IEICE TRANS. INF. & SYST., VOL.E102–D, NO.

12 DECEMBER 2019
2524

PAPER
Trust, Perceived Useful, Attitude and Continuance Intention to Use
E-Government Service: An Empirical Study in Taiwan
Hau-Dong TSUI†a) , Member

SUMMARY According to the official TDOAS 2009∼2017 survey, the first step towards wider and more meaningful government
penetration rate of social media in Taiwan has reached a record 96.8%, services. While, in continuance use behavior, trust is even
while the Internet access rate is as high as 99.7%. However, people us-
more important since e-government service delivery have
ing government online services access to relevant information has con-
tinued to decline over the years, from 50.8% in 2009 to 35.4% in 2017. an exclusive and irreplaceable status. Assuming that cit-
At the same time, the proportion of e-transaction users has also dropped izens lack of basic trust on government portals, users are
simultaneously from 30.3% to 27.7%. In particular, only 1.1% of them more likely to adopt informal channels such as social me-
are interested in government online forums, while the remaining 97.2% dia or even traditional means to obtain incorrect or falsifi-
are more willing to engage in social media as a source of personal refer-
ence. The study aims to explore why are users not interested in accessing
cation data. Therefore, there is no doubt that trust is one
e-government services? Are they affected by the popularity of social net- of the main factors for the success of e-government initia-
working applications? What are the key factors for users to continue to use tives [50], [70]. The next step is not only to ensure that all
e-government service? The research framework was adapted from expec- stakeholders are able to apply these tools, but its success
tation confirmation theory and model (ECT/ECM), technology acceptance
also depends on citizen satisfaction and the continued use of
model (TAM) with trust theories, in validating attitude measures for a bet-
ter understanding of continuance intention of using e-government service. these government online services.
In terms of measurement, the assessment used the structural equation mod- Taiwan National Development Council (TNDC) con-
eling method (SEM) to explore the views and preferences of 400 college ducts “Digital Opportunity Annual Survey (TDOAS)” since
students on e-government service. The study results identified that per- 2009 [79]. We have noticed from TDOAS report that
ceived usefulness not only plays a full mediating role, it is expected to be
the most important ex-post factor influencing user’s intention to continue
as early as 2009, the preliminary e-government informa-
using e-government service. It also clarifies that the intent to continue to tion and e-transaction access rate have reached a medium
use e-government services is not related to use any alternative means such level of 50.8% and 30.3% respectively. In the same year,
as social media application. Taiwan also ranked 8th in the Waseda - IAC International e-
key words: ECT, ECM, SEM, E-government, social media
Government Rankings Survey in Japan. However, these ba-
sic functions can be attributed to passive e-government ser-
1. Introduction vices, which should be regularly compared to other proac-
tive applications or services (such as social media penetra-
E-government initiative shall responsible for supporting tion and e-commerce adoption) to achieve a broader per-
transformation towards a sustainable and resilient society. spective and progress. Therefore, in addition to the ini-
For the past two decades, developing, engaging and retain- tial user acquisition, another key issue that arises here is
ing citizens in using e-government online services is being whether they (adopters) continue to use e-government ser-
recognized as a effectively means for reshaping governance vices. From the perspective of the national information in-
in the areas of public sector, such as transparency, efficiency, frastructure (NII) and customer (citizen) services orienta-
accountability, and civic engagement. This has driven most tion, users’ continuance behavior is the best guarantee for
governments to continue to use new IS and IT to create un- the success of e-government development.
precedented communication channels and expand interac- In essence, e-government service provision has exclu-
tion with the public [63]. The UN’s 2018 e-government sur- sively nature with non-substitutability, which drives our re-
vey also disclosed this developing trend continues to grow search interests to explore, why the most basic and irre-
globally, and it can be attributed to the large increase in gov- placeable e-government service access rate continue to drop
ernment open data and effective online delivery channels, sharply from 50.8% in 2009 to 35.4% in 2017? Why G2C
especially through social media networks [56]. e-transaction services without accompanied by the booming
The purpose of an e-government is to provide good of e-commerce, but also continues to drop from 30.3% in
public service and credible information to citizens, deploy 2009 to 27.7% in 2017? Why are users not interested in
those dedicated online deliberation instruments is only the accessing e-government services? Are they affected by the
Manuscript received January 18, 2019. popularity of social media applications? What are the key
Manuscript revised July 3, 2019. factors for users to continue to use e-government service? In
Manuscript publicized September 24, 2019. the same scenario, social media penetration rate has reached
†
The author is with the Department of Information Manage-
ment, Chinese Culture University, Taipei, 11114 Taiwan. a record 96.8% (please see Appendix A.2).
a) E-mail: chd2@faculty.pccu.edu.tw
DOI: 10.1587/transinf.2019EDP7055

Copyright 
c 2019 The Institute of Electronics, Information and Communication Engineers
TSUI: TRUST, PERCEIVED USEFUL, ATTITUDE AND CONTINUANCE INTENTION TO USE E-GOVERNMENT SERVICE: AN EMPIRICAL STUDY IN TAIWAN
2525

