli et al. 2019) intervals from bootstrapping (BCa) with 500 iterations.

Table 3 Assessment of the structural model, collinearity, path coefﬁcients, explanatory power, and
predictive relevance
Standard p- Lower Upper
Relationship β value bound bound VIF R2
Internal audit → GDPR 0.382 0.000 0.217 0.556 1.290 0.349
Management support → GDPR 0.304 0.002 0.107 0.488 1.290
Management support → Inter- 0.474 0.000 0.229 0.633 1.000 0.225
nal audit
Internal audit → Risk 0.562 0.000 0.417 0.707 1.000 0.315
management

Table 4 Manifest variables (MV) prediction summary with PLS-SEM prediction

Root mean square error (RMSE) RMSE (linear (PLS-SEM-
MV (PLS-SEM model) model [LM]) LM) Q2predict
f3 0.884 0.842 0.042 0.030
f4 0.930 0.901 0.029 0.138
f1 0.737 0.669 0.068 0.247
c4 0.978 0.941 0.037 0.123
c2 0.853 0.839 0.014 0.235
c1 0.945 0.933 0.012 0.104
e5 0.646 0.601 0.045 0.118
e4 0.818 0.724 0.094 0.098
e6 0.744 0.691 0.053 0.141
e1 0.684 0.670 0.014 0.137
e3 0.747 0.762 -0.015 0.120
596 S. Moschidis et al.

Table 5 Hypotheses testing

Standard
deviation p-
Relationship β Mean (SD) values Decision
Direct effects
H4 Management support → GDPR 0.486 0.498 0.067 0.000 Supported
H1 Management support → Internal 0.474 0.492 0.095 0.000 Supported
audit
H3 Internal audit → GDPR 0.382 0.385 0.087 0.000 Supported
H2 Internal audit → Risk management 0.562 0.576 0.071 0.000 Supported
Indirect effects
H5 Management support → Internal 0.181 0.191 0.063 0.004 Supported
audit → GDPR
H5 Management support → Internal 0.266 0.287 0.082 0.001 Supported
audit → Risk management

4 Conclusions

The implementation of GDPR rules has resulted in signiﬁcant changes in the way
businesses are audited. This chapter evaluates components such as management
support, internal audit, GDPR, and risk management. The corresponding research
hypotheses were developed through a relevant literature review and tested using a
structural equation model created by the PLS-SEM method. The proposed model
went under consideration related to measurement model and structural model. The
proposed model supports hypotheses H1–H5 and also indicates the existence of
partial mediation of internal audit between management support and GDPR rela-
tionship. The results come to supplement the existing knowledge, while an attempt
will be made to expand the current model at a later stage.

References

Alqudah HM, Amran NA, Hassan H (2019) Factors affecting the internal auditors’ effectiveness in
the Jordanian public sector: the moderating effect of task complexity. EuroMed J Bus 14(3):
251–273
Alzeban A, Gwilliam D (2014) Factors affecting the internal audit effectiveness: a survey of the
Saudi public sector. J Int Account Audit Tax 23(2):74–86
Asbahr K, Ruhnke K (2019) Real effects of reporting key audit matters on auditors’ judgment and
choice of action. Int J Audit 23(2):165–180
Brodin M (2019) A framework for GDPR compliance for small- and medium-sized enterprises. Eur
J Secur Res 4(2):243–264
Brody RG, Lowe DJ (2000) The new role of the internal auditor: implications for internal auditor
objectivity. Int J Audit 4(2):169–176
Caplan DH, Dutta SK, Liu AZ (2018) Are material weaknesses in internal controls associated with
poor M&A decisions? Evidence from goodwill impairment. AUDITING 37(4):49–74
A PLS-SEM Model on Auditing and GDPR: The Mediating Role of Internal. . . 597

