Information & Management 61 (2024) 104037

Contents lists available at ScienceDirect

Information & Management

journal homepage: www.elsevier.com/locate/im

Identifying the peripheral cues in the credibility assessment of online

health information
Jennifer L. Claggett a,* , Brent Kitchens b , Maria Paino c
a
School of Business, Wake Forest University, Winston-Salem, NC, United States
b
McIntire School of Commerce, University of Virginia, Charlottesville, VA, United States
c
Department of Sociology, Anthropology, Social Work and Criminal Justice, Oakland University, Rochester, MI, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Online health information varies, as well as what people choose to consume and believe. Previous research finds
Source credibility that hesitancy to follow health advice is often due to suspicion about credibility. The elaboration likelihood
Online health information model suggests credibility assessments use both argument quality and source credibility. One important facet for
Elaboration likelihood model
understanding how and why people cling to misinformation about health advice is uncovering what drives their
Information framing
Webpage format
credibility assessment in the first place. Yet, little research focuses on what peripheral cues influence source
Online experiment credibility in online health information. Our mixed-method study, which combines an online experiment and
Mixed methods qualitative analysis, explores how source, tone, and format affect credibility perceptions in health contexts. The
results confirm ELM relationships and indicate credibility increases when information is physician-authored and
objectively presented. Our findings address a gap in the literature by exploring what influences a person’s
credibility assessment of online health information, offering insights that could inform the design of future online
health resources.

1. Introduction conversation with their physicians [11]. Physician opinions and expe-
riences with patients using the internet for health information remains
Extensive health information is available to consumersd online, and mixed – sometimes the patient is more informed and the consultation is
what people choose to consume and believe varies widely. For a recent therefore perceived as more efficient and effective, and sometimes the
and poignant example, consider the variation in people’s willingness to patient is misinformed or resistant to new ideas or diagnoses [12]. In this
receive a COVID-19 vaccine, and the many alternate beliefs online [1] vein, health literacy is concerned with the degree to which individuals
about the potential consequences of receiving this vaccine that helped have the ability to find, understand, and use information and services to
convince 20 % of the US population to abstain [2]. While the global inform health-related decisions and actions for themselves and others
COVID pandemic perhaps brought this phenomenon into focus, hesitant [13]. A recent report finds that in a 6 month period, 58 % of American
attitudes towards vaccination have been prevalent and increasing since adults used the internet to look up health information [14], yet, only 12
the influenza pandemic of 2009 [3,4]. As with many health topics, an % of American adults are considered proficient in their ability to
abundant variety of information regarding vaccines is easily accessible interpret health information [15].
online [5–7]. While believing false information is troubling in many Health information is not unlike news in that the variety of available
contexts, it poses particular problems in the health arena, where it can sources has significantly expanded over the last couple of decades. In the
delay or prevent effective care, potentially threatening lives [8]. past, health information was available through a few highly controlled
The medical community acknowledges the importance of a patient’s sources, but now is a cacophony of online voices with few inherent
ability to self-advocate and manage their own health conditions as safety checks in the information provided [16]. There is a widening
medical resources are increasingly spread thin [9,10]. Understandably, breadth of quality, fueled by Web 2.0 technologies that encourage
patients increasingly consume health information online, which in- people to share their experiences and opinions [17]. For example, users
cludes seeking information to self-diagnose, self-treat, or prepare for a often generate content that tends to be experiential in tone, which is

* Corresponding author.
E-mail addresses: [email protected] (J.L. Claggett), [email protected] (B. Kitchens), [email protected] (M. Paino).
d
In our context we refer to “healthcare information consumer” as patients or non-professional caregivers seeking health information about a patient’s situation.

https://doi.org/10.1016/j.im.2024.104037
Received 14 September 2023; Received in revised form 5 August 2024; Accepted 3 September 2024
Available online 8 September 2024
0378-7206/© 2024 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
J.L. Claggett et al. Information & Management 61 (2024) 104037

somewhat analogous to the consumer product and service reviews that and defined features of online content that contribute to a person’s
often guide consumer purchasing behaviors [18,19]. Further, unlike willingness to believe and follow that health information. Practically,
traditional publishing (i.e., print), online information is easier to alter, the three traits tested (authorship, webpage format, and framing of
plagiarize, misrepresent, and create anonymously under false pretenses advice) are under the designer’s control and can be altered to improve
[20,21]. People’s critical assessment of social media websites, such as the likelihood that individuals will find the information credible.
Twitter (now X), is often lacking [22], even though it is gaining in Finally, we confirm that the dual-processing routes to persuasion
popularity as a way to share health information [23]. Recent studies described by the ELM are present and important in how people decide
increasingly highlight the percentage of false health information online whether to follow online health advice. The implications of these find-
[24], signaling a need to clearly understand what influences a person’s ings and future research directions are discussed at the end.
assessment of online health information.
Prior literature notes that visible signs of expertise [25,26] and 2. Literature review
favorable crowd-support signals [27,28] may improve credibility as-
sessments of health information, but these studies are often embedded in 2.1. Elaboration likelihood model
single contexts (e.g., Facebook or a specific patient forum) and do not
provide a complete view of what may serve as cues (or combinations of The elaboration likelihood model (ELM)e provides a dual-process
cues) that improve credibility of health information encountered over a framework for understanding the basic processes underlying how com-
wide variety of online platforms. In order to understand patient munications effectively persuade an individual and change their attitude
behavior, and thus design better online resources, we need a more by identifying two major routes to persuasion: the central route
complete picture of how people form credibility assessments of online (involving argument quality assessment) and the peripheral route
health information that in turn impact their health literacy [29]. (involving source credibility assessment) [30]. It is a useful tool for
Further, how people assess the credibility of online health information is understanding the influence processes across a variety of technology
likely to differ from other contexts due to the highly specialized level of scenarios [38,40–43]. The central route describes the cognitive response
expertise in healthcare, the complexity of health conditions and in- to information presented. In other words, this describes the assessment
teractions, the traditional evaluation processes of approved treatments, of the quality of the argument being made and uses knowledge about the
and the significant personal consequences. subject already possessed by the person. In related theories, this concept
In this paper, we draw upon the elaboration likelihood model (ELM) is referred to as message credibility [44], which is defined as “an in-
in order to understand how individuals assess the credibility of online dividual’s judgement of the veracity of the content of communication”
health webpages. The ELM lens contends that people use two assessment [40, p. 63]. In contrast, the peripheral route relies on general impres-
routes to determine credibility: (1) the central route (involving argu- sions and other cues or heuristics to judge the message as a proxy for
ment quality assessment) and (2) the peripheral route (involving source argument quality. This route looks for mental shortcuts to assess the
credibility assessment) [30]. While both routes are likely involved to credibility of the message, which may include judgments about the
some extent in any situation where a person may be persuaded, when message sender (e.g., authoritative, charismatic, etc.) or impressions
people are unable or unmotivated to judge the quality of an argument, about the message (e.g., professionally presented, celebrity endorse-
they will lean more heavily on peripheral cues not directly tied to the ment, etc.) [30]. Because the peripheral route does not rely on deep
argument or content to assess the credibility of the information pre- cognitive processing, people may not be fully conscious of the cues that
sented. This use of the peripheral route may be especially relevant in influence them through this route [45]. The ELM, summarized in Fig. 1,
health information scenarios where patients lack the medical training to offers a theoretical explanation for observed differences in how mes-
assess the argument quality of medical advice. Previous research finds sages have different levels of influence on recipients [38]. We unpack
that digital health information credibility is likely to have different pe- the peripheral route to better understand what serves as cues for cred-
ripheral cues that shape the credibility assessment, will involve traits of ibility in the context of online health websites, regardless of the argu-
the website where the information is found, and should be studied ment quality presented by the actual text.
directly [31,32]. It is an area where the importance of context to build The ELM assumes that people are motivated to be accurately
and extend theory is especially salient [33,34]. Further, the peripheral informed but that their willingness and ability to engage in evaluating
cues that influence the peripheral route to persuasion are inherently tied message varies by individual and situational factors [30]. The mixture of
to digital content presentation and can be altered separately from the the two routes in a situation represents a continuum of how a person
message content. Therefore, our research question is: What are the pe- assesses communication; one extreme would be to solely consider the
ripheral cues that impact a person’s credibility assessment of online health message via the central route (high elaboration) and the other would be
information? to solely consider the message via the peripheral route (low elabora-
To address the gap in the literature on health information credibility tion). However, an assessment is likely determined by some combination
assessment, we adopt a mixed methods approach. First, we conduct a of the two paths, resulting in a combination of argument quality and
brief qualitative study to explore what digital content traits serve as source credibility assessments [46].
peripheral cues to people reading online health information. Five traits The central route, leading to argument quality assessment, is
emerge: authorship, webpage format, framing of advice, brand recog- cognitively more taxing to engage in, and two key factors impact its use
nition, and length of material. Next, we conduct an online experiment (2 by a person: motivation and ability [30]. As the central route takes more
× 2 × 2 design) to test the importance of three of these traits (author- cognitive effort, motivation to engage with the communication is
ship, webpage format, framing of advice) in two different online health necessary to expend effort on considering the argument quality [47].
scenarios, controlling for brand recognition and length of material. We Motivation could be personal relevance or interest in the topic. A per-
find that authorship (physician, as opposed to patient) and framing son’s ability also determines the amount of cognitive effort that is
(objective, as opposed to experiential) are significant factors in
increasing a patient’s credibility assessment of online health informa-
tion, and subsequently their intention to follow that advice. Interest- e
Other dual-process theories of attitude formation and persuasion, offer
ingly, we find no evidence to suggest that the webpage format (article or
similar theoretical vantage points such as the heuristic-systematic model [35],
interactive forum) influences credibility assessments. controlled versus automatic processing [36], surface vs. earned credibility [37],
We contribute to the ELM and source credibility theory by extending and dual-process theory of website credibility assessment [21]. For our work,
our understanding of traits that serve as peripheral cues to credibility we agree with previous research [38,39] that they complement each other and
assessments within the health context. In doing so, we uncover specific contain core elements analogous to those discussed in the ELM.