institution-based trust into structural assurances and situa-

2. Theoretical Framework and Hypotheses tional normality beliefs, and both are considered to posi-
tively influence online shoppers’ trust in e-commerce set-
2.1 Trust Dimensions in E-Government Service ting [20]. This leads to the first hypothesis:
H1: Trust in e-government service is positively associated
E-government mainly refers to the use of the Internet and
with the attitude toward users’ intention to continue using
information communication technologies to improve the ef-
that service.
ficiency and effectiveness of public service delivery [64].
The UN 2016 e-government report also affirmed the imple- McKnight’s et al. and Gefen et al. both integrated TAM
mentation of e-governments made a significant contribution constructs with trust theories in validating trust measures for
to the stakeholders’ interests, and deepened the governance e-commerce [20], [67]. McKnight’s model aims to discover
process [56]. In this case, in order to be able to take advan- the antecedents (perception) associated with initial trust,
tage of the potential benefits of e-government service, gov- while Gefen’s model focuses on examining factors (mea-
ernment must ensure that citizens are willing to accept and sures) that influence continuous trust (knowledge-based
use it [26], [27], [38], [48]. However, it is undeniable that trust and calculative-based trust). However, their studies
the strength of trust beliefs in sharing information with gov- confirm that institution-based trust is expected to be the most
ernment agencies for online transactions may create obsta- significant element used to examine online trust. In the con-
cles and concerns about their use of e-government services. text of e-government or e-commerce environment, trust is
Fear could emerge from transmitting process, which is sub- critical to the initial use or continued use of online behavior,
ject to the uncertainty of risky cyber domains [49]. In e- both grounded on the development of information and com-
government context, trust for continued usage is even more munication technologies. In essence, e-government service
important as citizens are unlikely to find alternative online provision has exclusively nature with non-substitutability
service at the same purpose. The intention of trust probably and contrary e-commerce is substitutable [61]. The impact
be weakened, especially based on the expectation that the of trust intention on e-government adoption goes far beyond
other will perform a particular action important to the user the scope of e-commerce.
in vulnerable situations [20], [65], [66].
Trust should not be considered as just individuals’ be- 2.2 Perceived Usefulness in TAM
liefs about the other party only or technology itself, it should
involve beliefs about the Internet service (ISP) or content The technology acceptance model was introduced by Davis,
provider (ICP) who operate and administrate these tech- Bagozzi and Warshaw in 1989, which derived from the the-
nological tools [62]. Most of the later studies about e- ory of reasoned action (TRA) [17], and until now is one of
government or e-commerce were suggested, that the inten- the most widely used theories in IS literature [13]. The pur-
tion to adopt online applications should involve two ma- pose of this model is to predict the acceptability of a spe-
jor concerns: trust in the web service provider and the cific system and to identify the modification factors. TAM’s
electronic service medium [6], [44], [45]. McKnight et al. framework replaces TRA’s attitude measures with two par-
adopted three major components of trusting intentions: dis- ticular beliefs, perceived usefulness and perceived ease of
position to trust, institution-based trust, and trusting beliefs use, which in turn influence intended use of a technology
on e-commerce study [67]. The research framework de- and impacts the actual behavior of system usage [8], [12].
composed institution-based trust into structural assurance Perceived usefulness is the degree to which a user thinks a
and situational normality dimensions. The first represents technology would enhance performance or productivity in
an individual’s belief about having safe structures, such as the workplace. It effectively reflects the function of using
law enforcement, legal protection, government regulation IS and measure its final performance, productivity, effec-
and technical safeguards to ensure a safer network environ- tiveness, and efficiency. Perceived ease of use is the de-
ment. The second dimension represents a normal and fa- gree of lack of effort required by the user in adopting a
vorable condition of individual’s perception towards service given technology [55]; it will measure the extent of self-
providers equipped with competence and integrity charac- efficacy, including ease of learning, understanding, manip-
teristics. McKnight’s research finding shows that the per- ulation, and control. Perceived usefulness and perceived
ception of institution-based trust and trusting beliefs as both ease of use, both are used as significant beliefs to influence
important factors in trusting the online environment [67]. his or hers intention to adopt a newly developed technol-
Gefen’s et al. in 2003 integrating trust and TAM model ogy [12], [51], [52].
to exam three antecedents of trust construct: knowledge- The TAM model demonstrates the direct linkage be-
based trust, calculative-based trust and institution-based tween perceived usefulness and behavioral intention to use
trust. Knowledge-based trust is stems from previous pos- which is stronger than perceived ease of use [10]. It repre-
itive contact with a specific web vendor that leads to in- sents that perceived usefulness act as a major determinants
crease trust over time. Whereas, calculative-based trust character will influence actual system use [12]. However,
represents shoppers’ perceptions to web vendors’ honesty subsequent studies have also confirmed that perceived use-
and secure online environment. The study further divide fulness significantly affects individual attitudes during ante
 IEICE TRANS. INF. & SYST., VOL.E102–D, NO.12 DECEMBER 2019
2526