De Hert P, Papakonstantinou V, Malgieri G, Beslay L, Sanchez I (2018) The right to data portability
in the GDPR: towards user-centric interoperability of digital services. Comput Law Secur Rev
34(2):193–203
Diamantopoulos A, Siguaw JA (2006) Formative versus reﬂective indicators in organizational
measure development: a comparison and empirical illustration. Br J Manag 17(4):263–282
Eur Lex (2016) General data protection regulation (GDPR). https://eur-lex.europa.eu/EN/legal-
content/summary/general-data-protection-regulation-gdpr.html#:~:text=Regulation%20(EU)%
202016%2F679%20of%20the%20European%20Parliament%20and,(OJ%20L%20119%2C%
204.5. Accessed 16 Mar 2021
Frigo ML, Anderson RJ (2011) Strategic risk management: a foundation for improving enterprise
risk management and governance. J Corp Account Financ 22(3):81–88
George D, Theofanis K, Konstantinos A (2015) Factors associated with internal audit effectiveness:
evidence from Greece. J Account Tax 7(7):113–122
Hair JF, Ringle CM, Sarstedt M (2011) PLS-SEM: indeed a silver bullet. J Mark Theory Pract
19(2):139–152
Hair JF, Risher JJ, Sarstedt M, Ringle CM (2019) When to use and how to report the results of
PLS-SEM. Eur Bus Rev 31(1):2–24
Hair JF, Howard MC, Nitzl C (2020) Assessing measurement model quality in PLS-SEM using
conﬁrmatory composite analysis. J Bus Res 109:101–110
Henseler J, Ringle CM, Sarstedt M (2015) A new criterion for assessing discriminant validity in
variance-based structural equation modeling. J Acad Mark Sci 43(1):115–135
Kock N, Hadaya P (2018) Minimum sample size estimation in PLS-SEM: the inverse square root
and gamma-exponential methods: Sample size in PLS-based SEM. Inf Syst J 28(1):227–261
Labadie C, Legner C (2019) Understanding data protection regulations from a data management
perspective: a capability-based approach to EU-GDPR. In: Proceedings of the 14th International
Conference on Wirtschaftsinformatik, Siegen, Germany, 24–27 February. pp 1292–1306
Lois P, Drogalas G, Karagiorgos A, Tsikalakis K (2020) Internal audits in the digital era: oppor-
tunities risks and challenges. EuroMed J Bus 15(2):205–217
Rahman MM, Sobhan R, Islam MS (2019) Intellectual capital disclosure and its determinants:
empirical evidence from listed pharmaceutical and chemical industry of Bangladesh. Asian J
Bus Environ 9(2):35–46
Ringle CM, Wende S, Becker J-M et al (2015) SmartPLS 3. SmartPLS GmbH, Boenningstedt
Salehi T (2016) Investigation factors affecting the effectiveness of internal auditors in the company:
case study Iran. Rev Eur Stud 8(2):224–235
Sarstedt M, Cheah J-H (2019) Partial least squares structural equation modeling using SmartPLS: a
software review. J Market Anal 7:196–202
Sarstedt M, Ringle CM, Hair JF (2021) Partial least squares structural equation modeling. In:
Homburg C, Klarmann M, Vomberg AE (eds) Handbook of market research. Springer, Cham,
pp 1–47
Shmueli G, Sarstedt M, Hair JF, Cheah J-H, Ting H, Vaithilingam S, Ringle CM (2019) Predictive
model assessment in PLS-SEM: guidelines for using PLSpredict. Eur J Mark 53(11):2322–2347
Tamburri DA (2020) Design principles for the General Data Protection Regulation (GDPR): a
formal concept analysis and its evaluation. Inf Syst 91(July):101469
Zaeem RN, Barber KS (2020) The effect of the GDPR on privacy policies: recent progress and
future promise. ACM Trans Manag Inf Syst 12(1):1–20

