A Healthy Diet According to the Internet: The Effect of Expert and Layperson Sources on Source and Message Credibility among Older and Younger Adults

A Healthy Diet According to the Internet:

the Effect of Expert and Layperson Sources on Source and Message Credibility among Older and Younger Adults

Fabienne van Rooij (11150300) Master’s Thesis

Master Communication Science – Persuasive Communication Graduate School of Communication

University of Amsterdam

Supervisor: Margot van der Goot Date: June 25th ₂₀₂₀

Abstract

Health forums are a popular online communication venue used for finding answers to one’s personal health questions. Contributors to these forums are laypersons and experts, making the quality of the information presented there highly variable. Differences in information from expert and layperson sources may foster the already existing confusion about some health topics, such as the topic of nutrition. The current experimental study analyzed how people perceive the credibility of expert and layperson sources and the information they provide. Indicating the significance of authority cues for credibility

evaluations, this study showed that visitors of online health forums perceive expert sources as more credible than layperson sources. Source credibility in turn was positively correlated with message credibility. Therefore, expert sources are considered to be more credible sources and in turn have a higher message credibility than layperson sources. The research also analyzed the moderating role of age, as older adults grew up with the traditional hierarchical doctor-patient relationship in which people were passive recipients of health information, while younger adults grew up with the internet, enhancing their autonomy in receiving health information. The study showed that older and younger adults do not differ in how they

evaluate the credibility of expert and layperson sources. Finally, implications for research and practice are discussed.

A Healthy Diet According to the Internet: the Effect of Expert and Layperson Sources on Source and Message Credibility among Older and Younger Adults

The internet has enabled people to proactively search for and read about information on various kinds of topics. According to Statistics Netherlands, 70% of older adults and 80% of younger adults use the internet to read about health-related information (2019). The more active role that people take in finding health-related information online is in contrast with traditional health communication, in which people passively received information from health professionals who they trusted completely (Fan, Lederman, Smith & Chang, 2014; Petracci, Schwartz, Sánchez Antelo & Mendes Diz, 2019). While it seems beneficial that health professionals are no longer the only ones to provide information, the internet contains a lot of information that is unverified, inaccurate or biased (Linn, van Weert, Gebeyehu, Sanders, Diviani, Smit & van Dijk, 2018). People are thus required to evaluate the information they encounter critically.

A frequently searched for health-related topic is nutrition (Colditz, Woods & Primack, 2018; Heuberger & Ivanitskaya, 2011; Kaminski, Skonieczna-Żydecka, Nowak, &

Stachowska, 2020). Nutrition is important for maintaining health and to prevent diseases (Minich & Bland, 2013). Nutrition is particularly important for the older population, as the sufficient intake of healthy food is correlated with the prevention of health issues and positively affects one’s aging process (Lee, Huang, Zhang & Xu, 2015). However, nutrition information is confusing as conceptions of what a healthy diet entails are constantly changing and are sometimes even contradictory (Clark, Nagler & Niederdeppe, 2019; Ramachandran et al., 2018). For example, there is conflicting information on the benefits and risks of the consumption of products such as wine, coffee and fish (Greiner, Clegg Smith & Guallar, 2010; Nagler, 2010). Most problematically, this confusion may translate into doubts about general nutrition recommendations, such as sufficient vegetable intake (Nagler, 2010). This in

turn may result in adverse public health effects (Clark et al., 2019). It is therefore important to understand how people evaluate nutrition information online.

When turning to Google for information, one may encounter information from expert and layperson sources. In the current day, a type of website that frequently pops up with regards to health questions is forums (Fan et al., 2014). Online forums are virtual discussion groups where experts and laypersons may share their knowledge in response to questions (Zhang, Liu, Chen, Wang, Gao & Zhu, 2018). The information discussed there is useful to both participants of the forum and others who browse the internet in search for answers to their questions (Winzelberg, 1997; Zimmerman & Jucks, 2018). Experts that provide

information in these forums base their answers on scientific evidence. On the other hand, the majority of information is posted by laypersons, who base the information they provide on personal experiences (Fan et al., 2014). This makes the quality of information highly variable. Credibility is a factor that people use to assess the quality of sources and information (Ma & Atkin, 2017). Therefore, the first aim of the current study is to determine how expert and layperson sources in online health forums affect credibility perceptions.