and post acceptance stages of IS use, while perceived ease and contradictory concept of satisfaction, leading some au-
of use is non-significant associated with attitude specifically thors to regard satisfaction as a synonym for attitude. Satis-
in post stage [12], [58]. Dillon and Morris (1996) pointed faction is a short-term impact of specific experience, and at-
out that ease of use could have a direct impact on perceived titude is a relatively more lasting impact beyond all previous
usefulness, if two systems offering the same features, a user experience. In addition, ECT’s framework proposes a vague
will find more useful the one that he finds easier to use [15]. concept of expectation construct and therefore some studies
Summarizing the preceding discussion, we hypothesize: define expectation with pre-consumption beliefs, while oth-
ers define expectations as the individual beliefs or a sum-
H2: The perceived usefulness of an e-government service is
mary of these beliefs [58].
positively associated with the attitude towards users’ inten-
In this context, the following Bhattacherjee’s expec-
tion to continue using that service.
tation confirmation model (ECM) was mainly drawn upon
H3: The user’s attitude towards e-government service is ECT, the technology acceptance model (TAM), and the the-
positively associated with the intention to continue using ory of planned behavior (TPB) [1], [12], [21]. Until now,
that service. ECM is broadly applied to examine the continuance inten-
tion of IS users rather than just to explain satisfaction [60].
Opposite to TAM, TRA is mainly used to predict and
In ECM, satisfaction measures the individual’s psycholog-
understand human behavior, whereas TAM is applied in the
ical or affective state which is caused by previous cogni-
field of ICTs boundary seeks to explain and predict the crit-
tive assessments. It focuses on three main variables in-
ical factors that influence personal attitude to accept new
clude expectation, satisfaction, and confirmation in deter-
system [34], [39], [42]. TAM also provides a complete the-
mining continued usage intention. Confirmation indicates a
oretical framework to understand how the external variables
cognitive belief of an individual’s initial expectation about
to shape one’s cognitive beliefs, the attitude of use, be-
the performance of a specific system is being confirmed af-
havioral intention and other constructs that affect personal
ter having use experience, then form a level of satisfaction
behavior [20], [46]. However, TRA and TAM both have
based on their degree of confirmation and expectation, the
strong behavioral elements exist, assume that when some-
pre-consumption expectation in ECT is replaced by ECM’s
one forms an intention to act that they will be free to act
post-consumption expectations. The ex-post expectation is
without any limitation [43]. But in the real world may exist
represented by perceived usefulness, since perceived useful-
many constraints of subjective normative and exterior influ-
ness is a cognitive belief salient to IS use, which is believed
ences, which limited personal freedom to act [4].
to contribute a more meaningful dimension to explore at-
titude toward satisfaction and intention to continue using a
2.3 Expectation Confirmation Theory and Model specific product or service. Theoretically, expectations con-
struct may contain other beliefs, while perceived usefulness
The expectation confirmation theory (ECT) is a theory based in the IS continuance context is the only belief in TAM,
on cognitive dissonance theory, which being widely used in which continues to influence user’s intention [12], [58], [68].
the consumer marketing and behavior research [33], [69]. It Although ECM was mainly drawn upon ECT, ECM’s
seeks to identify what determines the individual consumer’s framework is distinct from ECT and others technology ac-
satisfaction and whether they will repurchase the product or ceptance models, since initial adoption does not automati-
service [57]. ECT model proposes a five-stage of decision- cally result in continued use behavior. Bhattacherjee sup-
making process for consumer repurchase intention, it as- plements that perceived usefulness just as the cognitive be-
suming that expectations and perceived performance will lief in IS acceptance contexts, those continuance contexts
lead to post-purchase satisfaction, which is mediated by of confirmation and perceived usefulness may also be re-
positive or negative disconfirmation between expectations lated [58]. From the cognitive dissonance theory, confirma-
and perceived performance [76]. Individual’s comparison tion will tend to promote users’ perceived usefulness and
of previous expectations of a certain commodity, such as disconfirmation will decrease such perceptions [58], [69];
a specific product or service with subsequent perceived per- which imply those two arguments are essentially consistent
formance will reflect his or her satisfaction after purchase, with each other. Stone and Baker adapted ECM to exam e-
and eventually forms a positive or negative disconfirmation book users’ continuance intentions [59]. The research find-
of beliefs. Oliver (1980) pointed out, positive disconfirma- ing shows confirmation is positively associated with per-
tion will lead to a high degree of satisfaction, while negative ceived usefulness and satisfaction, while satisfaction and
disconfirmation will lead to consumer’s dissatisfaction [57]. continued intention to use e-book are both positively related
Compared to the initial adoption of behavioral patterns to perceived usefulness. Hence, the context of preceding
such as TAM or TPB, the framework of ECT can be re- discussion leads to the following hypotheses:
garded as a dynamic process model for post-action research.
H4: The level of confirmation with an e-government service
However, Bhattacherjee argued that ECT mainly ignores po-
is positively associated with perceived usefulness of using
tential changes in consumers’ expectations after using new
that service.
products, and the impact of these changes on subsequent
cognitive processes. In essence, ECT presents a different H5: The level of confirmation with an e-government service
TSUI: TRUST, PERCEIVED USEFUL, ATTITUDE AND CONTINUANCE INTENTION TO USE E-GOVERNMENT SERVICE: AN EMPIRICAL STUDY IN TAIWAN
2527

is positively associated with users’ satisfaction toward that are coded as a five-point Likert scale from “strongly dis-
service. agree” to “strong agree”. According to the research proce-
dure, the study firstly conducted pre-test on the question-
Limayem and Cheung expanded the ECM framework
naire through two professors and three graduate students to
by adding a moderator of habit to IS continuance intention
check whether the questionnaire items were clear and easy
and IS continued use. The results have a strong support to
to understand without any unclear wording or ambiguity, to
ECM and moderator, both satisfaction and prior cognitive
ensure the measurement variables’ reliability and validity.
attitude to perceived usefulness were found to have posi-
After removing 3 low reliability items from 27 questions in
tively associated with IS continuance intention [53], [61].
total, the analysis showed that the total Cronbach’s Alpha
Teo, Srivastava and Jiang in 2008 conducted an empirical
value exceeded the standard 0.7 [80]. It indicates the hy-
study of trust and e-government success. The authors indi-
pothesis model has a parsimonious framework and is con-
cate that the intention to continue using e-government web
sistent with the previous research context; the measurement
site is similar to a user’s revisit specific commercial web
of the scale is applicable to the next phase of formal survey.
site. Especially, in the post-adoption stage, user’s inten-
The sample allocation of formal survey was conducted
tion to continue using a web site follows a repeat behav-
by stratified proportional sampling method based on the
ior and is influenced by the feedback mechanism. Through
2018 demographic data of the Ministry of Interior (MOI)
the e-government service feedback mechanism, satisfaction
Taiwan [81] (Please see Appendix A.1). The survey se-
as an affective state formed from prior experience, which
lected 400 PCCU business college students based on their
will influence user’s intention to continue using [61]. Ex-
trapolating from these arguments, we propose the following
research hypotheses: Table 1 Questionnaires items and sources.

H6: Perceived usefulness of an e-government service is

positively associated with users’ satisfaction toward that
service.
H7: Perceived usefulness of an e-government service is pos-
itively associated with users’ intention to continue using that
service.
H8: The level of satisfaction with an e-government service
is positively associated with intention to continue using that
service.

3. Method

The study adapted validated scales from existing literature.

The original questionnaire consisted of 27 questions which
attributed to six constructs, and the corresponding answers

Fig. 1 Proposed model and hypotheses.