Bayesian Statistical Methods
100% (10)
Bayesian Statistical Methods
288 pages
Machine Learning - An Applied Mathematics Introduction PDF
100% (13)
Machine Learning - An Applied Mathematics Introduction PDF
246 pages
Hang Li - Machine Learning Methods-Springer (2023) (Z-Lib - Io)
100% (9)
Hang Li - Machine Learning Methods-Springer (2023) (Z-Lib - Io)
530 pages
Linear Algebra Optimization Machine Learning PDF
100% (12)
Linear Algebra Optimization Machine Learning PDF
507 pages
Sarstedt, M., Hair, J. F., Pick, M., Liengaard, B. D., Radomir, L., & Ringle, C. M. (2022)
No ratings yet
Sarstedt, M., Hair, J. F., Pick, M., Liengaard, B. D., Radomir, L., & Ringle, C. M. (2022)
30 pages
A13、A13A ServiceManual PDF
50% (2)
A13、A13A ServiceManual PDF
1,333 pages
132438-2004 Yamaha Sx200txrc Outboard Service Repair Maintenance Manual Factory
100% (4)
132438-2004 Yamaha Sx200txrc Outboard Service Repair Maintenance Manual Factory
329 pages
Understanding Machine Learning
100% (69)
Understanding Machine Learning
416 pages
Learning Statistics
100% (27)
Learning Statistics
408 pages
Designing Controls For The Process Industries PDF
60% (5)
Designing Controls For The Process Industries PDF
429 pages
David R. Dowty, Robert E. Wall, Stanley Peters (Auth.) - Introduction To Montague Semantics-Springer Netherlands (1981)
No ratings yet
David R. Dowty, Robert E. Wall, Stanley Peters (Auth.) - Introduction To Montague Semantics-Springer Netherlands (1981)
325 pages
Artificial Intelligence With Python (Machine Learning Foundations, Methodologies, and Applications) (Teik Toe Teoh, Zheng Rong)
93% (15)
Artificial Intelligence With Python (Machine Learning Foundations, Methodologies, and Applications) (Teik Toe Teoh, Zheng Rong)
334 pages
Algorithms For Data Science 1st Brian Steele (WWW - Ebook DL - Com)
100% (15)
Algorithms For Data Science 1st Brian Steele (WWW - Ebook DL - Com)
438 pages
Machine Learning With Python
100% (14)
Machine Learning With Python
692 pages
HairJr2021 Book PartialLeastSquaresStructuralE
100% (2)
HairJr2021 Book PartialLeastSquaresStructuralE
208 pages
Computational Simulation Tools in Engineering
From Everand
Computational Simulation Tools in Engineering
V. Ramesh Kumar,
No ratings yet
Full Course of Machine Learning
100% (16)
Full Course of Machine Learning
660 pages
Control Engineering Solutions
100% (10)
Control Engineering Solutions
322 pages
Automatic Control With Experiments PDF
83% (6)
Automatic Control With Experiments PDF
996 pages
Hackers Guide To Machine Learning With Python PDF
100% (15)
Hackers Guide To Machine Learning With Python PDF
272 pages
Jackpot Trigger Device
100% (2)
Jackpot Trigger Device
25 pages
PLS Sem
No ratings yet
PLS Sem
45 pages
2021 Sarstedtetal. HandbookofMarketResearch
No ratings yet
2021 Sarstedtetal. HandbookofMarketResearch
47 pages
When To Use and How To Report The Results of PLS-SEM: Joseph F. Hair
No ratings yet
When To Use and How To Report The Results of PLS-SEM: Joseph F. Hair
23 pages
Partial Least Squares Structural Equation Modeling
No ratings yet
Partial Least Squares Structural Equation Modeling
16 pages
PLS-SEM Results Reporting Format
100% (1)
PLS-SEM Results Reporting Format
46 pages
Avkiran PLS Book Springer
No ratings yet
Avkiran PLS Book Springer
243 pages
PLS SEMPrimer BookFlyer
No ratings yet
PLS SEMPrimer BookFlyer
3 pages
Partial Least Squares Structural Equation Modeling (PLS-SEM) : An Emerging Tool For Business Research
No ratings yet
Partial Least Squares Structural Equation Modeling (PLS-SEM) : An Emerging Tool For Business Research
17 pages
PLS-SEMPrimer BookFlyer
No ratings yet
PLS-SEMPrimer BookFlyer
3 pages
European Business Review: Article Information
No ratings yet