As health communication has undergone numerous transformations over the past decades (Petracci et al., 2019), there is a difference between the older and younger population in how they were socialized with regards to making health-related decisions. Older adults, those people born before 1970, have grown up in the era when the traditional doctor-patient relationship dominated health communication (Huisman, Joye & Biltereyst, 2019). In

contrast, younger adults, specifically those people born since 1990, have been raised with the internet and generally have a stronger desire for autonomy in health decisions (Xie, Wang, Feldman & Zhou, 2012). This generation gap possibly results in differences in evaluations of expert and layperson sources too. Therefore, the second aim of the current study is to

expert and layperson sources. In sum, the current study addresses the following research question:

Research question: How do expert and layperson sources of online nutrition information affect credibility perceptions and is this effect moderated by age?

Academic relevance

The current study adds to prior research in several ways. Until now, research on online health forums has been predominantly qualitative (Lederman, Fan, Smith & Chang, 2013; Mendes, Abreu, Vilar-Correai & Borlido-Santos, 2016; Vennik, Adams, Fabe & Putters, 2014) and focused on contributors of health forums (Fan et al., 2014; Vennik et al., 2014; Zhang et al., 2018). The current study is novel as it is a quantitative analysis, providing more objective and generalizable evidence on the credibility of sources and information in online health forums. Additionally, this study pertains to visitors of online health forums rather than contributor. Moreover, while the link between source and message credibility has been well-established (Eastin 2006; Sundar, Knobloch-Westerwick & Hastall, 2007; Winter, Kramer, Appel & Schielke, 2010), to our awareness prior research has not yet investigated this

relationship in the context of expert and layperson sources in an online health forum. As Saleh noted (2016), studying the interrelationships between credibility concepts is important as this will show how each of these impacts the overall credibility.

This study also contributes to earlier studies as it provides an understanding of the target group of older adults. Many prior studies have looked into differences in internet skills (Chung, Park, Wang, Fulk &McLaughlin, 2010; Hart, Henwood & Wyatt, 2014; Xie, 2009), cognitive skills (Bol et al., 2016; Liao & Fu, 2012) and motivation (Carstensen, Isaacowitz & Charles, 1999; van der Goot, Bol & van Weert, 2019) between older and younger adults. As health communication has undergone several transformations over the past 40 years, this

study focuses on the generational difference between older and younger adults in terms of their socialization and tests whether this difference between older and younger adults translates to differences in how they perceive the credibility of expert and layperson sources online.

Societal relevance

The current study is socially relevant in two ways. Firstly, findings may help

communication professionals to choose between expert and layperson sources for conveying health information. As source and message credibility have been shown to affect behavior (Fan et al., 2014), choosing a credible source to convey health information may positively affect health behavior. Secondly, findings may show how people evaluate the quality of online sources and information. Generally, people draw upon credibility to assess the quality of sources and information (Ma & Atkin, 2017). As the quality of information in online health forums is highly variable (Fan et al., 2014) and the internet contains a lot of incorrect

information (Linn et al., 2018), findings thus may reveal how susceptible people of different ages are to believe potential incorrect information.

Theoretical Framework Sources of online nutrition information

While traditionally health professionals dominated the provision of health information (Hu & Sundar, 2009), the internet has enabled unqualified individuals to provide information too. Consequently, the main source types of online health information are both experts and laypersons (Hu & Sundar, 2009). Online forums are an example of an online communication venue where both experts and laypersons can provide information (Jung, Walsh-Childers and Kim, 2016). These forums are virtual discussion groups with members discussing various

health topics (Zhang et al, 2018). In some forums, experts provide advice to users, while in other forums laypersons exchange information with each other (Fan et al., 2014; Zimmerman & Jucks, 2018). The audience of an online forum is larger than merely the participants, as many users ‘lurk’ on forums to find information (Winzelberg, 1997; Zimmerman & Jucks, 2018).