 IEICE TRANS. INF. & SYST., VOL.E102–D, NO.12 DECEMBER 2019
2528

Table 2 Descriptive statistics of factor loading, CR, and AVE.

residency attribution and e-government experience. The
largest number of subjects from New Taipei City, account-
ing for 68 person; the smallest group was Kinmen County,
accounting for 2 out of 400 students (please see attached
sample allocation chart). Based on the time of the subject’s
class, the survey is conducted in a timely manner and ex-
plained in detail to avoid confusion and to ensure the va-
lidity of each item. Therefore, the survey results are in line
with the expected goals (time, controllability, resources) and
achieve an effective response rate of 100%.
Since E-government adopters in Taiwan are mainly dis-
tributed between the ages of 20 and 39, especially under 29
that accounting for 38.5% of the total users. Therefore, the
young respondents of the business college used in this study
not only meet the outline of TDOAS survey [79], but also
prevented potential biases from providing a lively and effec-
tive scale for our research.
The assessment tool adopts structural equation model
(SEM) technique to identify the critical factors of users’ in-
tention to continue using e-government service. Boudreau,
Gefen and Straub in “The validation in information sys-
tems research” pointed out, the advantage of using the SEM
method is that measurement and structural models can be in-
corporated into simultaneous analysis [16], [73]. The mea-
surement model includes items loading that are expected to Table 3 Correlation table.
converge on their proposed constructs. Whereas, structural
model involves examining the proposed hypotheses among
theoretical framework and providing an overall model fit.
Consequently, SEM is commonly justified in the social sci-
ences being capable of impute relationships between un-
observed constructs from observable variables [41], [72].
Based on the above, this paper uses quantitative methods
and a SEM assessment tools to understand the intent to
continue using e-government service should be appropri- Larcker (1981) has been commonly used to assess the de-
ately [71]. gree of shared variance between the latent variables of the
model, standardized factor loadings, average variance ex-
4. The Measurement Model tracted (AVE) and composite reliability (CR) [20]. Stan-
dardized factor loadings are representative of the level of
After the formal survey, the assessment of normality shows association between scale items and a single latent vari-
that all variables are conforms to the suggested standard able. AVE measures the level of variance captured by a
values, skewness < 2, and kurtosis < 7 [19], [36]. Multi- construct versus the level due to measurement error, and CR
variate kurtosis c.r. value 73.45 is also in line with Bollen’s is a less biased estimate of reliability than Cronbach’s Al-
(1993) suggested threshold [2], [3], [7], [47]. The prelim- pha value. The assessment (Table 2) shows the standardized
inary offending test illustrates all standard errors of vari- factor loadings of items to the corresponding variable are
ances are range from 0.01 to 0.06 of significance, and fac- highly significant that range from 0.70-0.90. CR on trust
tor loadings without exceeding or very close to value 1.0. is 0.890, perceived usefulness is 0.932, attitude is 0.878,
Further, as part of justifying in measurement model, the confirmation is 0.866, satisfaction is 0.868, and intention to
study will proceed with confirmatory factor analysis (CFA) continue using is 0.854. AVE values are 0.669, 0.775, 0.643,
to verify convergent validity and discriminant validity of the 0.619, 0.624, and 0.594 respectively. The convergent valid-
scale [18], [28], [37]. Convergent validity refers to the de- ity in three tests were supported, it indicates the observed
gree to which two measures of constructs that theoretically variables have good explanatory to the variation of latent
should be related are in fact related. In contrast, discrim- variables and a highly degree of reliability [9], [20], [31].
inant validity tests whether concepts or measurements that For discriminant validity, we use AVE analysis to see if
are supposed to be unrelated are in fact unrelated. Both the square root of every AVE value belonging to each latent
types of validity are a requirement for excellent construct construct is much larger than any correlation among any pair
validity [9], [14], [29]. of latent constructs [11], [20]. Through the correlation ma-
For convergent validity, the criterion of Fornell and trix, Table 3 shows the factor correlation between pairs of la-
TSUI: TRUST, PERCEIVED USEFUL, ATTITUDE AND CONTINUANCE INTENTION TO USE E-GOVERNMENT SERVICE: AN EMPIRICAL STUDY IN TAIWAN
2529

tent variable are less than the square root of the correspond- found to be significantly associated with attitude, which ex-
ing AVE. It indicates the discriminant validity was satisfied plaining R2 = 0.44 of citizen’s perceptions. H2 is supported
for every variable of the proposed model [22]–[24], [26]. (path = 0.69, t = 9.11, p < 0.001).
However, perceived usefulness is also found to be sig-
5. The Structural Model nificantly associated with the intention of continue using e-
government service. H7 is supported (path = 0.34, t = 4.75,
After proposing the modification process of the model, Ta- p < 0.001, R2 = 0.17). The other five paths were not found
ble 4 shows three fit indices including the absolute fit in- to be significant: trust to attitude, perceived usefulness to
dices, incremental fit indices, and parsimonious fit indices, satisfaction, confirmation to satisfaction, satisfaction to in-
all meet the recommended thresholds [32]. The following tention, and attitude to intention.
causal path analysis in Fig. 2, using SEM maximum like- As for the research question Q1 mentioned, whether
lihood estimation to exam the path coefficients of the pro- the low access rate of e-government service is affected by
posed model [30]. The estimates of regression weights indi- the popularity of social media application? The estimate
cate the level of confirmation with an e-government service of regression weights show that intention in the prediction
is significantly associated with perceived usefulness of using of questionnaires items ICU2 and ICU3, both were signif-
that service. Hence, H4 is supported (path = 0.27, t = 4.53, icantly different from zero at the 0.001 level (p < 0.001).
p < 0.001, R2 = 0.07). Further, perceived usefulness is It clarifies that the intent to continue to use e-government
services is not related to use any alternative means such as
social media application.
Table 4 The model fit and criteria of structural model. In addition, we also compare Bhattacherjee’s Expecta-
tion Confirmation Model (ECM) by removing the trust and
attitude constructs from the proposed model. The results in-
dicate that the level of confirmation with an e-government
service is significantly associated with perceived usefulness
of using that service (path = 0.30, t = 5.32, p < 0.001).
While, perceived usefulness is also found to be signifi-
cantly associated with the intention of continue using e-
government service (path = 0.41, t = 7.45, p < 0.001).
From the perspective of cognitive dissonance theory, posi-
tive confirmation will lead to a high degree of cognitive be-
lief about the usefulness of e-government service and con-
tinue to use in the future, while disconfirmation will de-
crease such perceptions. This finding is consistent with
Bhattacherjee’s research, that perceived usefulness just as
the cognitive belief in IS acceptance contexts, those in IS
contexts of confirmation and perceived usefulness may also
be related.