European Business Review: Article Information
39 pages
FLYER Sage Publications LTD - Advanced Issues in Partial Least Squares Structural Equation Modeling - 2023-11-08
No ratings yet
FLYER Sage Publications LTD - Advanced Issues in Partial Least Squares Structural Equation Modeling - 2023-11-08
4 pages
Marketing Sustainability
No ratings yet
Marketing Sustainability
89 pages
PLS SEMPrimer BookFlyer
No ratings yet
PLS SEMPrimer BookFlyer
3 pages
SSRN Id22337955421
No ratings yet
SSRN Id22337955421
12 pages
PLS SEMPrimer BookFlyer
No ratings yet
PLS SEMPrimer BookFlyer
3 pages
Robustness
No ratings yet
Robustness
10 pages
The Use of Partial Least Squares Structural Equation Modeling and Complementary Methods in International Management Research
No ratings yet
The Use of Partial Least Squares Structural Equation Modeling and Complementary Methods in International Management Research
22 pages
A Primer On Partial Leasr Squares Structural Equation Modeling (PLS-SEM)
No ratings yet
A Primer On Partial Leasr Squares Structural Equation Modeling (PLS-SEM)
4 pages
Avkiran 2018
No ratings yet
Avkiran 2018
29 pages
Rethinking Some of The Rethinking of Partial Least Squares
No ratings yet
Rethinking Some of The Rethinking of Partial Least Squares
19 pages
Special Issue Editorial - Advanced Partial Least Squares Structural Equation
No ratings yet
Special Issue Editorial - Advanced Partial Least Squares Structural Equation
11 pages
PLS-SEMs Most Wanted Guidance
No ratings yet
PLS-SEMs Most Wanted Guidance
26 pages
Partial Least Squares Structural Equation Modeling: September 2017
No ratings yet
Partial Least Squares Structural Equation Modeling: September 2017
41 pages
Partial Least Squares Path Modeling
No ratings yet
Partial Least Squares Path Modeling
2 pages
Journal of Business Research: Marko Sarstedt, L Acr Amioara Radomir, Ovidiu Ioan Moisescu, Christian M. Ringle
No ratings yet
Journal of Business Research: Marko Sarstedt, L Acr Amioara Radomir, Ovidiu Ioan Moisescu, Christian M. Ringle
10 pages
Shelah Ramayah 2023
No ratings yet
Shelah Ramayah 2023
20 pages
Structural Equation Modeling: Introduction and Overview: Joe F. Hair, JR
100% (1)
Structural Equation Modeling: Introduction and Overview: Joe F. Hair, JR
24 pages
A Primer On Partial Least Squares Structural Equation Modeling
No ratings yet
A Primer On Partial Least Squares Structural Equation Modeling
3 pages
PLS-SEM and Reflective Constructs - A Response To Recent Criticism and A Constructive Path Forward
No ratings yet
PLS-SEM and Reflective Constructs - A Response To Recent Criticism and A Constructive Path Forward
9 pages
2017 - Sarstedt Et Al. - Handbook of Market Research PDF
No ratings yet
2017 - Sarstedt Et Al. - Handbook of Market Research PDF
41 pages
Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R A Workbook
No ratings yet
Partial Least Squares Structural Equation Modeling (PLS-SEM) Using R A Workbook
209 pages
79 PLS Using PLS Path Modeling in New Technology Research - Updated Guidelines 2016
No ratings yet
79 PLS Using PLS Path Modeling in New Technology Research - Updated Guidelines 2016
22 pages
Partial Least Squares Structural Equations Modeling (PLS-SEM)
No ratings yet
Partial Least Squares Structural Equations Modeling (PLS-SEM)
4 pages
Free Online Course On PLS-SEM Using SmartPLS 3.0 - Higher Order
No ratings yet
Free Online Course On PLS-SEM Using SmartPLS 3.0 - Higher Order
23 pages
A Critical Look at The Use of PLS-SEM in MIS Quarterly
No ratings yet
A Critical Look at The Use of PLS-SEM in MIS Quarterly
18 pages
Ditor S Omments: A Critical Look at The Use of PLS-SEM in MIS Quarterly
No ratings yet
Ditor S Omments: A Critical Look at The Use of PLS-SEM in MIS Quarterly
18 pages
Em-Ejmj190102 1..26