Expert and layperson sources and the information they provide in online health forums differ in several ways. Expert sources have an official authority and often rely on professional expertise, such as a specific education or institutional affiliations (Flanagin & Metzger, 2013). Experts on nutrition information are dieticians and general practitioners (Dong, 2015;

Voedingscentrum, 2020). In health forums, expert information is often standard, objective and factual (Lederman et al., 2013) and is focused on informational support (Sanders et al., 2020). Moreover, it is generally knowledge-oriented, prescriptive in style and contains facts and opinions related to the expert’s work (Hartzler & Pratt, 2011).

Layperson sources are people that post information without having an official authority. In health forums, layperson sources provide information about personal experiences that others may relate to (Broom, 2005; Kanthawala, Vermeesch, Given & Huh, 2016; Hartzler & Pratt 2011; Hu & Sundar, 2009). The information is personal and contains a subjective narrative (Lederman et al., 2013) using informal language (Koepfler & Fleischmann, 2011) and focused on providing affective support (Sanders et al., 2020).

Source credibility

A first concept that may show how people evaluate expert and layperson sources is source credibility. Source credibility refers to the degree to which the information provider is

perceived as a credible source of information (Shan, 2016; Tseng & Fogg, 1999). Prior research has focused on a source’s expertise and trustworthiness as the two main components

of source credibility (Borah & Xiao, 2018; Dutta-Bergman, 2004; Hilligoss & Rieh, 2008; O’Keefe 2002; Tseng & Fogg, 1999). Perceived expertise is how users perceive the knowledge, skill and experience of the source, which is related to how they perceive a source’s ability to provide information that is accurate and valid (Hilligoss & Rieh, 2008; McGinnies & Ward, 1980; Shan, 2016). Trustworthiness pertains to the honesty and integrity of the source (McGinnies & Ward, 1980).

Looking at the two dimensions of source credibility, prior research showed that expert sources are perceived as more credible than layperson sources (Lee & Sundar, 2013; Machakova & Smahel, 2018; Sbaffi & Rowley, 2017; Winter et al., 2010). Expert sources often contain authority cues, which are important for conveying credibility (Lee & Sundar, 2013; Sillence & Briggs, 2007). A cue can be defined as a piece of information that helps people to evaluate the information (Kim & Sundar, 2011; Lee & Sundar, 2013). Authority cues refer to a source’s professional expertise and authority by referring to the sources’ knowledge and skills (Lee & Sundar, 2013; Sillence & Briggs, 2007). The Elaboration Likelihood Model of persuasion (ELM) explains how people process these cues. ELM proposes that people may process information centrally, using many cognitive resources, or peripherally, relying on heuristics (Wilson, 2007). According to this model, authority cues are relevant when people process information peripherally (Lee & Sundar, 2013). Authority cues have been shown to positively affect both dimensions of source credibility (i.e., expertise and trustworthiness) (Sillence & Briggs, 2007; Thon & Jucks, 2017). In the case of health forums, prior studies have shown that people considered expert sources to be most credible providers of information (Lee et al., 2015; Winter et al., 2010). Hence, the link between expert sources and source credibility is well-established.

Layperson sources derive their credibility from experiential expertise, which means that they base their knowledge about a health topic on personal experience (Hu & Sundar, 2009;

Lederman et al., 2013; Metzger, Flanagin & Medders, 2010; Sillence and Briggs, 2007). Comparing the effects of expert and layperson sources, Flanagin & Metzger (2013) showed that expert sources containing authority cues are perceived as more credible than layperson sources with experiential expertise. In other words, authority cues are more influential in the evaluation of credibility than experiential expertise is. Similarly, the study by Ramachandran et al., (2018) showed that expert sources are more influential to communicate nutrition information, because there is a lot of confusion about the topic (Clark et al., 2019; Greiner et al., 2010; Nagler, 2010; Ramachandran et al., 2018). Therefore, focusing on the two

dimensions of source credibility, namely, expertise and trustworthiness (Borah & Xiao, 2018; Dutta-Bergman, 2004; Hilligoss & Rieh, 2008; O’Keefe 2002; Tseng & Fogg, 1999), the current study expects that expert sources will be perceived as having more source credibility than layperson sources.