Fig. 2 Path coefficients and hypotheses test. Fig. 3 Expectation confirmation model (ECM) test.
 IEICE TRANS. INF. & SYST., VOL.E102–D, NO.12 DECEMBER 2019
2530

associated with perceived usefulness of using e-government

6. Discussions and Implications service, and perceived usefulness is also found to be sig-
nificantly associated with the intention of continue that ser-
6.1 Discussions vice. The Bhattacherjee’s model shows the robustness of
perceived usefulness, which is directly related to the con-
The study conducted an in-depth analysis to explore why firmation and the continuance of intention. Encouragingly,
users are not interested in accessing e-government online our findings is fully in line with Bhattacherjee’s assertion
services? Are they affected by the popularity of social me- that positive confirmation will lead to a high degree of cog-
dia applications? What are the key factors for users to con- nitive belief in the usefulness. From cognitive dissonance
tinue to use e-government service? The research frame- theory viewpoint, confirmation will lead to promote users’
work was adapted from expectation confirmation theory and perceived usefulness and disconfirmation will decrease such
model (ECT/ECM), technology acceptance model (TAM) perceptions [69]. Both models were explicitly define the key
with trust theories, in validating attitude measures for a position of the perceived usefulness and the degree of con-
better understanding of continuance intention of using e- firmation [58], and the user’s intention to continue using the
government service. system is largely determined by the power of this belief.
However, the theoretical expansion of a preliminary
model usually requires an empirical test of its robustness 6.2 Implications for Practice
and predictability in a broader context. Therefore, based on
the purpose of this study, we propose some modifications Since the study found that perceived usefulness plays an
and extensions to the original IT continuity model. The re- important role in the user’s intention to continue to use e-
search posits four exogenous variables which namely: trust government services. Therefore, we proposed some practi-
(in e-government service), attitude (using e-government ser- cal implications, particularly those concerned with attitude
vice), perceived usefulness (e-government service perfor- and the continuance intention.
mance), confirmation (comparison of previous expectations First, the perceived usefulness of an e-government ser-
and post hoc performance) and satisfaction (the outcome), vice is positively associated with the attitude towards users’
to explore the key factors for users’ intention to continue to intention to continue using that service. People may be in-
use e-government service. clined to give high evaluations and comments on the de-
Through the model test, we found that confirmation is ployment of new technologies or innovative applications,
significantly associated with perceived usefulness of user’s if the user finds that service is as useful as perceived then
initial expectation about the performance of e-government will forms a positive attitude towards reusing. For those
service. Furthermore, perceived usefulness is found to be continuances behavior contexts, this cognitive process may
significantly associated with attitude and intention of con- entail the recall of previous experience or the evaluation
tinue using e-government service. It indicates that perceived of the objects perceived, and then aggregate to each cor-
usefulness plays a major role of full mediation [25], [35] for responding cognitive belief. However, some external vari-
direct catalytic effect on the users’ intention to continue to ables may have a significant impact on the portfolios of
use e-government service. On the contrary, the effects of trust, attitude, self-efficacy, and perceived behavioral con-
attitude on intention of continue using is found to be non- trol [61], [77], [78], which encompassing cybersecurity, le-
partially mediated. While the other five paths including: gal protection, facilitating conditions and innovative appli-
trust to attitude, perceived usefulness to satisfaction, confir- cation issues. In practice, it is difficult to construct a cogni-
mation to satisfaction, satisfaction to intention, and attitude tive belief in attitude just like a mediator, trying to find out
to intention were found not significant. his or her inner psychology [5], [40].
In addition to these findings, the most important thing Second, perceived usefulness of an e-government ser-
is that we also confirm that the intention of continue us- vice is positively associated with users’ intention to con-
ing e-government service is not related to use any alterna- tinue using that service. It indicates perceived usefulness
tive means such as social media application. This implies is expected to be the most salient ex-post factor influenc-
even though an individual’s attitude can be influenced by the ing user’s continuance behavior. However, Bhattacherjee
overall perception to access such services and beneficial to et al. point out that sometimes users’ perceptions of use-
the users; but in essence, the continuance acts still depend- fulness will directly drive their continued intent, because
ing on its post-stage performance of usefulness. Several they find that the service meets their needs, then it is pos-
empirical studies support this argument regarding the rela- sible to bypass the validation process and continue to use
tionship between perceived usefulness and attitude to adopt it directly [58]. Similarly, in our study, the measurements
new systems or services. Whereas, for those connecting to also proved (identified) that perceive usefulness bypassed
the context of self-efficacy facilitating conditions, it perhaps the attitude path and was significantly related to the contin-
goes beyond our scope [51], [52], [54]. ued use of e-government service. In practice, compared with
The study further compared the proposed model with the e-commerce environment, online services and message
the original expected confirmation model (ECM). The ECM responses provided by government authorities are some-
test results confirm the level of confirmation is significantly times insufficient for the needs of citizens and lack conve-
TSUI: TRUST, PERCEIVED USEFUL, ATTITUDE AND CONTINUANCE INTENTION TO USE E-GOVERNMENT SERVICE: AN EMPIRICAL STUDY IN TAIWAN
2531