No ratings yet
Em-Ejmj190102 1..26
26 pages
Advanced Marketing Analytics Using Partial Least Squares Structural Equation Modeling (PLS SEM)
No ratings yet
Advanced Marketing Analytics Using Partial Least Squares Structural Equation Modeling (PLS SEM)
5 pages
Partial Least Squares
No ratings yet
Partial Least Squares
32 pages
SEM-ML, Sample Paper
No ratings yet
SEM-ML, Sample Paper
18 pages
Industrial Management & Data Systems: Article Information
No ratings yet
Industrial Management & Data Systems: Article Information
30 pages
Capítulo 6 - Analizando Los Resultados en PLS SEM
No ratings yet
Capítulo 6 - Analizando Los Resultados en PLS SEM
37 pages
Hair 2017
No ratings yet
Hair 2017
22 pages
Partial Least Squares Structural Equation Modeling...
No ratings yet
Partial Least Squares Structural Equation Modeling...
208 pages
Alietal 2018
No ratings yet
Alietal 2018
26 pages
Industrial Management & Data Systems: Article Information
No ratings yet
Industrial Management & Data Systems: Article Information
42 pages
Sempls: Structural Equation Modeling Using Partial Least Squares
No ratings yet
Sempls: Structural Equation Modeling Using Partial Least Squares
34 pages
10 1108 - TQM 06 2022 0197
No ratings yet
10 1108 - TQM 06 2022 0197
10 pages
0 The Use and Misuse of SEM in Social Science Researches
No ratings yet
0 The Use and Misuse of SEM in Social Science Researches
30 pages
International Journal of Mathematics and Statistics Invention (IJMSI)
No ratings yet
International Journal of Mathematics and Statistics Invention (IJMSI)
17 pages
A Primer On Partial Least Squares Struct PDF
No ratings yet
A Primer On Partial Least Squares Struct PDF
2 pages
Office, Ijaa2166
No ratings yet
Office, Ijaa2166
14 pages
2020 - This Fast Car Can Move Faster A Review of PLS-SEM Application in Higher Eduation Research
No ratings yet
2020 - This Fast Car Can Move Faster A Review of PLS-SEM Application in Higher Eduation Research
32 pages
Mastering Partial Least Squares Structural Equation Modeling (Pls-Sem) with Smartpls in 38 Hours
From Everand
Mastering Partial Least Squares Structural Equation Modeling (Pls-Sem) with Smartpls in 38 Hours
Ken Kwong-Kay Wong
3/5 (1)
ICT Project Management: Framework for ICT-based Pedagogy System: Development, Operation, and Management
From Everand
ICT Project Management: Framework for ICT-based Pedagogy System: Development, Operation, and Management
Suman Ahmmed
No ratings yet
Techniques in Mathematical Modelling
From Everand
Techniques in Mathematical Modelling
Gautami Devar
No ratings yet
Ultimate Enterprise Data Analysis and Forecasting using Python: Leverage Cloud platforms with Azure Time Series Insights and AWS Forecast Components for Deep learning Modeling using Python (English Edition)
From Everand
Ultimate Enterprise Data Analysis and Forecasting using Python: Leverage Cloud platforms with Azure Time Series Insights and AWS Forecast Components for Deep learning Modeling using Python (English Edition)
Shanthababu Pandian
No ratings yet
Introduction To Control Engineering Modeling, Analysis and Design - Ajit K. Mandal
100% (4)
Introduction To Control Engineering Modeling, Analysis and Design - Ajit K. Mandal
633 pages
Deep Learning - Fundamentals, Theory and Applications 2019 PDF
100% (10)
Deep Learning - Fundamentals, Theory and Applications 2019 PDF
168 pages
Control Engineering - Theory, Worked Examples and Problems
90% (10)
Control Engineering - Theory, Worked Examples and Problems
117 pages
Robust and Optimal Control
100% (2)
Robust and Optimal Control
346 pages
Jay H Lee - MPC Lecture Notes
100% (2)
Jay H Lee - MPC Lecture Notes
137 pages
Learning Simulink
100% (28)
Learning Simulink
572 pages
Deep Learning in Natural Language Processing PDF