Hypothesis 1: Expert sources are perceived to have more expertise and be more trustworthy than layperson sources (i.e., source credibility is higher for expert sources than for layperson sources).

Message credibility

A second concept that may show how people evaluate expert and layperson sources and the information they provide is message credibility. In contrast to source credibility that focuses on the source of information, message credibility pertains to aspects of the

information in the message. The elements of message credibility are believability, accuracy and authenticity of information (Appelman & Sundar, 2016; Hu & Sundar, 2009). The message can be perceived as highly credible regardless of the source that produced the information (Ma & Atkin, 2017). Indeed, research showed that the source does not influence message credibility when the information is clear (Hu & Sundar, 2009; Saleh, 2016). In

contrast, when information is unclear, source credibility becomes important to judge message credibility (Saleh, 2016). For example, the research by Go, Jung and Wu (2014) showed that when information is unclear, source cues are important for the evaluation of credibility in addition to the message arguments. As nutrition information is often considered to be

confusing (Clark et al., 2019; Greiner et al., 2010; Nagler, 2010; Ramachandran et al., 2018), source credibility is thought to play an important role in the perceived message credibility of nutrition information in online health forums too.

The mediating role of source credibility. Many studies have shown that message credibility relies on source credibility (Eastin 2006; Sundar, Knobloch-Westerwick & Hastall, 2007; Winter et al., 2010). Generally, when people regard a source to be credible, the health information presented is considered to be credible too (Jung et al., 2016; Lederman et al., 2013; Machackova & Smahel, 2018; Thon & Jucks, 2017). As explained by Dong (2015), this link relies on the heuristic that credible sources provide stronger arguments than low credible sources. Thon and Jucks (2017) for example showed that statements from professionals were considered to be more credible as the sources contained authority cues. Similarly, Lee and Sundar (2013) showed that when an expert posted information, the message credibility was higher. Therefore, the current study expects that the information produced by experts is

perceived as more believable, accurate and authentic than information produced by laypersons and that this effect is mediated by source credibility.

Hypothesis 2: Information from expert sources is perceived as more believable, accurate and authentic than information from layperson sources (i.e. message credibility is higher for information from expert sources than for information from layperson sources), and this effect is mediated by source credibility.

The moderating role of age: exploring generational differences

Over the past decades, health communication has transformed tremendously (Petracci et al., 2017). The internet has increased people’s autonomy in their access to health

information, which has, among others, contributed to higher patient satisfaction (Sanders & Linn, 2018). In former times, health professionals dominated the provision of health

information (Hu & Sundar, 2009; Makoul, 1998). As a result, the doctor-patient relationship was characterized by an unequal power balance and information asymmetry in which health professionals were authoritative (Broom, 2005; Huisman et al., 2019) and considered paternal figures whose advice was unquestionable (Petracci et al., 2017). Patients, on the other hand, were passive and compliant (Brody, 1980; Makoul, 1998; Xie, 2009). The traditional doctor-patient relationship began changing from the 1970s onwards (Xie et al., 2012). Older adults, those people born before 1970, were socialized in the era when the traditional doctor-patient relationship dominated, meaning that they grew up with a paternalistic relationship with healthcare providers (Eliassen, 2015). As a result, in today’s society older adults may potentially still exhibit passivity in interaction with health professionals and when encountering online information from expert sources (Eliassen, 2015).

Indeed, reflecting the traditional doctor-patient relationship, prior research showed that older adults perceive online expert sources as more credible than layperson sources. For example, in their study among older adults, Zulman, Kirch, Zheng & An (2011) found that older adults are more trusting of online health information when it is easy to identify the expertise of the source. Studying the effects of nutrition information posted in a health forum, Lee et al. (2015) found that when nutrition information is conflicting, older adults want the input of a health professional to provide clarity. All in all, these findings show that older adults perceive expert sources as more credible than layperson sources.