nient alternatives and delivery channels. Perhaps in Taiwan, government service is not related to the personal adoption
e-government services may regard as a symbol of digital au- of social media application nor any alternative means. Per-
thoritarian with mandatory and exclusive nature, for most ceived usefulness plays a full mediating role that directly
users this probably the only alternative to use [61]. catalyzes users’ intention to continue to use e-government
service.
6.3 Implications for Policy Although the survey data was collected from college
young respondents to meet the outline of the TDOAS sur-
Through this empirical study, we find that perceived use- vey [79], to prevent potential biases from providing a lively
fulness not only serves as a full mediating role, but is also and effective scale for our research; however, the ques-
expected to be the most important ex-post factor affect- tionnaire items may not be comprehensive enough to mea-
ing user’s intention to continue using e-government ser- sure their cognitive beliefs. The future research (if) could
vice. Overall, we still need to emphasize the success of take some other constructs into account, such as by adding
e-government initiatives is contingent upon citizen’s sat- certain facilitating conditions, self-efficacy or adapt per-
isfaction, willingness and continued usage of such gov- ceived risk issues to deepen the scale of measurement.
ernment online service, rather than simply establishing It’s obviously will getting more public concerns, especially
unique choices and irreplaceable service delivery in the at the critical moment of sustainable development of e-
bureaucracy. government services.
From a policy perspective, it clearly demonstrates the
importance of citizen perception, which focus on usefulness References
of online content and innovative technology implementation
[1] I. Ajzen, “The theory of planned behavior,” Org. Behavior and Hu-
over e-government service, especially for positive “useful” man Decision Processes, vol.50, no.2, pp.179–211, 1991.
feelings. Government should strive to create more accurate, [2] J.C. Anderson and D.W. Gerbing, “Structural equation modeling in
friendly, and functional web content, such as implements practice: A review and recommended two-step approach,” Psycho-
multi-level of social media connections at the public net- logical Bulletin, vol.103, no.3, pp.411–423, 1988.
work infrastructure, similar to cloud computing providing [3] R.P. Bagozzi and Y. Yi, “On the evaluation of structural equation
models,” Journal of the academy of marketing science, vol.16, no.1,
an “on demand” service models for the stakeholders.
pp.74–94, 1988.
Though, the initiative of e-government service advo- [4] R.P. Bagozzi, F.D. Davis, and P.R. Warshaw, “Development and test
cates “everything as a service”, policy makers should notice of a theory of technological learning and usage,” Human Relations,
in particular to their continuance intention under the premise vol.45, no.7, pp.660–686, 1992.
of sustainable development. As our research findings in- [5] R.M. Baron and D.A. Kenny, “The moderator-mediator variable dis-
dicate, the initial adoption does not automatically result in tinction in social psychological research: Conceptual, strategic, and
statistical considerations,” Journal of Personality and Social Psy-
continued use behavior, and its initial aggregated attitude chology, vol.51, no.6, pp.1173–1182, 1986.
can only serve as a partial mediating role to strengthen their [6] F. Bélanger and L. Carter, “Trust and Risk in e-Government Adop-
cognitive beliefs. Instead, perceived usefulness directly cat- tion,” J. of Strategic Info. Systems, vol.17, no.2, pp.165–176, 2008.
alyzes the user’s intent to continue to use e-government [7] K.A. Bollen and R.A. Stine, Bootstrapping goodness-of-fit measures
services. in SEM: Testing structural equation models, K.A. Bollen and J.S.
Long, Eds., pp.111–135, Sage, 1993.
[8] R.E. Burnkrant and J.P. Jr. Thomas, “The structure and antecedents
7. Conclusions and Limitations of the normative and attitudinal components of Fishbein’s theory
of reasoned action,” Journal of Experimental Social Psychology,
This paper emphasizes the importance of continuous use of vol.24, pp.66–87, 1988.
specific information systems behavioral research. From a [9] D.T. Campbell and D.W. Fiske, “Convergent and discriminant val-
practical point of view, regardless of the different hypotheti- idation by the multitrait–multimethod matrix,” Psychological Bul-
letin, vol.56, no.2, pp.81–105, 1959.
cal models and analytical methods, relevant research results [10] P.-Y. Chu, N. Hsiao, F.-W. Lee, and C.-W. Chen, “Exploring success
can be used for further exploration, especially in connection factors for Taiwan’s government electronic tendering system: behav-
with previous theoretical and practical verification. ioral perspectives from end users,” GIQ, vol.21, no.2, pp.219–234,
The result indicates that perceived usefulness is the 2004.
only factor that to be significantly associated with the in- [11] J. Cohen, P. Cohen, S.G. West, and L.S. Aiken, Applied Multi-
tention of continue using e-government service; and it also regression: Correlation Analysis for the Behavior Sciences, 3rd ed.,
Lawrence Erlbaum Associates, 2003.
found to be significantly associated with the attitude. The [12] F.D. Davis, “Perceived usefulness, perceived ease of use, and user
third supported path is the correspondence between con- acceptance of information technology,” MIS Quarterly, vol.13, no.3,
firmations of an individual’s initial expectation about the pp.319–340, 1989.
performance of e-government service to the ex-post eval- [13] F.D. Davis, R.P. Bagozzi, and P.R. Warshaw, “User acceptance of
uation of perceived usefulness. As other five casual paths computer technology: a comparison of two theoretical models,”
Management Science, vol.35, no.8, pp.982–1003, 1989.
are not supported including: trust to attitude, attitude to in-
[14] R.F. DeVellis, Scale Development: Theory and applications, 2nd
tention, perceived usefulness to satisfaction, confirmation to ed., SAGE Publication, 2003.
satisfaction, and satisfaction to intention. Most importantly, [15] A. Dillon and M.G. Morris, “User acceptance of information tech-
this study clarifies that the intention of continue using e- nology: Theories and models,” Annual Review of Information Sci-
 IEICE TRANS. INF. & SYST., VOL.E102–D, NO.12 DECEMBER 2019
2532