100% (9)
Deep Learning in Natural Language Processing PDF
338 pages
Bayesian Inference Data Evaluation and Decisions Second Edition
100% (2)
Bayesian Inference Data Evaluation and Decisions Second Edition
245 pages
Statistical Regression and Classification - From Linear Models To Machine Learning
100% (10)
Statistical Regression and Classification - From Linear Models To Machine Learning
532 pages
Modern Control System Theory
100% (2)
Modern Control System Theory
716 pages
2016 Book SearchAndOptimizationByMetaheu
100% (1)
2016 Book SearchAndOptimizationByMetaheu
437 pages
Fuzzy Modeling and Control - Andrzej Piegat PDF
100% (2)
Fuzzy Modeling and Control - Andrzej Piegat PDF
737 pages
A Primer On Partial Least Squares Structural Equation Modeling Pls Sem
100% (1)
A Primer On Partial Least Squares Structural Equation Modeling Pls Sem
148 pages
Advance Statistical Methods in Data Science Chen
100% (5)
Advance Statistical Methods in Data Science Chen
229 pages
Elements of Nonlinear Series Analysis and Forecasting PDF
100% (8)
Elements of Nonlinear Series Analysis and Forecasting PDF
626 pages
Model Predictive Control Notes
100% (6)
Model Predictive Control Notes
135 pages
20.009 (Remote Interface Manual v3.x)
No ratings yet
20.009 (Remote Interface Manual v3.x)
50 pages
Ees452 2021 2.1
No ratings yet
Ees452 2021 2.1
9 pages
LITESTAR 4D v. 4.00: User Manual Litecalc - Lighting Design Module
No ratings yet
LITESTAR 4D v. 4.00: User Manual Litecalc - Lighting Design Module
117 pages
Grade 8th Maths Mock 3 Mid Term 1 Answerkey
No ratings yet
Grade 8th Maths Mock 3 Mid Term 1 Answerkey
3 pages
II 2 Reservoir Rock Properties Compressibility
100% (1)
II 2 Reservoir Rock Properties Compressibility
13 pages
28 Saturday: Py-Hon in Slallalion Ooxkfng Progam Py-Then Koswords Jdentievs
No ratings yet
28 Saturday: Py-Hon in Slallalion Ooxkfng Progam Py-Then Koswords Jdentievs
25 pages
DX180LC-3: Crawler Excavator
No ratings yet
DX180LC-3: Crawler Excavator
24 pages
Solid Mechanics - Chapter-13-Riveted, Bolted, and Welded Connections Singer - Pytel (IEM 2-2 KUET)
100% (1)
Solid Mechanics - Chapter-13-Riveted, Bolted, and Welded Connections Singer - Pytel (IEM 2-2 KUET)
61 pages
Data Sheet: HCPL-3120/J312 HCNW3120
No ratings yet
Data Sheet: HCPL-3120/J312 HCNW3120
26 pages
Mas
No ratings yet
Mas
1 page
F.4 Mathematics BTE (3) More About Polynomial
No ratings yet
F.4 Mathematics BTE (3) More About Polynomial
2 pages
Chapter 2 - Storage Methods
No ratings yet
Chapter 2 - Storage Methods
13 pages
Formato Articulo Dyna
No ratings yet
Formato Articulo Dyna
2 pages
Module 5 - Forecasting
No ratings yet
Module 5 - Forecasting
13 pages
June 2014 QP
No ratings yet
June 2014 QP
16 pages
Playing (Mathematics) Games - When Is A Game Not A Game (Gough)
No ratings yet
Playing (Mathematics) Games - When Is A Game Not A Game (Gough)
5 pages
Olefini Catalogue 2015 - Industrial
No ratings yet
Olefini Catalogue 2015 - Industrial
2 pages
APMR Series: Packaged Air Conditioners
100% (1)
APMR Series: Packaged Air Conditioners
28 pages
SP-5 Manual - EN
No ratings yet
SP-5 Manual - EN
32 pages
Summative Test Inmathematics 3 Second Quarter Julio Arzaga Elementary School
No ratings yet
Summative Test Inmathematics 3 Second Quarter Julio Arzaga Elementary School
2 pages
22 Chapter 4 Data Management
No ratings yet
22 Chapter 4 Data Management
75 pages
Grammar Year 6 KSSR ENGLISH 2015
No ratings yet
Grammar Year 6 KSSR ENGLISH 2015
17 pages
Year 6 Converting Units Knowledge Organiser
No ratings yet
Year 6 Converting Units Knowledge Organiser
2 pages
Arithmetic 20 Dec
No ratings yet
Arithmetic 20 Dec
43 pages
Tissue Healing Timeline
No ratings yet
Tissue Healing Timeline
1 page