2
J.L. Claggett et al. Information & Management 61 (2024) 104037

Fig. 1. The Dual-Processing Nature of the Elaboration Likelihood Model (summarizing [30]).

required to engage in the central route. For example, if a person is a that positive influence is reduced in a scenario where male consumers
healthcare professional, then processing information about a health are matched with male streamers [50]. Other user-generated content
condition will require fewer cognitive resources than someone without work in online shopping explores how domain relevance and domain
training, as they already have mental scaffolding in place to understand specificity impact argument quality assessments [51]. While considering
and assess the argument. In this way, people with high motivation and crowdfunding, research suggests that, if relevant to the proposal, the
high ability are likely to rely primarily on the central route when pre- source’s education credentials matter during funding decisions [52]. In
sented with a persuasive message. People without ability (e.g., they do the context of fake news, recent work explores the impact of warning
not have the mental resources or training to understand the argument) labels and the removal of social endorsement on a user’s likelihood of
are limited to primarily using the peripheral route, and people without sharing the false information [53]. This growing body of work demon-
motivation are likely to be “cognitive misers” and rely on peripheral strates the importance of considering context-specific nuances to how
cues to drive their assessment [48]. However, as people and situations people process information. There is a need to directly unpack what
vary, a person making an assessment will use a combination of these peripheral cues are salient in online health information assessment.
pathways, with the weights of their reliance on one path over the other People likely assess credibility of online heath information in ways that
depending, in part, on a combination of ability and motivation. The differ from other contexts. The traditional evaluation of treatments (e.g.,
extent to which individuals choose to consider cues from each route is drug trials and published studies), the high level of specialized expertise,
referred to as their elaboration likelihood [41]. the complexity of health conditions and interactions, and the significant
The ELM offers a theoretical explanation for observed differences in personal consequences to health and well-being are all unique to the
how messages have different amounts of influence on recipients [38]. health information context. Several studies call for more work to directly
When a message is presented to individuals, they will vary in how much address how credibility is formed in online health contexts [32,54], yet
cognitive energy they devote to the assessment [30]. In the case of the only a handful of studies attempt to shed light on this phenomenon to
central pathway, their learning content, structure, and processes will date (summarized in Table 1).
influence their conclusion about the argument quality [39]. When Of the studies looking at source credibility assessments of online
engaging in the peripheral pathway, individuals may vary on what health information, three of them manipulate a cue that signals expertise
heuristics they apply to the communication and what serves as periph- or not, yet this was often the presence or absence of an “M.D.” credential
eral cues that signal source credibility. Our focus is to unpack the latter after the author’s name instead of presenting alternative roles such as
to better understand what may serve as peripheral cues in the context of patients. Further, three of the studies [25,26,28] frame an experiment
online health websites. within a real website context (e.g., Facebook or WebMD), which makes
it difficult to isolate the potential effects of the platform brand as a
confounding peripheral cue. Our study extends this work in two distinct
2.2. Online source credibility assessment
ways. First, we take a deep dive into the salient traits of online health
resources via a qualitative study to identify digital content traits that are
At the heart of the peripheral route within the ELM is the idea of
applicable to a wide array of online resources. Second, we manipulate
using peripheral cues to assess the credibility of the information. Cred-
three of the traits that can signal source credibility independently
ibility is a perceptual variable rather than an objective measure of
(authorship, format, and framing) in order to disentangle them from
quality and refers to the believability of information [37,49]. Source
brand or platform recognition in an experiment designed to assess the
credibility is separate from the logical assessment of the argument
impact of these website traits on credibility assessment as well as
quality, and it is formed as an amalgamation of judgements and in-
behavioral intention.
terpretations of cues derived from the presentation of the digital con-
tent. Early work on website source credibility focused on how to
3. Study 1: identifying online health resource traits
cultivate digital literacy when the internet was in its nascent stages and
applied broad concepts of credibility assessment from print media to an
To first address our research question, we need to explore what
online context. For a review of early work, see [21], which includes a list
might be acting as peripheral cues during a person’s credibility assess-
of 25 potential factors that may influence source credibility assessments
ment of online health information. Given the lack of prior work in
of online information. Like Metzger [21], we note that these lists are
studying source credibility in the online health information context, we
unwieldy, lack empirical evidence, are dated, and often were initially
conduct a qualitative study to inform our experimental study design.
identified for use in printed materials; therefore current website capa-
Our mixed-method approach is best categorized as developmental [56],
bilities and usage patterns should be studied.
as this study informs what consumers of health information may be
Recent research continues to explore and expand the Elaboration
noticing when considering the source credibility of online information.
Likelihood Model (ELM) within diverse contexts. In online purchasing
Specifically, our method in Study 1 is best described as a vignette-based
decisions influenced by live streaming, content that purports the func-
qualitative study, where we present example health information
tionality of a product positively impacts user interest in a product, yet

3
J.L. Claggett et al. Information & Management 61 (2024) 104037

Table 1 source credibility and better understand what digital content traits are
Empirical studies considering credibility assessments in healthcare contexts. salient to people when browsing health information online.
Study Context DV Manipulation Key Finding
3.1. Data collection
[26] Physical Credibility • Frame (gain Gain-framed
activity and Perception vs. loss messages and
alcohol message) messages from We invited faculty, staff, and students from a large public university
consumption • Source experts were to participate in the qualitative study, taking care to select people with
health (expert vs. both perceived
varied backgrounds. We selected a variety of ages, evenly spread be-
information non-expert) as more
shared on • Social credible.
tween the age groups 20–29 years old, 30–39 years old, 40–49 years old,
Facebook to endorsement and over 50 years old. Of the 20 participants interviewed, 12 are female
young adults. (high vs. low and 8 are male. None of the participants have any formal medical
like count) training. We identified three websites that offered descriptions and
[25] Q&A format Credibility • Attractive Attractive
advice about the condition of leg cramps, and these websites are very
of a patient Assessment & formatting formatting and
asking advice Behavioral options (e.g., the MD different in terms of lay-out, tone, and functionality. We use an article
about Intention bullet points) credential led to from the Mayo Clinic, a blog post with a forum discussion from a phy-
depression on • MD higher sician’s assistant posted on Clinical Advisor, and a page from a user-
WebMD. credential credibility
forum on Patient Forums. The authors have no affiliation with any of
listed (vs. not assessment and
listed) behavioral
these organizations and made no alterations to the content provided on
intentions the websites to preserve maximum realism. Each participant saw iden-
[27] Twenty-eight Analysis of • Expertise cue Expertise cues tical webpages during their interaction with an author, and preserved
medical credibility of (e.g., Dr. and crowd copies of each website are available upon request.
statement statement, title) support cues
One author conducted individual, in-person sessions with each
snippets author • Crowd were associated
across a wide trustworthiness support (# of with higher participant. The participant sessions varied in length but took approxi-
variety of positive credibility mately 12–15 min each. Each session began with an explanation that we
medical ratings) assessments are gathering data about how people assess online health information,
topics. Half of
specifically what website traits are noticed by people as they browse
the
statements
information. Each participant received a hypothetical situation to
were true. consider: “Someone in your family has been suffering from leg muscle
[28] Examined Coded parent N/A Identified cramps, so you have decided to research leg cramps online. Assume you
online forum forum peripheral cues find this website during your online search. Please spend about 20 s
discussions responses to see within the
browsing this page to find out more information about leg cramps.”
between what they chose website,
parents about to elaborate on. including After the 20 sf passed, they were asked the following open-ended
vaccines. “recommended questions:
answer” badge,
and “thumbs 1. Can you describe the website to me in terms of the type of webpage it
up” approval
from other
is and what you notice about it?
parents 2. Does the website inspire credibility? Why or why not? What are the
[55] Experiment to Credibility • “Contact Us” Study found that biggest factors or traits this website has that you think feeds into your
analyze assessment Link a “contact us” assessment?
effects of • Street link worked as a
contact Address peripheral cue
information • Email Links to increase Given the straight-forward nature of the participant responses, a
provision. perceptions of single author took notes about their statements and the reasons that
credibility, but affected their credibility assessments. This process was repeated with
that street the remaining two webpages, so that each participant saw and answered
address or email
links did not
these questions about all three webpages. The order of the website
significantly presentation was randomized each session, as anchoring and compari-
affect the sons between pages naturally occurred (e.g., “I don’t trust this website
credibility [patient forum] as much as the Mayo Clinic. The Mayo Clinic was a
rating
brand I recognized.”)

websites to participants paired with short interviews to understand the 3.2. Data analysis
judgment process and perceptions that participants ascribe to online
information during their credibility assessments. Vignette studies pre- We collected and codified the responses captured in our session notes
sent short descriptions of scenarios (either written, audio, or video until we achieved theoretical saturation, as is recommended for a study
material) to participants and elicit responses about their judgments, of this type. We follow standard procedures for developing codes and
decisions, and perceptions of the scenario [57]. These vignettes are categories, placing similar themed comments into larger categories until
selected to include specific variables or elements that the research aims prominent themes emerge to form our list of codes [60]. Given the direct
to investigate, allowing for the examination of how these variables in- nature of the answers provided by the participants (i.e., they clearly
fluence participants’ reactions or decisions. By presenting participants stated what they noticed and if it improved or eroded credibility), the
with actual website options, this vignette methodology provides high two authors that reviewed the codes did not have disagreements about
realism to the activity, strengthening content and external validity [58]. the classifications. The set of codes, the frequencies of their appearance,
The scenarios are typically followed by a series of open-ended questions and examples of participant comments are provided in Appendix A.
or prompts that encourage participants to reflect on the vignette and
express their thoughts, feelings, and perceptions. By following this
f
methodology, we identify key website traits used in the assessment of Twenty seconds was selected because user studies have indicated that
people often leave a web page within 10-20 seconds after skimming it [59].

4
J.L. Claggett et al. Information & Management 61 (2024) 104037

3.3. Results online health information, we selected three website traits that emerged
in our first study and empirically tested if people were influenced by
Five digital content traits emerged during our participant sessions: these factors during an experiment focused on the assessment of online
(1) author type, (2) webpage format, (3) framing of advice, (4) brand health information. We selected authorship, format, and framing, as
recognition, and (5) length of the information presented. The first three these are easily decoupled from the health content being communicated
are the focus of our experimental study (Study 2), as these three traits (i.e., a piece of advice can be presented from a physician or a patient, can
may be reasonably isolated and manipulated while maintaining con- be framed as objective or experiential, etc.) and are rooted in the digital
sistency in content. The brand recognition around Mayo Clinic is presentation. Length of the health content was kept constant, to reduce
mentioned in nearly every participant session. But brand recognition is variation in the amount of content to consider in the argument (central
difficult to manipulate and measure within an experiment, and therefore route), and brand was removed to avoid the complexity of capturing
needs to be thoughtfully controlled. In the subsequent experimental existing brand image and exposure across each participant. Iterating
study, our manipulations have no logos or site names to avoid any between our earlier insights and the literature, we develop hypotheses
confounding effects with brand recognition or pre-conceived notions regarding the influence of these traits. See Fig. 2 for a summary of our
about a site or company. Similarly, the information layout (including theoretical model.
approximate word-count, headers, and bullet pointed lists) is mentioned Authorship: A major component of an individual’s source credi-
by participants, and generally indicates they prefer shorter web re- bility assessment is the consideration of the author or person providing
sources. Recent work in neuroscience shows that different message the information [62–64]. As internet resources are almost always pro-
lengths may engage different parts of the brain during argument quality vided by people not known to the reader, the role or title of the author
assessment [61]. Therefore, care is taken to keep the word counts and becomes a prominent trait when assessing credibility. We consider two
page lengths nearly identical in our experimental design. common roles in the health information context: healthcare provider
One interesting occurrence is that different factors are noticeable to and fellow patient. Healthcare providers are medical experts trained to
participants based on the type of page they are shown. In other words, provide medical care, while patients are perceived as peers who may
what is a strong factor on one example page might not be mentioned at deal with the same types of health concerns. Previous research demon-
all (the absence or presence of the factor) on another page. For example, strates some ambivalence regarding which role is viewed more
branding and professional layout are commonly described traits of the favorably.
Mayo Clinic article, but the (lack of) brand recognition or (less desirable) On one hand, in many online contexts, other consumers are consid-
lay-out is rarely mentioned on the other two websites. In contrast, cre- ered a more trustworthy source than figures of authority or the people
dentials are frequently noted on the physician assistant’s blog, and the trying to “sell” the product or service [65]. Consumers increasingly rely
lack of any verifiable credentials is noted on the patient forum, but it on the recommendations and ratings of others when making online
rarely came up when discussing the Mayo Clinic article (which is decisions in a wide variety of online contexts [66]. In addition to the
credited to “Mayo Clinic Staff” at the bottom of the webpage). Our recommendation phenomenon, people may also feel comforted by the
qualitative study suggests that one or two traits are especially salient for perceived shared experiences of fellow patients. Research has noted the
each webpage we showed the participants, which increases our interest emotional and psychological benefits of finding patients with similar
in the research question and our motivation to design a second study to illnesses in online communities [67].
isolate the effects of the actual traits. The sequential combination of However, healthcare providers are trained in the medical field and
qualitative and quantitative studies allows us to further explore our are considered experts [67]. Expertise is strongly associated with
research question by taking insights found in the qualitative data and competence and authoritativeness [66,68], which are traits that yield
designing an experiment to explore the relationships identified. credibility to statements made within their area of expertise. Previous
research shows that perceived levels of expertise positively impact a
4. Study 2: testing the effects of three online health resource person’s willingness to follow recommendations [26,27]. Presumably,
traits on source credibility and willingness to follow advice this positive impact is due to the perception that expertise leads to
credible advice. For example, participants in one study were more likely
4.1. Theoretical model and hypotheses to show interest in a movie that was recommended by a movie critic over
other movie patrons, especially in situations where multiple opinions
To further explore our research question, seeking to understand what were not aggregated [16]. Personal healthcare is an arena with serious
are the peripheral cues that impact a person’s credibility assessment of consequences and concerns, and we believe this will increase the desire

Fig. 2. Research Model.