The internet emerged in the mid-1990s (Simpson, 2004), enabling people to go online to find health information and giving them more control over health decisions (Broom, 2005; Hu & Sundar, 2009). Therefore, in contrast to older adults, younger adults, those people born after 1990, were socialized to have more autonomy and rely less on health professionals to receive information on health topics (Hu & Sundar, 2009; Xie et al., 2012). As a result, health professionals are increasingly viewed as advisors (Petracci et al., 2017). Similar to older adults, younger adults perceive online expert sources as credible (Gray, Klein, Noyce and Sessel, 2005; Mendes et al., 2016) and regard these sources as having expertise and being trustworthy (Dong, 2015; Mendes et al., 2016). However, other studies conducted among younger adults showed that they perceive layperson sources in health forums as having experiential expertise and therefore to be credible sources too (Hirvonen, Tirroniemi & Kortelainen, 2019; Hu & Sundar, 2009). This suggests that younger adults may perceive layperson sources as credible too while older adults perceive expert sources as more credible than layperson sources.

In sum, older adults have been socialized in an era when the paternalistic doctor-patient relationship dominated, while younger adults have grown up with more autonomy in their search for health information. Prior studies have shown that older adults perceive expert sources as more credible (Lee et al., 2015; Zulman et al., 2011) while younger adults perceive both expert and layperson sources as credible (Dong, 2015; Mendes et al., 2016; McGinnies & Ward, 1980; Hirvonen et al., 2019). Therefore, the current study expects that expert sources are perceived as more credible than layperson sources, that this effect is stronger for older than for younger adults, and that higher source credibility in turn leads to higher message credibility. The conceptual model of this study is displayed in Figure 1.

Hypothesis 3: The mediated relationship between source type and message credibility through source credibility is moderated by age, in such a way that the effect of source type on source credibility is stronger for older adults than younger adults.

Figure 1: Conceptual Model

Method Experimental design

This study used a 2 (source type: expert versus layperson) x 2 (age: older versus younger adults) between-subjects factorial experimental design to assess whether the source of online information affects source credibility, if this effect is moderated by age and whether source credibility mediates the effect of source type on message credibility.

Participants

In order to test the predictions, 193 Dutch participants engaged in an online

experiment. 16 participants were excluded from the sample as they fell beyond the required age ranges and their data was not used in the analyses. The final sample consisted of 177 participants (53 males, 122 females and 2 persons identified differently) who all completed the survey. The study distinguishes between two age groups: older adults, those people born before 1970, and younger adults, those people born after 1990. The ages of participants in the older age group ranged from 50 to 79 and the mean age of this group was 58.51 years (SD =

Age: older versus younger adults

Source type: expert

versus layperson Message credibility Source credibility:

expertise and trustworthiness

5.87). The ages of participants in the younger age group ranged from 18 to 30 and the mean age of this age group was 24.29 years (SD = 2.47).

The participants were recruited by means of convenience sampling and snowball sampling. A link to the survey was spread through the researcher’s social media channels and among friends and family. Participants were asked to spread the survey in their personal networks. While a random sampling method would reduce the risk of bias, convenience sampling was considered to be the most practical and cost-effective method for obtaining samples of the target populations. Moreover, this approach ensures that participants are familiar with the internet and health forums. This strengthens the external validity of the research. None of the participants received an incentive to participate.

Stimuli development

To test the expectations, a mock health forum was created. The nutrition topic

discussed in this mock forum was butters and oils. These are products that are frequently used by both older and younger age groups (National Institute for Public Health and the

Environment, 2011). Moreover, this is a health topic about which there is a lot of contradicting information available (O’Connor, 2009) and has been discussed in online forums before. As nutrition information is often considered to be confusing (Clark et al., 2019; Greiner et al., 2010; Nagler, 2010; Ramachandran et al., 2018), which triggers the relevance of source cues in addition to message arguments for the evaluation of credibility (Go et al., 2015), this was considered as an appropriate nutrition topic. Followingly, the information presented would also contain some contradictions.