ence and Technology, vol.31, pp.3–32, 1996. blogs: Lifestyle and purchase behavior,” The Journal of Advertising
[16] B. Efron, “Better Bootstrap Confidence Intervals,” Journal of the and Public Relations, vol.32, pp.91–123, 2009.
American Statistical Asso., vol.82, no.397, pp.171–185, 1987. [40] K.J. Preacher and A.F. Hayes, “Asymptotic and resampling strate-
[17] M. Fishbein and I. Ajzen, Belief, Attitude, Intention and Behavior: gies for assessing and comparing indirect effects in multiple media-
An Introduction to Theory and Research, Addison-Wesley, Reading, tor models,” Behavior Research. Methods, vol.40, no.3, pp.879–891,
Massachusetts, 1975. 2008.
[18] C. Fornell and D.F. Larcker, “Evaluating structural equation mod- [41] Y. Reisinger and T. Lindsay, “Structural equation modeling, SEM
els with unobservable variables and measurement error,” Journal of with Lisrel: application in tourism,” Tourism Management, vol.20,
marketing research, vol.18, no.1, pp.39–50, 1981. no.1, pp.71–88, 1999.
[19] S. Gao, P. Mokhtarian, and R. Johnston, “Nonnormality of data in [42] E.M. Rogers, Diffusion of Innovations, 5th ed., Simon and Schuster
SEM transportation research record,” Journal of the Transportation Publishing, 2003.
Research Board, vol.2082, pp.116–124, 2008. [43] A.B. Sahli and L. Patrick, “Sing the DTPB to explain the intention
[20] D. Gefen, E. Karahanna, and D. Straub, “Trust and TAM in on- to book tourism products online,” International Journal of Online
line shopping: an integrated model,” MIS Quarterly, vol.27, no.1, Marketing, vol.4, no.1, pp.1–10, 2014.
pp.51–90, 2003. [44] L.C. Schaupp, L. Carter, and M.E. McBride, “E-file adoption: A
[21] F.G. Joey, “The theory of planned behavior and Internet purchasing,” study of U.S. taxpayers’ intentions,” Computers in Human Behavior,
Internet Research, vol.14, no.3, pp.198–212, 2004. vol.26, no.4, pp.636–644, 2010.
[22] J.F. Hair Jr, R.E. Anderson, R.L. Tatham, and W.C. Black, Multi- [45] P.G. Schierz, O. Schilke, and B.W. Wirtz, “Understanding consumer
variate Data Analysis, 5th ed., Prentice-Hall, 1998. acceptance of mobile payment services: An empirical analysis,”
[23] L.L. Hallow, Behavior of Some Elliptical Theory Estimators with E-Com. Research and App., vol.9, no.3, pp.209–216, 2010.
Non-normality Data in a Covariance Structures Framework: A [46] P.E. Shrout and N. Bolger, “Mediation in experimental and nonex-
Monte Carlo study, Ph.D. Dissertation, UCLA, 1985. perimental studies: New procedures and recommendations,” Psy-
[24] J.F. Hair Jr, W.C. Black, B.J. Babin, R.E. Anderson, and R.L. chological Methods, vol.7, no.4, pp.422–445, 2002.
Tatham, Multivariate Data Analysis, 6th ed., Prentice-Hall, 2006. [47] J.P. Stevens, Applied multivariate statistics for the social sciences,
[25] A.F. Hayes, “Beyond Baron and Kenny: Statistical mediation anal- 3rd ed., Lawrence Erlbaum Associates, 1996.
ysis in the new millennium,” Communication Monographs, vol.76, [48] K. Sugiyama and T. Andree, The Dentsu Way: Secrets of Cross
no.4, pp.408–420, 2009. Switch Marketing from the World’s Most Innovative Advertising
[26] C.-L. Hsu, H.-P. Lu, and H.-H. Hsu, “Adoption of the mobile In- Agency, McGraw-Hill Education, 2010.
ternet: An empirical study of multimedia message service (MMS),” [49] U. Sumita and R. Sogai, “Development of e-marketing contract
Omega, vol.35, no.6, pp.715–726, 2007. structure based on consumer-generated contents and its optimal
[27] S.-Y. Hung, C.-M. Chang, and T.-J. Yu, “Determinants of user ac- strategy,” Department of Social Systems and Management Discus-
ceptance of the e-Government services: The case of online tax filing sion Paper Series, University of Tsukuba, 2009.
and payment system,” Gov. Info. Quarterly, vol.23, no.1, pp.97–122, [50] M. Tan and T. Teo, “Factors influencing the adoption of internet
2006. banking,” Journal of the Association for Information Systems, vol.1,
[28] D.L. Jackson, J.A. Gillaspy Jr, and R. Purc-Stephenson, “Reporting no.5, pp.1–42, 2000.
Practices in confirmatory factor analysis: An Overview and some [51] S. Taylor and P. Todd, “Decomposition and crossover effects in the
Recommendations,” Psychological Methods, vol.14, no.1, pp.6–23, theory of planned behavior: a study of consumer adoption inten-
2009. tions,” International Journal of Research in Marketing, vol.12, no.2,
[29] K.G. Jöreskog, “A general approach to confirmatory maximum like- pp.137–155, 1995a.
lihood factor analysis,” ETS Research Bulletin Series, vol.1967, [52] S. Taylor and P.A. Todd, “Understanding information technology
no.2, pp.183–202, 1967. usage: A test of competing models,” Information Systems Research,
[30] K.G. Jöreskog and D. Sörbom, Analysis of Linear Structural Re- vol.6, no.2, pp.144–176, 1995b.
lationships by Maximum Likelihood, Instrumental Variables, and [53] T.S.H. Teo and S.H. Pok, “Adoption of WAP-enabled mobile phones
Least Squares Methods LISREL VI, 7th ed., 2000. among Internet users,” Omega, vol.31, no.6, pp.483–498, 2003.
[31] D.A. Kenny and C.M. Judd, “Estimating the nonlinear and interac- [54] L.R. Vijayasarathy, “Predicting consumer intentions to use on-line
tive effects of latent variables,” Psychological Bulletin, vol.96, no.1, shopping: the case for an augmented technology acceptance model,”
pp.201–210, 1984. Journal of Information & Management, vol.41, no.6, pp.747–762,
[32] D.A. Kenny and D.B. McCoach, “Effect of the number of vari- 2004.
ables on measures of fit in structural equation modeling,” Struc- [55] P.S. Wei and L.H. Peng, “An examination of the celebrity endorse-
tural Equation Modeling: A Multidisciplinary Journal, vol.10, no.3, ments and online customer reviews influence female consumers’
pp.333–351, 2003. shopping behavior,” Computers in Human Behavior, vol.29, pp.193–
[33] P. Kotler, H. Kartajaya, and I. Setiawan, Marketing 3.0: From Prod- 201, 2013.
ucts to Customers to Human Spirit, Wiley & Sons, 2010. [56] UN, E-Government Survey 2018, Department of Economic and So-
[34] C. López-Nicolás, F.J. Molina-Castillo, and H. Bouwman, “An as- cial Affairs (DESA), UN Publication, 2016.
sessment of advanced mobile services acceptance: Contributions [57] R.L. Oliver, “A Cognitive Model of the Antecedents and Conse-
from TAM and diffusion theory models,” Information & Manage- quences of Satisfaction Decisions,” Journal of Marketing Research,
ment, vol.45, no.6, pp.359–364, 2008. vol.17, no.4, pp.460–469, 1980.
[35] D.P. MacKinnon, Introduction to Statistical Mediation Analysis, [58] A. Bhattacherjee, “Understanding Information Systems Continu-
Lawrence Erlbaum Associates, 2008. ance: An Expectation-Confirmation Model,” MIS Quarterly, vol.25,
[36] K.V. Mardia, “Measures of multivariate skewness and kurtosis with no.3, pp.351–370, 2001.
applications,” Biometrika, vol.57, no.3, pp.519–530, 1970. [59] R.W. Stone and L. Baker-Eveleth, “Students’ expectation, confirma-
[37] B.D. Medlin, The factors that may influence a faculty member’s de- tion, and continuance intention to use electronic textbooks,” Com-
cision to adopt electronic technologies in instruction, Ph.D. disser- puters in Human Behavior, vol.29, no.3, pp.984–990, 2013.
tation, Virginia Poly Institute and State University, 2001. [60] M. Limayem and C.M.K. Cheung, “Understanding information
[38] T. O’Reilly, What Is Web2.0: Design Patterns and Business Models systems continuance: The case of Internet-based learning technolo-
for the Next Gen. of Software, O’Reilly Media, 2005. gies,” Information & Management, vol.45, no.4, pp.227–232, 2008.
[39] Y. Peng and S. Tao, “A study of information searchers on cosmetic [61] S.H.T. Teo, C.S. Srivastava, and L. Jiang, “Trust and electronic gov-
TSUI: TRUST, PERCEIVED USEFUL, ATTITUDE AND CONTINUANCE INTENTION TO USE E-GOVERNMENT SERVICE: AN EMPIRICAL STUDY IN TAIWAN
2533