5
J.L. Claggett et al. Information & Management 61 (2024) 104037

to find and follow information presented by an expert. Therefore, we their willingness to follow advice due to an increased perception of
posit that expertise will be favored over peer advice when considering source credibility, as it is a more traditional format associated with
the credibility of online health information. Further, the positive influ- scientific research.
ence on a person’s willingness to follow advice because of their
H2(a). Online health resources presented in an article format (as opposed
perceived expertise will be mediated by source credibility. In our study,
to a forum format) will have higher source credibility assessments.
we select the physician as a classic and easily recognizable example of a
healthcare provider with relevant expertise.
H2(b). The impact of article format on a person’s willingness to follow
H1(a). Online health resources authored by a physician (as opposed to a online health advice is mediated by their source credibility assessment.
fellow patient) will have higher source credibility assessments.
Framing: The third website trait we test that can vary between on-
line health resources is the framing of the presentation. Many informa-
H1(b). The impact of physician authorship on a person’s willingness to
tional sources have an objective tone, where information is presented in
follow online health advice is mediated by their source credibility assessment.
a factual manner devoid of any personal insights. For example, “mag-
Format: Another salient trait that may inform source credibility nesium supplements may ease muscular cramps in patients.” In contrast,
assessment is the format of the online resource. Traditional medical information can be presented as personal experiences of the author. For
resources such as peer-reviewed manuscripts and textbooks are pre- example, “I started taking magnesium supplements and it helped with
sented as information from one authorship source. Even when multiple my muscular cramps.” This distinction is of particular interest to online
authors are involved (e.g., co-authors), the voice of the resource is health resources, as social media and other ways to publish information
consistent, and it is presented as a one-directional flow of information in online have led to an explosion of information shared as personal ac-
a traditional, read-only format. We refer to this as article format. counts [54,76]. Further, there is often an implication that experiential
However, the internet provides an opportunity to create platforms with tone is synonymous with patient voices (instead of healthcare provider
rich user interaction. The term Web 2.0 was coined by O’Reilly [69] to voices) and discussion forums (instead of article format). However, the
refer to back-and-forth interaction as internet consumers transitioned framing of the information is separate from the authorship and the
from passive readers of material to active participants sharing their own format, and examples exist with all combinations (e.g., physicians can
data and thoughts to create internet content. use an experiential tone in an article, and patients can respond to a
The results of having a Web 2.0 format change how people share and discussion forum with objectively presented information). Therefore, we
process information, and the impact is studied in political elections and recognize and isolate the effects of experiential framing on how people
campaign effectiveness [70], merger and acquisition activities [71], and assess source credibility.
education activities [72]. The potential benefits of actively engaging in Marketing research notes that the use of experiential framing can
Web 2.0 online health resources (e.g., asking a question or requesting appeal to feelings and emotions [77] and may drive feelings of loyalty
clarification from the author) are outside of the scope of our study. and satisfaction [78]. Yet these insights are garnered from studies
However, the passive use of Web 2.0, where a reader peruses past considering consumer brand and product opinions, not healthcare
interactive conversations, is a common internet use case, as the nature of advice. Further, the experiential tone in these conditions is likely part of
the internet is to archive and preserve all content. These conversational an orchestrated experiential marketing campaign by design [77] and
formats create environments with potential opinion congruence, where may have different effects than shared experiences over the web that
multiple voices intertwine in one web resource [16]. In one study happen organically (e.g., an individual sharing their health experi-
focusing on health advice platforms, physicians and nurses were the ences). While some patients may appreciate hearing about a similar
primary voices in 20 % of those sampled, indicating that these conver- health experience, individual experiences are often interpreted as sub-
sational formats are embraced by physicians and patients [73]. In recent jective information [79], which may harm credibility in a health
years, some of these conversations have migrated from online discussion context. Further, the medical field has a long history of evidence-based
forums (e.g., Reddit, Stack Overflow, etc.) to social media platforms (e. practices, where numerous participants are studied to statistically
g., Facebook, Twitter, etc.), yet 20 % of U.S. online users still participate determine the efficacy of a treatment. An objective framing may be
in forums, and they are routinely included in search results [74]. interpreted as a result of evidence-based studies, as opposed to an
Further, the back-and-forth discussion format is still present in many anecdote which may only be one person’s experience. Therefore, we
social media platforms, as users respond to posts or tweets. For posit that consumers will assess health information presented in an
simplicity, we refer to this Web 2.0 type format, which allows multiple objective frame as being more credible.
authorship sources and a recorded “back and forth” type conversation as
H3(a). Online health resources presented in an objective frame (as opposed
a forum format.
to experiential) will have higher source credibility assessments.
It remains unclear how format (article vs. forum) plays into the
credibility assessment and ultimate decision-making process of internet
H3(b). The impact of framing on a person’s willingness to follow online
users. A report from the American Press Institute (2016) suggests that
health advice is mediated by their source credibility assessment.
half of Americans get news on social media (a forum-like Web 2.0
format), but the levels of skepticism vary. That same report underscores
the importance of context on people’s acceptance of forum formats, 4.2. Methodology
because a website format that allows for public comment is seen as
especially important for crime and public safety, weather or traffic, and To test our hypotheses, we conduct a web-based experiment that
domestic issues (healthcare information is not included in their results). proceeds in four steps. First, participants filled out a survey indicating
To our knowledge, research has not yet isolated and studied the impact their demographic background, information about their previous
of website format on willingness to follow health advice or their source healthcare experiences, and their general trust in healthcare. Next, we
credibility assessment of it. Our own qualitative study results indicate presented participants with a hypothetical health complaint situation.
that participants are more skeptical when presented with healthcare Then, we showed each participant a single webpage (of the possible
information from an online blog or forum. Although not about health eight experimental conditions) offering health advice that would
information specifically, a recent study shows that people are less likely directly address the health complaint. Finally, we measured assessments
to find scientific information credible when shared over Twitter in of argument quality, source credibility, and willingness to follow the
comparison to other platforms [75]. Due to the gravity of healthcare advice. We evaluate the data via OLS regression and conduct mediation
information, we posit that people will find an article format increases analyses.

6
J.L. Claggett et al. Information & Management 61 (2024) 104037

4.2.1. Experimental design

We presented each participant with a hypothetical health prompt to
begin the experiment, in the following format: “Your friend has recently
been battling [a health condition]. They ask you to help them research the
condition online, and you find the following resource. Please read the
web page presented and then answer the questions related to your
experience with the online resource.” We use two health conditions to
create a more robust test of our hypotheses. In one scenario, we ask
about leg cramps and present online health information that offers
advice to use magnesium supplements to ease leg cramps. In the other
scenario, we ask about moderate acne and present online health infor-
mation that offers advice to try a low-sugar diet to decrease acne out-
breaks. These two health situations, leg cramps and acne, were selected
because the treatment recommendations made are nontrivial but could
nonetheless be administered without formally seeking professional
medical interventions (a diet change and an over-the-counter supple-
ment). These are also typically not categorized as serious or chronic
conditions, such as diabetes, heart disease, or cancer. Many patients
faced with these illnesses would be highly motivated to become
educated on many facets of the disease and develop a level of expertise
that would significantly lessen the use of a peripheral path to decision
making [10,30]. We designed the treatment recommendations under the
guidance of a medical doctor who specializes in family medicine. Fig. 3B. Acne treatment advice presented from physician author, using
For each scenario, we created eight webpages to use in the experi- objective framing, in an article format.
ment, utilizing a 2 × 2 × 2 experimental design. These eight webpages
represent each possible permutation of the three website traits of in-
terest: authorship (physician or patient), format (article or discussion
forum), and framing (objective or experiential). Each webpage,
regardless of the combination of traits, offers the same advice, presented
in similar language (varied slightly for experiential vs. objective
framing), and uses equivalent word counts. Each experiment participant
was randomly assigned one of the eight different webpages. Because two
health scenarios were used (leg cramps and acne), a total of 16 pages
were generated. Figure 3 shows four webpages used in the experiment
(two from each health scenario), of two opposite trait permutations, to
demonstrate how the manipulations are designed. A complete set of the
16 webpages (eight for acne advice and eight for leg cramp advice) can
be found in Appendix B.
Fig. 3A, 3B, 3C and 3D
In study one, we note the positive perception of a specific brand or

Fig. 3C. Leg cramp treatment advice presented from patient authors, using
experiential framing, over a forum format.

the reputation of a website, which aligns with previous studies assessing

the impact of healthcare branding [80]. Our qualitative study partici-
pants also remarked on a preference for “skimmable” webpages. For
instance, bullet points receive positive comments, while long blocks of
text are dismissed; this pattern is consistent with previous research [81,
82]. To avoid confounding effects rooted in brand recognition, text
length, or readability, our experimental design avoids any branding, and
Fig. 3A. Leg cramp treatment advice presented from physician author, using we standardize the text length and grammatical presentation in all
objective framing, in an article format.