The lay-out of the mock health forum was the same in the two experimental conditions and showed the question by a user named ‘Bloemenzee’ asking whether margarine, low-fat margarine or olive oil is healthiest. The response to this question was the same in both

conditions and was posted by a user named NettyvanH. The only element that differed between the conditions was the cue. As expert sources often contain authority cues referring to the source’s professional expertise (Kim & Sundar, 2011; Lee & Sundar, 2013; Sillence & Briggs, 2007), the cue used in the expert condition was ‘general practitioner’. The cue that was initially chosen for the layperson source was ‘active contributor in this category’. This was a cue that was considered to reflect the experiential expertise that layperson sources are known for (Hu & Sundar, 2009; Lederman et al., 2013; Metzger et al., 2010; Sillence and Briggs, 2007). A pretest was conducted to test whether these cues sufficiently reflected expert and layperson sources.

Pretest. A quantitative and qualitative pretest was conducted using the survey

software Qualtrics to assess whether the authority cue ‘general practitioner’ used in the expert condition and the cue ‘active contributor in this category’ used in the layperson condition sufficiently reflected an expert and layperson source.

Participants saw both the expert and layperson condition and answered a small questionnaire. Respondents had to rate on a 7-point scale (1 = Totally disagree, 7 = Totally agree) the extent to which they agreed the source of information was an expert or layperson respectively. By means of an open-ended question, respondents were asked how they interpreted the cue ‘active contributor in this category’ that was used in the layperson condition. No such open-ended question was posed for the cue ‘general practitioner’ in the expert condition. Finally, the clarity of the information was asked on a 7-point semantic scale (1 = Totally unclear, 7 = Totally clear). The entire pretest questionnaire can be found in Appendix A.

A total of 15 respondents participated in the pretest (7 males, 8 females). While this is a small sample, it does indicate whether the desired perception is present in the population. The older adults group consisted of 7 individuals, with ages ranging from 53 to 73 and a mean

age of 59 years (SD = 6.40). The younger age group consisted of 8 individuals, with ages ranging from 22 to 28 and a mean age of 24.88 years (SD = 1.73).

A one-sample t-test was conducted to test whether the stimulus in the expert condition was perceived as reflecting an expert source, using the variable ‘expert’ as the test variable and 4 as the test-value, which represents the neutral value. The effect was statistically significant with t (14) = 2.19, p = 0.046, 95% CI [0.02, 1.98]. The mean was 5.0 (SD = 1.78) indicating that people slightly agreed with the statement that the information they saw came from an expert source. This result shows that the stimulus was perceived as coming from an expert source. A one-way ANOVA with ‘expert’ as the test variable and ‘age’ as the factor further showed that older and younger adults did not significantly differ in the extent to which they thought the information was from an expert source with F (1, 13) = 0.32, p = 0.58.

Similarly, a one-sample t-test was executed to test whether the layperson condition was perceived as reflecting the layperson source with the variable ‘layperson’ as the testing variable and a test-value of 4, which represents the neutral value. The effect was not

statistically significant with t (13) = 0.77, p = 0.455, 95% CI [-0.65, 1.36]. The scores ranged from 1 to 5, illustrating that very few participants considered the condition to reflect a

layperson source. Therefore, the cue ‘active contributor in this category’ does not

significantly reflect that the source is a layperson. A one-way ANOVA with ‘layperson’ as the test variable and ‘age’ as the factor further showed that older and younger adults did not differ in the extent to which they perceived the layperson stimulus to reflect a layperson source with F (1, 12) = 0.57, p = 0.464.This result was also reflected in people’s responses to the open-ended question ‘What do you think is meant with the cue ‘active contributor in this category’’? Some participants pointed out that the person commenting was an expert in the field of nutrition while others indicated that they thought the person was just an active

contributor in online health forums. It shows that participants did not clearly perceive the cue ‘active contributor in this category’ to be representing a layperson source.