ernment success: An empirical study,” Journal of MIS, vol.25, no.3,

pp.99–132, 2008. Appendix: Sample Allocation and TDOAS Survey
[62] A.I. Alzahrani, Web-based e-Government Services Acceptance for
G2C-A Structural Equation Modelling Approach, Ph.D. disserta- A.1 The Sample Allocation
tion, De Montfort University, 2011.
[63] C.H. Jeong, Fundamental of Development Administration, Table A· 1 Stratified proportional sampling (MOI 2018)
Selangor: Scholar Press, 2007.
[64] R.C. Mayer, J.H. Davis, and F.D. Schoorman, “An integrative model
of organizational trust,” Academy of Management Review, vol.20,
no.3, pp.709–734, 1995.
[65] Y.H. Kim, D.J. Kim, and Y. Hwang, “Exploring Online Transaction
Self-Efficacy in Trust Building in B2C E-Commerce,” Journal of
Organizational and End User Computing, vol.21, no.1, pp.37–59,
2009.
[66] M.K.O. Lee and E. Turban, “A trust model for consumer internet
shopping,” Int. Jour. of E-Com., vol.6, no.1, pp.75–91, 2001.
[67] D.H. McKnight, V. Choudhury, and C. Kacmar, “Developing and
validating trust measures for e-commerce: an integrative typology,”
Info. Systems Res, vol.13, no.3, pp.334–359, 2002.
[68] E. Karahanna, D.W. Straub, and N.L. Chervany, “Information tech-
nology adoption across time: A cross-sectional comparison of pre-
adoption and post-adoption beliefs,” MIS Quarterly, vol.23, no.2,
pp.183–213, 1999.
[69] L. Festlnger, A Theory of Cognitive Dissonance, Stanford Univer-
sity Press, 1957.
[70] M. Warkentin, D. Gefen, P.A. Pavlou, and G.M. Rose, “Encourag-
ing citizen adoption of e-government by building trust,” Electronic
Markets, vol.12, no.3, pp.157–162, 2002.
[71] A. Bryman, Social Research Methods, 2nd edition, Oxford Univer-
sity Press, 2004.
[72] G.R. Hancock, “Fortune Cookies, Measurement Error, and Exper-
imental Design,” Journal of Modern Applied Statistical Methods,
vol.2, no.2, p.293, 2003.
[73] M.-C. Boudreau, D. Gefen, and D.W. Straub, “Validation in infor-
mation systems research: A state-of-the-art assessment,” MIS Quar-
terly, vol.25, no.1, pp.1–16, 2001.
[74] R.L. Wakefield, M.H. Stocks, and W.M. Wilder, “The role of Web
site characteristics in initial trust formation,” Journal of Computer
Information Systems, vol.45, no.1, pp.94–103, 2004.
[75] J.F. George, “The TPA and Internet purchasing,” Internet Research,
vol.14, no.3, pp.198–212, 2004.
[76] L. Festinger, “Cognitive dissonance,” Scientific American, vol.207,
no.4, pp.93–107, 1962.
[77] A. Bhattacherjee, J. Perols, and C. Sanford, “Information technology
continuance: A theoretic extension and empirical test,” The Journal
of Computer Information Systems, vol.49, no.1, pp.17–26, 2008.
[78] A. Bhattacherjee, “An empirical analysis of the antecedents of elec- A.2 TDOAS Annual Survey
tronic commerce service continuance,” Decision Support Systems,
vol.32, no.2, pp.201–214, 2001. Table A· 2 Taiwan e-government service rate (2009–2017)
[79] TDOAS, “Taiwan digital opportunity annual survey,” National De-
velopment Council, Taiwan, 2017.
[80] J.M. Cortina, “What is coefficient alpha? An examination of the-
ory and applications,” Journal of Applied Psychology, vol.78, no.1,
pp.98–107, 1993.
[81] MOI, “Statistical Yearbook of Interior,” The Department of Statis-
tics, Ministry of Interior (MOI), Taiwan, 2018.
 IEICE TRANS. INF. & SYST., VOL.E102–D, NO.12 DECEMBER 2019
2534

Hau-Dong Tsui currently as an assis-

tant professor in the Department of Informa-
tion Management, Chinese Culture University,
Taipei, Taiwan. He received Ph.D. degree
of Political Science (e-Government) in PCCU
Taiwan, and MS degree from the University of
West Florida, USA. His research interests cover:
Electronic Government/Governance, Social Sci-
ence Methodology, E-Commerce, and Behav-
ioral Science Research.