7
J.L. Claggett et al. Information & Management 61 (2024) 104037

Table 2
Descriptive statistics of each health condition experiment scenario.
Variable Leg Cramp Acne Scenario
Scenario

Sample Size N = 1683 N = 1679

Race Demographics
Asian or Pacific Islander 127 (7.5 %) 115 (6.8 %)
Black 363 (21.6 %) 355 (21.1 %)
Hispanic or Latino 147 (8.7 %) 143 (8.5 %)
Other 67 (4.0 %) 65 (3.9 %)
White 979 (58.2 %) 1001 (59.6 %)
Urbanicity Demographics
Rural 406 (24.1 %) 406 (24.2 %)
Suburban 765 (45.5 %) 787 (46.9 %)
Urban 512 (30.4 %) 486 (28.9 %)
Age Demographics
Young (18–34 years) 566 (33.6 %) 541 (32.2 %)
Middle-aged (35–64 years) 877 (52.1 %) 879 (52.4 %)
Older (65+ years) 240 (14.3 %) 259 (15.4 %)
Other Demographics
Female 989 (58.8 %) 985 (58.7 %)
Bachelor’s degree or higher 537 (31.9 %) 521 (31.0 %)
Income bracket mean=4.08 (s. mean=4.11 (s.
d.=3.18) d.=3.15)
Number of health visits in the last 2 mean=4.43 (s. mean=4.45 (s.
years d.=3.38) d.=3.44)
Viewed on Mobile Device (as opposed 929 (55.2 %) 927 (55.2 %)
to computer browser)
Other Controls
Healthcare Trust mean=2.69 (s. mean=2.68 (s.
d.=0.84) d.=0.86)
Argument Quality Assessment mean=3.02 (s. mean=2.94 (s.
d.=0.78) d.=0.79)
Measured Constructs of Interest
(Continuous variable, ranging from 0 to 4: see Appendix D for measurement items)
Source Credibility Assessment mean=2.77 (s. mean=2.70 (s.
d.=0.78) d.=0.79)
Willingness to Follow Advice mean=2.81 (s. mean=3.06 (s.
Fig. 3D. Acne treatment advice presented from patient authors, using experi- d.=0.863) d.=0.80)
Experimental Conditions (2 × 2 × 2 design – random assignment to 1 of 8 permutations)
ential framing, over a forum format.
Physician (as opposed to patient 818 (48.6 %) 814 (48.5 %)
author)
iterations. Article (as opposed to forum format) 850 (50.5 %) 862 (51.3 %)
Objective (as opposed to experiential 830 (49.3 %) 846 (50.4 %)
framing)
4.2.2. Sampling and data collection
We designed the web-based experiment to be embedded in a Qual-
trics survey. We also used Qualtrics panels to recruit participants that cramp scenario. Respondents report the number of health visits they
are demographically representative of the population of internet users in attend per year, and we use this information to control for their level of
the United States, which is a strength of Qualtrics solutions [83]. We engagement with personal health conditions. In order to rule out the
exclude participants who work in a medical profession and those who possibility of device effect [85], we capture the device the web-based
have previous personal experience with the health condition being experiment is viewed on (i.e., mobile browser vs. computer browser)
considered (i.e., suffered from leg cramps or acne) to remove partici- but do not find evidence this affected a person’s likeliness to follow
pants with previous expertise in the study’s context, as these partici- online health information. Healthcare trust is associated with patient
pants are likely heavily opinionated regarding their ideal treatment of satisfaction and broader health outcomes [86], and patients with higher
the health condition in question and expertise is more likely to make one levels of healthcare trust are more likely to comply with treatment plans
focus on the central route of argument quality assessment [30]. [87], so we also control for a person’s level of healthcare trust. Previous
Our Qualtrics survey solution does not record incomplete responses work with the ELM explains the two assessment paths used in decision
(i.e., people that did not complete the survey are not included in the making: argument quality (central) and source credibility (peripheral)
results) or responses that took an impossibly short amount of time to [30]. We do not include hypotheses for the direct positive effects of
complete. Additionally, we removed responses with invariant answers argument quality assessment or source credibility assessment on will-
(e.g., all responses were a “4″ on the numerous Likert scale questions), ingness to follow advice because these are posited and tested in prior
indicating that the person likely completed the survey without reading work [38,52], but, as expected, we empirically confirm these relation-
the questions. For the scenario using leg cramps as the health condition, ships in our study (see Table 5).
our sample size includes 1683 responses, and for the scenario using acne Measurements: We use a four item version of the healthcare trust
as the health condition, our sample size includes 1679 responses. scale, derived from Anderson and Dedrick [88] and modify it to refer to
Descriptive statistics of each analyzed sample are included in Table 2. healthcare professionals instead of the term “doctor.” Source credibility
Controls: We use various demographic information (sex, race, ed- and argument quality is assessed using the same modified versions of
ucation level, income, age, and urbanicity: whether they lived in a rural, Sussman and Siegal’s [89] scales that were used by Bhattacherjee and
suburban, or urban area) as control variables in our model. Previous Sanford [40]. To measure a participant’s willingness to follow the pre-
work finds evidence that older individuals are more likely to find online sented health advice, we developed a four-item scale that directly
health advice credible, regardless of features and credibility cues [84]. referenced the advice presented in the scenario. Details on our scale
We observe this pattern in the acne condition scenario, but not the leg development process can be seen in Appendix C, and measurement

8
J.L. Claggett et al. Information & Management 61 (2024) 104037

scales are presented in Appendix D. Table 4

The leg cramp (acne) scenario data has good model adequacy, with a OLS Regression of Source Credibility Regressed on Experimental Conditions.
Kaiser-Meyer-Olkin value of 0.896 (0.901) and a significant Bartlett’s Leg Cramp Acne
Test of Sphericity (sig.<0.001 for both). Reliability is high, as all Scenario Scenario
Cronbach’s Alpha values are between 0.84 and 0.91 (0.83 and 0.89). A Experimental Conditions
summary of our confirmatory factor analysis, including composite reli- Physician Author 0.174*** 0.157***
ability, average variance extracted, maximum shared variance, and Article Format 0.024 − 0.017
correlations between constructs can be found in Table 3. Appendix E Objective Framing 0.072** 0.055**
Control Variables
contains item and cross loadings. Healthcare Trust 0.330*** 0.300***
Common method bias is a research design concern that refers to the Viewed on a Mobile Device − 0.025 − 0.031
likelihood that common variance is shared between the independent Rural Setting − 0.081** − 0.046
variables (IVs) and dependent variables (DVs) due to the fact that re- Suburban Setting − 0.063* − 0.010
(Reference=Urban Setting)
spondents are being asked about the constructs in the same way (e.g., an
Female 0.004 0.017
online survey) and at the same time. We conduct Lindell and Whitney’s Asian − 0.15 0.017
[90] test, a recommended approach to rigorously assess common Black − 0.006 0.040
method bias [91] and find no evidence of common method bias (see Hispanic 0.003 0.024
Appendix F). Other Race − 0.018 − 0.005
(Reference=White)
Bachelor’s Degree or Higher − 0.053* − 0.081**
(Reference=Completed less than
4.3. Results Bachelor’s)
Income − 0.009 − 0.009
The two scenarios (leg cramp and acne) yield similar results Young (18–34) − 0.038 − 0.029
Middle-Age (35–64) 0.013 0.006
regarding the support of our hypotheses (Table 4). Participants are more
(Reference=Older (65+)
likely to assess source credibility more favorably when presented with Number of Health Visits − 0.001 − 0.023
health advice that is authored by a physician (instead of a patient) and Sample Size n = 1683 n = 1679
framed using an objective tone (instead of an experience framing), Adjusted R-Squared 0.149 0.125
which supports hypotheses 1a and 3a. The format of the health advice *
p<.05.
(forum or article presentation) does not seem to affect the source cred- **
p<.01.
ibility assessment; thus, hypothesis 2a is not supported in either ***
p<.001.
scenario.g
In order to verify the importance of both the argument quality
assessment (central route) and source credibility assessment (peripheral Table 5
route), we run an OLS regression model showing that both are signifi- OLS Regression of Willingness to Follow Advice Regressed on Argument Quality
cant predictors of a person’s willingness to follow the health advice and Source Credibility Assessments (both health scenarios).
(Table 5). The two experimental conditions show similar results, Leg Cramp Scenario Acne Scenario

Decision Making Paths

Table 3 Argument Quality Assessment 0.482*** 0.491***
Summary Results of Confirmatory Factor Analysis. Source Credibility Assessment 0.169*** 0.190***
Control Variables
Leg Cramp Scenario Viewed on a Mobile Device − 0.002 − 0.027
CR AVE MSV 1 2 3 4 Rural Setting 0.015 − 0.024
Suburban Setting 0.006 − 0.014
1. Healthcare Trust 0.84 0.57 0.11 0.75 (Reference=Urban Setting)
(Control) Female − 0.002 0.006
2. Argument Credibility 0.87 0.64 0.58 0.25 0.80 Asian − 0.024 0.002
Assessment (Control) Black − 0.045* 0.026
3. Source Credibility 0.85 0.66 0.58 0.33 0.76 0.81 Hispanic 0.005 0.000
Assessment Other Race 0.005 0.018
4. Willingness to Follow 0.91 0.71 0.38 0.19 0.54 0.62 0.84 (Reference=White)
Advice Bachelor’s Degree or Higher − 0.075*** − 0.038
Acne Scenario (Reference=Less than BA)
Income − 0.011 0.065**
1. Healthcare Trust 0.84 0.58 0.09 0.76 Young (18–34) − 0.010 − 0.067*
(Control) Middle-Age (35–64) 0.011 − 0.076**
2. Argument Credibility 0.89 0.66 0.59 0.21 0.81 (Reference=Older (65+)
Assessment (Control Number of Health Visits 0.031 0.039*
3. Source Credibility 0.84 0.64 0.59 0.30 0.77 0.80 Sample Size N = 1683 N = 1679
Assessment Adjusted R-Squared 0.396 0.434
4. Willingness to Follow 0.89 0.67 0.41 0.23 0.56 0.64 0.82 *
Advice p<.05.
**
p<.01.
CR – Composite Reliability; AVE – Average Variance Extracted; MSV – Maximum ***
p<.001.
Shared Variance;.
Leading diagonal (bold) shows the square root of the variance shared between
the constructs and their measures. indicating that the central route has the most influence regarding the
decision, but that both routes are used.

4.3.1. Mediation tests

We hypothesize that the experimental conditions serving as periph-
g
In a post-hoc analysis, we conducted two-way and three-way moderation eral cues (authorship, format, and framing) will impact source credi-
tests on our experimental conditions (authorship, format, and framing). We did bility assessment, and that the subsequent effects on a person’s
not find any significant moderation results.

9
J.L. Claggett et al. Information & Management 61 (2024) 104037

Table 6
Summary of Mediation Test Results for Physician Authorship.
Mediation Test

Leg Cramp Scenario

Sum of Estimated Indirect Effects (a’ þ b’)* Effect SE Lower CI Upper CI
0.1902 0.03 0.1392 0.2433
Estimated Direct Effects (c’) Effect SE Lower CI Upper CI
− 0.0397 (n.s.) 0.04 − 0.1183 0.0388
Total Net Estimated Effects 0.1505 (p=.0012)
Sobel Test The Sobel test statistic is 10.15, with a standard error of 0.019 and a p-value of < 0.001, lending further support for the proposed
mediation effect.
Interpretation Full Mediation
Acne Scenario
Sum of Estimated Indirect Effects (a’ þ b’) Effect SE Lower CI Upper CI
0.1817 0.03 0.1281 0.2370
Estimated Direct Effects (c’) Effect SE Lower CI Upper CI
0.0185 (n.s.) 0.04 − 0.0567 0.0936
Total Net Estimated Effects .2002 (p<.001)
Sobel Test The Sobel test statistic is 6.76, with a standard error of 0.027 and a p-value of < 0.001, lending further support for the proposed
mediation effect.
Interpretation Full Mediation
*
Indirect effects calculated directly during bootstrap estimation in SPSS (Hayes 2013); Lower and upper limit of 95 % confidence intervals reported.

willingness to follow advice will be mediated by source credibility foundational work detailing the causal step approach, and recent ad-
assessment (H1b-H3b). Of the experimental conditions, we find direct vances in statistical methods using bootstrapping provide an additional
effects on source credibility assessment for authorship (physician vs. way to estimate indirect effects [93]. We present the bootstrap method
patient) and framing (objective vs. experiential) but not for format test, which uses resampling to calculate path coefficients, standard er-
(article vs. forum). Therefore, we conduct a mediation test on authorship rors, and confidence intervals for the intervening effect of a mediator
and framing. variable [94]. For completeness, we also conduct the Sobel “delta”
Mediation analysis is well described in Baron and Kenny’s [92] method, which calculates the indirect effects as the product of the two

Table 7
Summary of Mediation Test Results for Objective Framing.
Mediation Test

Leg Cramp Scenario

Sum of Estimated Indirect Effects (a’ þ b’)* Effect SE Lower CI Upper CI
0.0661 0.03 0.0175 0.1160
Estimated Direct Effects (c’) Effect SE Lower CI Upper CI
0.074 (n.s.) 0.04 − 0.0033 0.1513
Total Net Estimated Effects 0.1401 (p=.0026)
Sobel Test The Sobel test statistic is 3.76, with a standard error of 0.018and a p-value of < 0.001, lending further support for the proposed
mediation effect.
Interpretation Full Mediation
Acne Scenario
Sum of Estimated Indirect Effects (a’ þ b’) Effect SE Lower CI Upper CI
0.0615 0.03 0.0095 0.1140
Estimated Direct Effects (c’) Effect SE Lower CI Upper CI
0.095 (p=.012) 0.07 0.0209 0.1689
Total Net Estimated Effects 0.1564 (p<.001)
Sobel Test The Sobel test statistic is 2.35, with a standard error of 0.026 and a p-value of < 0.05, lending further support for the proposed
mediation effect.
Interpretation Partial Mediation
*
Indirect effects calculated directly during bootstrap estimation in SPSS (Hayes 2013); Lower and upper limit of 95 % confidence intervals reported.

10
J.L. Claggett et al. Information & Management 61 (2024) 104037

paths a’ and b’ [95]. Details are in Tables 6 and 7. We find support that between format perceptions than our experimental study was able to
both authorship and framing cues influence a person’s willingness to detect or that future studies should select different framing/wording
follow advice are mediated by source credibility in both health sce- when asking about potential cues to form a more nuanced understand-
narios. Therefore, we find support for hypotheses H1b and H3b. ing of the effects of format. To rule out the possibility that formatting
Full mediation is a situation where the independent variable predicts differences across devices might contribute to this, we conducted post-
the dependent variable without considering the mediator, but when the hoc analyses to ensure the article format condition is not affecting
mediator is added the direct effect is no longer significant. However, users of mobile devices differently than those on traditional web
when the independent variable has a diminished, but not insignificant browsers. Across all devices used and health conditions described, no
power to predict the dependent variable in the presence of the mediator, difference between an article and a forum format is detected upon
that is considered partial mediation [92]. Within the leg cramp health source credibility assessment or willingness to follow online health
scenario, we observe full mediation in both mediation tests. Within the advice. One explanation for this is that Web 2.0 formats providing a
acne health scenario, source credibility is observed as fully mediating “back and forth” exchange of information between multiple authors via
authorship but only partially mediating the effects of formatting. forums, social media platforms, and the like, have become so ubiquitous
that consumers do not assign any value to format when assessing cred-
5. Discussion ibility, despite noting it when asked to describe an example webpage.

5.1. Identification of peripheral cues in online health information 5.2. Theoretical implications

The use of the internet to source online health information is Our findings fill a gap in the research that explores what impacts a
becoming more common, especially as patients are put into a position to person’s credibility assessment of online health information [32,54].
manage substantial aspects of their own health due to strained health- The ELM provides scaffolding to understand the importance of argument
care systems [96,97]. Yet, what people choose to believe in the online quality and source credibility on the decision-making process, but the
health arena varies greatly, as evidenced by the strong emotions and nuances of context are important [30]. In addition to the general good
split opinions about vaccine efficacy and safety during the COVID-19 website design principles that influence user perceptions [41,100], we
pandemic [98,99]. One important facet for understanding how and seek to understand what context-specific traits serve as peripheral cues
why people cling to misinformation about health advice, such as vaccine in the online health information context. Our use of mixed methods
recommendations, is uncovering what drives their credibility assess- allows us to triangulate the factors that influence source credibility, as
ment in the first place. Our identification of peripheral cues used within well as find evidence of the importance of several factors across methods
the online health context contributes to this understanding. Previous [101].
research establishes many website elements that positively sway user In their work positing what may be important credibility cues for
opinion, such as navigation functionality, embedded capabilities, health information, Hocevar et al. [31] discuss credentials, which is
attractive graphics, and a feeling of consumer support [41,100]. By consistent with our findings – the presence of a physician credential (M.
examining the healthcare information context specifically, we comple- D.) improves a consumer’s source credibility assessment. They also
ment this knowledge by uncovering that authorship and framing influ- suggest reputation, along with the absence of commercial motive, could
ence source credibility assessments (see Table 8). Healthcare favorably signal credibility. While our qualitative study did not include
organizations can leverage this to effectively design health information anyone voicing a concern about commercial motive, it may be a factor in
websites to inform people who are seeking health information. assessments of the importance of brand and reputation.
Surprisingly, we find no support for our hypotheses that predict the We find evidence that consumers considering online health infor-
online format of health information will serve as a peripheral cue. In the mation react to objective framing with a more favorable source credi-
preliminary qualitative study, we noticed that when shown an example bility assessment than experiential framing. This exemplifies the
of a healthcare webpage with a forum format, people commented about importance of context-specific examinations of peripheral cues, as pre-
the format as the first thing they noticed approximately one-third of the vious research in consumer review interpretation and crowdfunding
time. In contrast, format was rarely mentioned if the example webpage websites indicates that a personal story and experience often yields
was in an article format. This may indicate a more complex relationship higher assessments of source credibility [52,77,78]. Yet, when consid-
ering health information, an objective tone serves as a credibility cue.
Our study also recognizes that general healthcare trust may prime
consumers to be predisposed to finding online health information
Table 8
credible. We measure and control for the participant’s level of general
Hypotheses Summarized.
healthcare trust and find that it significantly impacts a person’s assess-
Hypothesis Supported?
ment of the source credibility. As health information consumers
H1(a): Online health resources authored by a physician (as opposed to Yes approach health-related internet searches with a preexisting trust in
a fellow patient) will have higher source credibility assessments. healthcare, web designers cannot control this factor. However, it may
H1(b): The impact of physician authorship on a person’s willingness to Yes
follow online health advice is mediated by their source credibility
help explain the divergent opinions on how and why the population
assessment. responds to public health messages differently.
H2(a): Online health resources presented in an article format (as No In addition to this extension of the ELM to investigate what, specif-
opposed to a forum format) will have higher source credibility ically, serves as peripheral cues in this context, we provide additional
assessments.
empirical support of the explanatory power of the ELM framework as we
H2(b): The impact of article format on a person’s willingness to follow No
online health advice is mediated by their source credibility retest the core relationships described by the central and peripheral
assessment. routes within our models. Our study confirms the presence and impor-
H3(a): Online health resources presented in an objective frame (as Yes tance of the argument quality assessment and source credibility assess-
opposed to experiential) will have higher source credibility ment in how a consumer of online health information decides whether
assessments.
H3(b): The impact of framing on a person’s willingness to follow Yes
or not to follow online health information advice.
online health advice is mediated by their source credibility
assessment.

11
J.L. Claggett et al. Information & Management 61 (2024) 104037

5.3. Practical implications in general agreement about a condition (see Figure 3). One direction of
future research could be to study a higher variation of opinions in a
The use of the internet to source online health information is forum setting. In reality, Web 2.0 formats that invite multiple voices
becoming more common [102], especially as patients are put into a often result in discourse, or even open disagreement. Previous work
position to manage substantial aspects of their own health due to notes that consumers’ preference for expert advice (instead of crowd-
strained healthcare systems [96,97]. Yet, what people choose to believe sourced) about movie reviews shifts as the volume and valence of the
in the online health arena varies greatly, as evidenced by the strong reviews change [16]. The amount of discourse within forum formats
emotions and split opinions about vaccine efficacy and safety during the may be a peripheral cue to explore.
COVID-19 pandemic [1,98,99,103]. One important facet for under- The authorship condition tested in the experiment is dichotomous –
standing how and why people cling to misinformation about health either authored by a physician or a fellow patient. However, numerous
advice, such as vaccine recommendations, is uncovering what drives other possibilities exist, including a recent trend by some major
their credibility assessment in the first place. We believe our identifi- healthcare providers (e.g., MayoClinic and Healthline) to display
cation of peripheral cues used within the online health context con- “medically reviewed by [name], M.D.” to indicate a credentialed pro-
tributes to this understanding. Previous research establishes many fessional vetted the material but did not write it. Some health infor-
website elements that positively sway user opinion, such as navigation mation online is also simply attributed to a “staff member,” relying more
functionality, embedded capabilities, attractive graphics, and a feeling on brand image than individual credentials. Future research should
of consumer support [41,100]. By examining the healthcare information consider a more nuanced approach to authorship to tease apart how this
context specifically, we complement this knowledge by uncovering that peripheral cue works in more complex scenarios.
authorship and framing influence the credibility assessments of users. Additionally, the presence of visible sponsorships could significantly
Healthcare organizations can leverage this knowledge to effectively influence the interpretation of peripheral cues by consumers. For
design health information websites to inform people who are seeking instance, content perceived as sponsored by a pharmaceutical company
health information. may lead consumers to question the authenticity of experiential narra-
However, a potential downside of our findings is that these design tives, suspecting them to be commercially motivated rather than
traits could be exploited by illegitimate parties to disseminate false in- genuine patient experiences. Similarly, physician endorsements that are
formation. Currently, the title "M.D." can be easily fabricated in many financially backed by a pharmaceutical entity might be viewed as
online resources, highlighting the need for future technological solu- biased, casting doubt on their credibility. Such dynamics underscore the
tions to prioritize credential validation. Ensuring a robust system to complexity of consumer behavior in interpreting online health infor-
verify the credentials of authors and information sources will become mation, particularly in commercial contexts where the lines between
increasingly important. Additionally, users will need to be educated as a unbiased information and promotional content can blur. This intersec-
component of health literacy on how to interpret credential validation tion warrants further investigation, as it may reveal nuanced consumer
on digital health information platforms. responses to sponsored health content, aligning more closely with be-
haviors observed in commercial review systems.
6. Limitations and future research directions In our experimental design, we intentionally selected health sce-
narios that were nontrivial yet not life-threatening—such as leg cramps
Our use of mixed methods allows us to triangulate the factors that and moderate acne—to investigate the influence of peripheral cues on
influence source credibility. As with any research, these chosen meth- credibility assessments. Given our experimental design and prompt, it
odologies involve a host of strengths and limitations, which suggest was important our health information advice could be administered
future research directions. Our research design, akin to most experi- without a medical professional’s direct involvement. This approach was
mental studies, prioritizes strong internal validity [104], ensuring that also based on the understanding that patients facing life-threatening
the observed effects can be directly attributed to the manipulation of diseases might exhibit a higher motivation and potentially develop
peripheral cues rather than external factors. One strength of our expertise, leading them to rely more heavily on the central pathway for
experimental design is the use of two different health scenarios to test decision-making regarding their condition. Such high-stakes situations
our hypotheses. The relationships from our conceptual model are could necessitate a deeper, more deliberate processing of information
generally identical across both scenarios tested. This congruency sug- (argument quality assessment), as posited by the ELM [30]. Conse-
gests our findings are robust and are not dependent on an idiosyncrasy quently, this leaves open an intriguing avenue for future research:
within our experiment’s health condition or treatment description, investigating whether the peripheral cues identified in our study hold
further strengthening external validity. A challenge in experimental the same significance in these more critical health scenarios could pro-
studies is ecological validity, or realism [104]. A limitation of our study vide further insights into how individuals navigate and assess online
is that, to isolate the three peripheral cues being tested, we did not health information under varying levels of health-related stress and
display a brand or URL address in our online health information dis- urgency.
plays. The presence of a brand to judge may alter a consumer’s assess- Our study presents a foundational understanding of how peripheral
ment, and future research should explore this complex variable. In cues affect credibility assessments of online health information within a
addition, our prompt design stated, “Your friend has recently been U.S.A.based participant pool. We acknowledge the limitations of this
battling [a health condition],” which we believe causes people to assess approach, as cultural dimensions have been shown to impact trust for-
health information in a manner similar to how they would in a mation and the evaluation of online information [105] and the use of
real-world situation. Yet consumers’ assessment processes may be technology artifacts [106,107], suggesting that different cultures may
different in a non-experimental setting or if they are personally suffering prioritize different cues when assessing credibility. Therefore, an
from the condition being discussed. Both sets of health scenarios and extension of our work would involve examining how specific cultural
advice suggestions were developed with the help of a physician to be contexts mediate the relationship between peripheral cues and credi-
realistic and plausible. bility assessments. Such research could offer invaluable insights into
One of the website conditions we test is format, and we alter the same designing and presenting online health information that is effectively
health advice to be presented as either an article or a brief “back-and- tailored to diverse global audiences.
forth” forum exchange. We do not find evidence that this format dif- Our research focuses on the assessment of specific health information
ference impacts how consumers assess source credibility or ultimately resources that were chosen for the participants. This boundary is
changes their willingness to follow the advice. However, to keep the necessary to isolate and study the assessment process of consumers as it
content nearly identical, the forum treatments portray two participants pertains to peripheral cues and source credibility of online health

12
J.L. Claggett et al. Information & Management 61 (2024) 104037

information. However, we acknowledge it is unnatural to separate the [3] L. Thunström, M. Ashworth, D. Finnoff, S.C. Newbold, Hesitancy Toward a
COVID-19 Vaccine, EcoHealth 18 (2021) 44–60, https://doi.org/10.1007/
online search process from the assessment of the results. A natural
s10393-021-01524-0.
extension of this work would be to study search behaviors and how they [4] O. Yaqub, S. Castle-Clarke, N. Sevdalis, J. Chataway, Attitudes to vaccination: A
interact with credibility assessments. Earlier studies suggest that people critical review, Soc. Sci. Med. (1967) 112 (2014) 1–11, https://doi.org/10.1016/
rarely go beyond the first couple of search results, even when highly j.socscimed.2014.04.018.
[5] F. Farooq, F.A. Rathore, COVID-19 Vaccination and the Challenge of Infodemic
motivated [108,109]. In information-abundant environments, this sat- and Disinformation, J. Korean Med. Sci. 36 (2021), https://doi.org/10.3346/
isficing behavior is rational [110] and necessary to leave cognitive en- jkms.2021.36.e78.
ergy to process the information [54]. However, the relationship between [6] S.M. Jungmann, M. Witthöft, Health anxiety, cyberchondria, and coping in the
current COVID-19 pandemic: Which factors are related to coronavirus anxiety?
credibility assessment and the decision to stop searching is an interesting J. Anxiety. Disord. 73 (2020) 102239 https://doi.org/10.1016/j.
one to explore. janxdis.2020.102239.
We limited our study to Web 2.0 technologies, specifically websites [7] A. Farooq, S. Laato, A.K.M.N. Islam, J. Isoaho, Understanding the impact of
information sources on COVID-19 related preventive measures in Finland,
with forum or article presentations of text-based health information. Technology in Society 65 (2021) 101573, https://doi.org/10.1016/j.
However, as novel Web 3.0 formats, such as short videos on social media techsoc.2021.101573.
platforms, continue to gain popularity, future research should consider [8] Y. Wang, M. McKee, A. Torbica, D. Stuckler, Systematic Literature Review on the
Spread of Health-related Misinformation on Social Media, Soc. Sci. Med. 240
how peripheral cues affecting credibility assessment may change in (2019) 112552, https://doi.org/10.1016/j.socscimed.2019.112552.
these new formats. Another format to consider is AI-based chatbots, [9] S. Dave, S. Reed, Promoting self-advocacy for young adults with chronic and rare
given the rapid advancement of AI-generated content. This AI source conditions, Journal of Pediatric Nursing: Nursing Care of Children and Families
65 (2022) A8–A11, https://doi.org/10.1016/j.pedn.2022.04.017.
may warrant new research to understand how it is perceived compared
[10] D. Schulman-Green, S. Jaser, F. Martin, A. Alonzo, M. Grey, R. McCorkle, N.
to expert or peer-sourced content. Additionally, the tone and format of S. Redeker, N. Reynolds, R. Whittemore, Processes of Self-Management in
AI-generated output may be interpreted differently than traditional Web Chronic Illness, J. Nurs. Scholarsh. 44 (2012) 136–144, https://doi.org/10.1111/
2.0 formats. j.1547-5069.2012.01444.x.
[11] B.W. Hesse, D.E. Nelson, G.L. Kreps, R.T. Croyle, N.K. Arora, B.K. Rimer,
K. Viswanath, Trust and sources of health information: the impact of the Internet
CRediT authorship contribution statement and its implications for health care providers: findings from the first Health
Information National Trends Survey, Arch. Intern. Med. 165 (2005) 2618–2624,
https://doi.org/10.1001/archinte.165.22.2618.
Jennifer L. Claggett: Writing – review & editing, Writing – original [12] S.S.L. Tan, N. Goonawardene, Internet Health Information Seeking and the
draft, Visualization, Project administration, Methodology, Investigation, Patient-Physician Relationship: A Systematic Review, J. Med. Internet. Res. 19
Formal analysis, Data curation, Conceptualization. Brent Kitchens: (2017) e5729, https://doi.org/10.2196/jmir.5729.
[13] K. Sørensen, S. Van den Broucke, J. Fullam, G. Doyle, J. Pelikan, Z. Slonska,
Writing – review & editing, Software, Resources, Data curation, H. Brand, (HLS-EU) Consortium Health Literacy Project European, Health literacy
Conceptualization. Maria Paino: Writing – review & editing, Formal and public health: A systematic review and integration of definitions and models,
analysis, Data curation. BMC. Public Health 12 (2012) 80, https://doi.org/10.1186/1471-2458-12-80.
[14] X. Wang, R.A. Cohen, Health Information Technology Use Among Adults: United
States, July–December 2022, National Center for Health Statistics, 2023, https://
Author Statement doi.org/10.15620/cdc:133700.
[15] M. Kutner, E. Greenberg, Y. Jin, C. Paulsen, The Health Literacy of America’s
Adults, U.S. Department of Education, 2006.
Jennifer Claggett (Corresponding Author) contributed significantly
[16] A.J. Flanagin, M.J. Metzger, Trusting expert-versus user-generated ratings online:
to the research project by leading the conceptualization, methodology The role of information volume, valence, and consumer characteristics, Comput.
development, and drafting of the manuscript. JC was instrumental in Human. Behav. 29 (2013) 1626–1634.
curating qualitative data, conducting formal analysis, and overseeing [17] X. Zha, J. Zhang, Y. Yan, D. Zha, Sound information seeking in Web 2.0 virtual
communities: the moderating effect of mindfulness, Behav. Inf. Technol. 34
the investigation process. As the project administrator, JC managed the (2015) 920–935, https://doi.org/10.1080/0144929X.2015.1027876.
project’s progression and coordinated the research team’s efforts. JC [18] M. Househ, E. Borycki, A. Kushniruk, Empowering patients through social media:
also took the lead in visualizing the research data and reviewing and The benefits and challenges, Health Informatics. J. 20 (2014) 50–58, https://doi.
org/10.1177/1460458213476969.
editing the manuscript to ensure its accuracy and integrity. [19] D. Mazzoni, E. Cicognani, Sharing experiences and social support requests in an
Brent Kitchens played a crucial role in the project by contributing to Internet forum for patients with systemic lupus erythematosus, J. Health Psychol.
the conceptualization and providing essential resources and software 19 (2014) 689–696, https://doi.org/10.1177/1359105313477674.
[20] J.F. George, G. Giordano, P.A. Tilley, Website credibility and deceiver credibility:
that facilitated the research. BK was responsible for the curation of Expanding Prominence-Interpretation Theory, Comput. Human. Behav. 54
quantitative data, which was vital for the project’s success. Additionally, (2016) 83–93, https://doi.org/10.1016/j.chb.2015.07.065.
BK participated in the review and editing process of the manuscript, [21] M.J. Metzger, Making sense of credibility on the web: Models for evaluating
online information and recommendations for future research, J. Am. Soc. Inf. Sci.
contributing to the refinement and enhancement of the final publication. Technol. 58 (2007) 2078–2091, https://doi.org/10.1002/asi.20672.
Maria Paino focused on the formal analysis of the research data, [22] S.M. Shariff, X. Zhang, M. Sanderson, On the credibility perception of news on
applying statistical and computational techniques to interpret the find- Twitter: Readers, topics and features, Comput. Hum. Behav 75 (2017) 785–796,
https://doi.org/10.1016/j.chb.2017.06.026.
ings accurately. MP also contributed to the review and editing of the
[23] F. Karim, A.A. Oyewande, L.F. Abdalla, R. Chaudhry Ehsanullah, S. Khan, Social
manuscript, ensuring the analytical methods and conclusions were Media Use and Its Connection to Mental Health: A Systematic Review, Cureus. 12
clearly presented and supported by the data. (2020) e8627, https://doi.org/10.7759/cureus.8627.
[24] S.B. Swetland, A.N. Rothrock, H. Andris, B. Davis, L. Nguyen, P. Davis, S.
G. Rothrock, Accuracy of health-related information regarding COVID-19 on
Supplementary materials Twitter during a global pandemic, World Med. Health Policy. 13 (2021) 503–517,
https://doi.org/10.1002/wmh3.468.
Supplementary material associated with this article can be found, in [25] W.S. Jung, M.Y. Chung, E.S. Rhee, The Effects of Attractiveness and Source
Expertise on Online Health Sites, Health Commun. 33 (2018) 962–971, https://
the online version, at doi:10.1016/j.im.2024.104037. doi.org/10.1080/10410236.2017.1323364.
[26] P. Borah, X. Xiao, The Importance of “Likes”: The Interplay of Message Framing,
References Source, and Social Endorsement on Credibility Perceptions of Health Information
on Facebook, J. Health Commun. 23 (2018) 399–411, https://doi.org/10.1080/
10810730.2018.1455770.
[1] S. Laato, A.K.M.N. Islam, M.N. Islam, E. Whelan, What drives unverified
[27] R. Jucks, F.M. Thon, Better to have many opinions than one from an expert?
information sharing and cyberchondria during the COVID-19 pandemic?
Social validation by one trustworthy source versus the masses in online health
European Journal of Information Systems 29 (2020) 288–305, https://doi.org/
forums, Comput. Human. Behav. 70 (2017) 375–381, https://doi.org/10.1016/j.
10.1080/0960085X.2020.1770632.
chb.2017.01.019.
[2] CDC, COVID Data Tracker, 2022. https://covid.cdc.gov/covid-data-tracker
[28] D. Goh, J. Chi, Central or peripheral? Information elaboration cues on childhood
(accessed October 24, 2022).
vaccination in an online parenting forum, Comput. Hum. Behav. 69 (2017)
181–188, https://doi.org/10.1016/j.chb.2016.11.066.

13
J.L. Claggett et al. Information & Management 61 (2024) 104037

[29] T.G. Lindgren, D. Reyes, L. Eller, D. Wantland, C. Portillo, W.L. Holzemer, [56] V. Venkatesh, S.A. Brown, H. Bala, Bridging the Qualitative-Quantitative Divide:
E. Matshediso, I. Corless, J. Kemppainen, K.D. Mogobe, A. Webel, K. Nokes, Guidelines for Conducting Mixed Methods Research in Information Systems, MIS.
P. Nicholas, M.Rivero Mendez, P. Reid, S.S. Baez, M.O. Johnson, Y. Cuca, C. Q. 37 (2013) 21–54.
D. Rose, Understanding Health Literacy for People Living With HIV: Locations of [57] R. Hughes, M. Huby, The application of vignettes in social and nursing research,
Learning, Journal of the Association of Nurses in AIDS Care 29 (2018) 190–203, J. Adv. Nurs. 37 (2002) 382–386, https://doi.org/10.1046/j.1365-
https://doi.org/10.1016/j.jana.2017.10.007. 2648.2002.02100.x.
[30] R.E. Petty, J.T. Cacioppo, The Elaboration Likelihood Model of Persuasion, in: R. [58] H. Aguinis, K.J. Bradley, Best Practice Recommendations for Designing and
E. Petty, J.T. Cacioppo (Eds.), Communication and Persuasion: Central and Implementing Experimental Vignette Methodology Studies, Organ. Res. Methods
Peripheral Routes to Attitude Change, Springer, New York, NY, 1986, pp. 1–24, 17 (2014) 351–371, https://doi.org/10.1177/1094428114547952.
https://doi.org/10.1007/978-1-4612-4964-1_1. [59] J. Nielson, How Long Do Users Stay On Web Pages? Nielsen Norman Group, 2011.
[31] K.P. Hocevar, M. Metzger, A.J. Flanagin, Source Credibility, Expertise, and Trust https://www.nngroup.com/articles/how-long-do-users-stay-on-web-pages/
in Health and Risk Messaging, Oxford Research Encyclopedia of Communication (accessed April 23, 2023).
(2017), https://doi.org/10.1093/acrefore/9780190228613.013.287. [60] M.B. Miles, M. Huberman, Qualitative Data Analysis: An Expanded Sourcebook,
[32] L. Sbaffi, J. Rowley, Trust and Credibility in Web-Based Health Information: A 2nd ed., Sage Publications, Inc, 1994.
Review and Agenda for Future Research, J. Med. Internet Res. 19 (2017) e218, [61] L. Kwasniewicz, G.M. Wojcik, P. Schneider, A. Kawiak, A. Wierzbicki, What to
https://doi.org/10.2196/jmir.7579. Believe? Impact of Knowledge and Message Length on Neural Activity in Message
[33] C. Avgerou, Contextual explanation: alternative approaches and persistent Credibility Evaluation, Front. Hum. Neurosci 15 (2021) 659243, https://doi.org/
challenges, MIS. Q. 43 (2019) 977–1006, https://doi.org/10.25300/MISQ/2019/ 10.3389/fnhum.2021.659243.
13990. [62] R.B. Cialdini, N.J. Goldstein, Social influence: Compliance and conformity, Annu.
[34] G. Johns, The Essential Impact of Context on Organizational Behavior, The Rev. Psychol. 55 (2004) 591–621, https://doi.org/10.1146/annurev.
Academy of Management Review 31 (2006) 386–408. psych.55.090902.142015.
[35] S. Chaiken, Heuristic versus systematic information processing and the use of [63] B.J. Fogg, Prominence-interpretation theory: explaining how people assess
source versus message cues in persuasion, J. Pers. Soc. Psychol. 39 (1980) credibility online. CHI ’03 Extended Abstracts On Human Factors in Computing
752–766, https://doi.org/10.1037/0022-3514.39.5.752. Systems, Association for Computing Machinery, New York, NY, USA, 2003,
[36] W. Schneider, R.M. Shiffrin, Controlled and automatic human information pp. 722–723, https://doi.org/10.1145/765891.765951.
processing: I. Detection, search, and attention, Psychol. Rev. 84 (1977) 1–66, [64] S. Newell, J. Swan, Trust and inter-organizational networking, Human Relations
https://doi.org/10.1037/0033-295X.84.1.1. 53 (2000) 1287–1328, https://doi.org/10.1177/a014106.
[37] S. Tseng, B.J. Fogg, Credibility and computing technology, Commun. ACM 42 [65] S. Senecal, J. Nantel, The influence of online product recommendations on
(1999) 39–44, https://doi.org/10.1145/301353.301402. consumers’ online choices, J. Retail. 80 (2004) 159–169, https://doi.org/
[38] C.M. Angst, R. Agarwal, Adoption of Electronic Health Records in the Presence of 10.1016/j.jretai.2004.04.001.
Privacy Concerns: The Elaboration Likelihood Model and Individual Persuasion, [66] J. Wang, M.D. Molina, S.S. Sundar, When expert recommendation contradicts
MIS. Q. 33 (2009) 339–370. peer opinion: Relative social influence of valence, group identity and artificial
[39] S. Chaiken, A.H. Eagly, Communication modality as a determinant of message intelligence, Comput. Hum. Behav. 107 (2020) 106278, https://doi.org/
persuasiveness and message comprehensibility, J. Pers. Soc. Psychol. 34 (1976) 10.1016/j.chb.2020.106278.
605–614, https://doi.org/10.1037/0022-3514.34.4.605. [67] N. Liu, Y. Tong, H.C. Chan, Dual effects of social support seeking in patient-
[40] A. Bhattacherjee, C. Sanford, Influence processes for information technology centric online healthcare communities: A longitudinal study, Inf. Manage. 57
acceptance: An elaboration likelihood model, MIS. Q. 30 (2006) 805–825. (2020) 103270, https://doi.org/10.1016/j.im.2020.103270.
[41] D. Cyr, M. Head, E. Lim, A. Stibe, Using the elaboration likelihood model to [68] V. Hachaturyan, M. Adam, C. Favaretti, M. Greuel, J. Gates, T. Baernighausen,
examine online persuasion through website design, Information & Management A. Vandormael, Reactance to Social Authority in a Sugar Reduction Informational
55 (2018) 807–821, https://doi.org/10.1016/j.im.2018.03.009. Video: Web-Based Randomized Controlled Trial of 4013 Participants, J. Med.
[42] S.Y. Ho, D. Bodoff, The Effects of Web Personalization on User Attitude and Internet Res. 23 (2021) e29664, https://doi.org/10.2196/29664.
Behavior: An Integration of the Elaboration Likelihood Model and Consumer [69] T. O’Reilly, What is Web 2.0: Design Patterns and Business Models for the Next
Search Theory, MIS. Q. 38 (2014), https://doi.org/10.25300/MISQ/2014/ Generation of Software, Communications & Strategies (2007). http://papers.ssrn.
38.2.08, 497-+. com/abstract=1008839 (accessed June 20, 2016).
[43] C.N. Wathen, J. Burkell, Believe it or not: Factors influencing credibility on the [70] S. Wattal, D. Schuff, M. Mandviwalla, C.B. Williams, Web 2.0 and Politics: The
Web, J. Am. Soc. Inf. Sci. Technol. 53 (2002) 134–144, https://doi.org/10.1002/ 2008 Us Presidential Election and an E-Politics Research Agenda, MIS. Q. 34
asi.10016. (2010) 669–688.
[44] A. Appelman, S.S. Sundar, Measuring Message Credibility: Construction and [71] R.Y.K. Lau, S.S.Y. Liao, K.F. Wong, D.K.W. Chiu, Web 2.0 Environmental
Validation of an Exclusive Scale, Journal. Mass Commun. Q. 93 (2016) 59–79, Scanning and Adaptive Decision Support for Business Mergers and Acquisitions,
https://doi.org/10.1177/1077699015606057. MIS. Q. 36 (2012) 1239–1268.
[45] J.P. Dillard, L. Shen (Eds.), The SAGE Handbook of Persuasion: Developments in [72] G.C. Kane, R.G. Fichman, The Shoemaker’s Children: Using Wikis for Information
Theory and Practice, Second edition, SAGE Publications, Inc, Los Angeles London Systems Teaching, Research, and Publication, MIS. Q. 33 (2009) 1–17.
New Delhi Singapore Washington DC, 2012. [73] T. Lagu, E.J. Kaufman, D.A. Asch, K. Armstrong, Content of Weblogs Written by
[46] R.E. Petty, J.A. Kasmer, C.P. Haugtvedt, J.T. Cacioppo, Source and message Health Professionals, J. Gen. Intern. Med. 23 (2008) 1642, https://doi.org/
factors in persuasion: A reply to stiff’s critique of the elaboration likelihood 10.1007/s11606-008-0726-6.
model, Commun. Monogr. 54 (1987) 233–249, https://doi.org/10.1080/ [74] L.F. Pendry, J. Salvatore, Individual and social benefits of online discussion
03637758709390229. forums, Comput. Human. Behav. 50 (2015) 211–220, https://doi.org/10.1016/j.
[47] S. Taylor, M. Graff, R. Taylor, How can you persuade me online? The impact of chb.2015.03.067.
goal-driven motivations on attention to online information, Comput. Human. [75] C. Boothby, D. Murray, A.P. Waggy, A. Tsou, C.R. Sugimoto, Credibility of
Behav. 105 (2020) 106210, https://doi.org/10.1016/j.chb.2019.106210. scientific information on social media: Variation by platform, genre and presence
[48] M.S. North, S.T. Fiske, A history of social cognition. Handbook of the History of of formal credibility cues, Quant. Sci. Stud. 2 (2021) 00151, https://doi.org/
Social Psychology, Psychology Press, New York, NY, US, 2012, pp. 81–99. 10.1162/qss_a_00151.
[49] W. Choi, B. Stvilia, Web credibility assessment: Conceptualization, [76] A.J. Flanagin, M.J. Metzger, The role of site features, user attributes, and
operationalization, variability, and models, J. Assoc. Inf. Sci. Technol. 66 (2015) information verification behaviors on the perceived credibility of web-based
2399–2414, https://doi.org/10.1002/asi.23543. information, New. Media Soc. 9 (2007) 319–342, https://doi.org/10.1177/
[50] S. Li, Y. Liu, J. Su, L. Duan, H. Fu, Men resist men: streamer-consumer gender 1461444807075015.
match for advertising the functional benefits of heterogeneously priced utilitarian [77] B.H. Schmitt, Customer Experience Management: A Revolutionary Approach to
products, Hum. Soc. Sci. Commun. 11 (2024) 550, https://doi.org/10.1057/ Connecting with Your Customers, John Wiley & Sons, 2010.
s41599-024-03039-9. [78] A.M. Urdea, C.P. Constantin, Experts’ Perspective on the Development of
[51] J. She, T. Zhang, J. Zhang, Q. Chang, Q. Chen, W. Fan, Y. Li, How do post content Experiential Marketing Strategy: Implementation Steps, Benefits, and Challenges,
and poster characteristics affect the perceived usefulness of user-generated J. Risk. Financ. Manage 14 (2021) 502, https://doi.org/10.3390/jrfm14100502.
content? Electron. Commer. Res. Appl. 65 (2024) 101395 https://doi.org/ [79] L. Leahu, S. Schwenk, P. Sengers, Subjective objectivity: negotiating emotional
10.1016/j.elerap.2024.101395. meaning, in: Proceedings of the 7th ACM Conference on Designing Interactive
[52] T.H. Allison, B.C. Davis, J.W. Webb, J.C. Short, Persuasion in crowdfunding: An Systems, Association for Computing Machinery, New York, NY, USA, 2008,
elaboration likelihood model of crowdfunding performance, J. Bus. Ventur. 32 pp. 425–434, https://doi.org/10.1145/1394445.1394491.
(2017) 707–725, https://doi.org/10.1016/j.jbusvent.2017.09.002. [80] B. Akdeniz, R.J. Calantone, C.M. Voorhees, Effectiveness of Marketing Cues on
[53] T.K. Koch, L. Frischlich, E. Lermer, Effects of fact-checking warning labels and Consumer Perceptions of Quality: The Moderating Roles of Brand Reputation and
social endorsement cues on climate change fake news credibility and engagement Third-Party Information, Psychol. Mark. 30 (2013) 76–89, https://doi.org/
on social media, J. Appl. Soc. Psychol. 53 (2023) 495–507, https://doi.org/ 10.1002/mar.20590.
10.1111/jasp.12959. [81] J. Estill, Shorter is Better, Copyblogger, 2008. https://copyblogger.com/shorter-i
[54] M.J. Metzger, A.J. Flanagin, Using Web 2.0 technologies to enhance evidence- s-better/ (accessed July 28, 2020).
based medical information, J. Health Commun. 16 (2011) 45–58. [82] N. Ibrahim, M.F. Shiratuddin, K.W. Wong, Persuasion techniques for tourism
[55] K.S. Freeman, J.H. Spyridakis, Effect of Contact Information on the Credibility of website design, in: Proceedings of the International Conference on E-
Online Health Information, IEEE Trans. Prof. Commun. 52 (2009) 152–166, Technologies and Business on the Web, Bangkok, Thailand, 2013. https://resear
https://doi.org/10.1109/TPC.2009.2017992. chrepository.murdoch.edu.au/id/eprint/22861/ (accessed August 10, 2020).

14
J.L. Claggett et al. Information & Management 61 (2024) 104037

[83] T.C. Boas, D.P. Christenson, D.M. Glick, Recruiting large online samples in the [101] V. Venkatesh, S.A. Brown, Y.W. Sullivan, Guidelines for Conducting Mixed-
United States and India: Facebook, Mechanical Turk, and Qualtrics, Political. Sci. methods Research: An Extension and Illustration, J. Assoc. Inf. Syst. 17 (2016)
Res. Methods (2018) 1–19, https://doi.org/10.1017/psrm.2018.28. 435–494, https://doi.org/10.17705/1jais.00433.
[84] Q.V. Liao, W.T. Fu, Age differences in credibility judgments of online health [102] S. Fox, M. Duggan, Health Online 2013 | Pew Research Center, Pew Research
information, ACM Trans. Comput.Hum. Interact. 21 (2014) 1–23, https://doi. Center, 2013. https://www.pewinternet.org/2013/01/15/health-online-2013/
org/10.1145/2534410. (accessed April 3, 2019).
[85] A. Ghose, A. Goldfarb, S.P. Han, How Is the Mobile Internet Different? Search [103] S. Loomba, A. de Figueiredo, S.J. Piatek, K. de Graaf, H.J. Larson, Measuring the
Costs and Local Activities, Inf. Syst. Res 24 (2013) 613–631, https://doi.org/ impact of COVID-19 vaccine misinformation on vaccination intent in the UK and
10.1287/isre.1120.0453. USA, Nat. Hum. Behav. 5 (2021) 337–348, https://doi.org/10.1038/s41562-021-
[86] K. Fiscella, S. Meldrum, P. Franks, C.G. Shields, P. Duberstein, S.H. McDaniel, R. 01056-1.
M. Epstein, Patient Trust: Is It Related to Patient-Centered Behavior of Primary [104] J.E. McGrath, Dilemmatics: The Study of Research Choices and Dilemmas,
Care Physicians? Med. Care 42 (2004) 1049–1055. American Behavioral Scientist 25 (1981) 179–197.
[87] D.H. Thom, M.A. Hall, L.G. Pawlson, Measuring Patients’ Trust In Physicians [105] M. Khosrowjerdi, National culture and trust in online health information, Journal
When Assessing Quality Of Care, Health Aff. 23 (2004) 124–132, https://doi.org/ of Librarianship and Information Science 52 (2020) 509–528, https://doi.org/
10.1377/hlthaff.23.4.124. 10.1177/0961000619836716.
[88] L.A. Anderson, R.F. Dedrick, Development of the Trust in Physician Scale: A [106] J. Liu, P.L. Rau, N. Wendler, Trust and online information-sharing in close
Measure to Assess Interpersonal Trust in Patient-Physician Relationships, Psychol. relationships: A cross-cultural perspective, Behaviour & Information Technology
Rep. 67 (1990) 1091–1100, https://doi.org/10.2466/pr0.1990.67.3f.1091. 34 (2014) 1–12, https://doi.org/10.1080/0144929X.2014.937458.
[89] S.W. Sussman, W.S. Siegal, Informational influence in organizations: An [107] A. Vance, C. Elie-Dit-Cosaque, D. Straub, Examining Trust in Information
integrated approach to knowledge adoption, Inf. Syst. Res. 14 (2003) 47–65, Technology Artifacts: The Effects of System Quality and Culture, J. Manage. Inf.
https://doi.org/10.1287/isre.14.1.47.14767. Syst. 24 (2008) 73–100, https://doi.org/10.2753/MIS0742-1222240403.
[90] M.K. Lindell, D.J. Whitney, Accounting for common method variance in cross- [108] J.M. Morahan-Martin, How Internet Users Find, Evaluate, and Use Online Health
sectional research designs, Journal of Applied Psychology 86 (2001) 114–121, Information: A Cross-Cultural Review, CyberPsychology & Behavior 7 (2004)
https://doi.org/10.1037/0021-9010.86.1.114. 497–510, https://doi.org/10.1089/cpb.2004.7.497.
[91] W.W. Chin, J.B. Thatcher, R.T. Wright, Assessing Common Method Bias: [109] C. Prabha, L. Silipigni Connaway, L. Olszewski, L.R. Jenkins, What is enough?
Problems with the ULMC Technique, MIS. Q. 36 (2012) 1003–1019. Satisficing information needs, Journal of Documentation 63 (2007) 74–89,
[92] R.M. Baron, D.A. Kenny, The moderator-mediator variable distinction in social https://doi.org/10.1108/00220410710723894.
psychological research: conceptual, strategic, and statistical considerations, [110] H.A. Simon, Administrative Behavior, 2nd Edition, The Free Press, 1957.
J. Pers. Soc. Psychol. 51 (1986) 1173–1182.
[93] A.F. Hayes, Introduction to Mediation, Moderation, and Conditional Process
JENNIFER CLAGGETT ([email protected]; corresponding author) is an associate professor of
Analysis: A Regression-Based Approach, 1 Edition, The Guilford Press, New York,
Management Information Systems and the AT&T Faculty Fellow at the School of Business
2013.
at Wake Forest University. She is also a Champion Member of the Center for Health
[94] P.E. Shrout, N. Bolger, Mediation in experimental and nonexperimental studies:
Innovation at the Wake Forest University School of Medicine. Her research primarily fo-
new procedures and recommendations, Psychol. Methods 7 (2002) 422–445.
cuses on health information technology, investigating how individuals adopt and utilize
[95] M.E. Sobel, Asymptotic Confidence Intervals for Indirect Effects in Structural
technology to transform processes and enhance outcomes. This work bridges technology
Equation Models, Sociol. Methodol. 13 (1982) 290–312, https://doi.org/
and management, aiming to advance business practices through innovative information
10.2307/270723.
systems solutions.
[96] S. Dineen-Griffin, V. Garcia-Cardenas, K. Williams, S.I. Benrimoj, Helping patients
help themselves: A systematic review of self-management support strategies in
primary health care practice, PLoS. One 14 (2019) e0220116, https://doi.org/ BRENT KITCHENS is the William Stamps Farish Associate Professor of Commerce and Director
10.1371/journal.pone.0220116. of the Center for Business Analytics at the McIntire School of Commerce at the University
[97] T. Song, F. Liu, N. Deng, S. Qian, T. Cui, Y. Guan, L. Arnolda, Z. Zhang, P. Yu, of Virginia. He received his Ph.D. from the Warrington College of Business at the Uni-
A Comprehensive 6A Framework for Improving Patient Self-Management of versity of Florida. His research focuses on how online platforms influence information
Hypertension Using mHealth Services: Qualitative Thematic Analysis, J. Med. consumption behaviors, as well as how online platforms and the ways information is
Internet. Res. 23 (2021) e25522, https://doi.org/10.2196/25522. shared on them can be designed and used for societal benefit. His research has appeared in
[98] M.S. Islam, A.H.M. Kamal, A. Kabir, D.L. Southern, S.H. Khan, S.M.M. Hasan, various high quality publications, including MIS Quarterly, Information Systems Research,
T. Sarkar, S. Sharmin, S. Das, T. Roy, M.G.D. Harun, A.A. Chughtai, N. Homaira, Journal of Management Information Systems, IEEE Transactions on Knowledge and Data En-
H. Seale, COVID-19 vaccine rumors and conspiracy theories: The need for gineering, and the Journal of Medical Internet Research. His research has been awarded the
cognitive inoculation against misinformation to improve vaccine adherence, AIS Senior Scholars and MIS Quarterly best paper awards, and has been featured in Har-
PLoS. One 16 (2021) e0251605, https://doi.org/10.1371/journal.pone.0251605. vard Business Review, CBS News, The New York Times, The Washington Post, Wired, and other
[99] S. Pluviano, C. Watt, S. Pompéia, R. Ekuni, S.Della Sala, Forming and updating outlets.
vaccination beliefs: does the continued effect of misinformation depend on what
we think we know? Cogn. Process. 23 (2022) 367–378, https://doi.org/10.1007/
MARIA PAINO is an Associate Professor in the Department of Anthropology, Social Work and
s10339-022-01093-2.
Criminal Justice at Oakland University. Her research focuses on inequalities and organi-
[100] B.J. Fogg, Persuasive Technology: Using Computers to Change What We Think
zational processes particularly in education, healthcare, and clinical contexts.
and Do, 1st edition, Morgan Kaufmann, Amsterdam ; Boston, 2002.

15