As nutrition information is often confusing (Clark et al., 2019; Greiner et al., 2010; Nagler, 2010; Ramachandran et al., 2018), the pretest was also conducted to test whether the information in the current study would be ambiguous enough. A one-sample t-test with the test variable ‘information clarity’ and a test-value of 4, representing a neutral score, showed a non-significant effect with t (14), 1.52, p = 0.151, 95% CI [-0.30, 1.77] in the expert condition and t (13) = 1.61, p = 1.32, 95% CI [-0.29, 2.01] in the layperson condition. Therefore, these results show that the information is not significantly clear nor unclear. A one-way ANOVA also showed that older adults did not significantly perceive the information as more confusing than younger adults in both conditions, with F (1, 13) = 1.34, p = 0.268 in the expert

condition and F (1, 12) = 0.00, p = 1.000 in the layperson condition. These results show that the information is nor clear nor unclear, which suggests that the message argument is

ambiguous enough to be used for the main experiment.

In sum, it was decided that both the expert and layperson condition would contain the question ‘What is the healthiest: margarine, low-fat margarine of olive oil?’. The response to this question was the same in both conditions and was posted by a user named NettyvanH. Based on the results of the pretest, the response by NettyvanH would contain the authority cue ‘general practitioner’ in the expert condition, indicating the professional expertise (Lee & Sundar, 2013; Sillence & Briggs, 2007), while this response would contain no cue in the layperson condition. Appendix B shows the definitive stimuli used in the experiment.

Procedure

Before starting the experiment, in line with the ethical procedures, participants were informed about the terms and conditions of the research. Followingly, they could choose to

agree or disagree with a consent statement for their participation. Upon agreement,

participants had to select the age group they belonged to. Afterwards, they were randomly allocated to one of both experimental conditions. Respondents were asked to read the stimulus carefully after which they were redirected to the questionnaire. Firstly, they had to answer items for the measurement of source credibility and message credibility. This was followed by the items for the manipulation check. Finally, they had to answer some demographic

questions. Upon completion of the study, participants were thanked for their participation. The entire questionnaire can be found in Appendix C. The data collection took place over a 1-week timeframe in May 2020.

Measures

Source Credibility. Two dimensions of source credibility were measured: expertise and trustworthiness (Borah & Xiao, 2018; Dutta-Bergman, 2004; Hilligoss & Rieh, 2008; O’Keefe 2002; Tseng & Fogg, 1999). To measure both dimensions, participants were asked: “I think the author of the information is …”. In response to these items, they had to rate on a seven-point semantic differential scale (1 = negative valence words, 7 = positive valence words) a total of ten items. The first dimension of source credibility, expertise, was measured using five items from an expertise scale (Ohanian, 1990): ‘not expert – expert’,

‘inexperienced – experienced’, ‘unknowledgeable – knowledgeable’, ‘unqualified –

qualified’, ‘unskilled – skilled’. A principal component analysis with an orthogonal rotation showed that all five items load on one underlying construct with an eigenvalue greater than 1 (eigenvalue 3.78), represented by a clear point of inflexion after this point in the scree plot. The variance that it explained was 75.6%. The reliability of the scale was good, Cronbach’s Alpha = .92. The mean of these five items showed a score for expertise. Higher scores

indicated higher levels of expertise while lower scores indicated lower levels of expertise. The resulting scale had a mean of 4.52 (SD = 1.04).

The second dimension of source credibility, trustworthiness, was measured using five items from a trustworthiness scale (Ohanian, 1990): ‘undependable – dependable’, ‘dishonest - honest’, ‘unreliable - reliable’, ‘insincere – sincere’ and ‘untrustworthy – trustworthy’. A principal component analysis with an orthogonal rotation showed that the five items load on one factor: only one component had an eigenvalue greater than 1 (eigenvalue 3.98), which was also shown by a point of inflexion after this point in the scree plot. The explained

variance of this component was 79.69%. The resulting scale was reliable, Cronbach’s Alpha = .91. The average of these five items indicated a score for trustworthiness. Higher scores indicated more higher levels of trustworthiness and lower scores indicated lower levels of trustworthiness. The resulting scale had a mean of 4.80 (SD = 1.22).

Message Credibility. This variable can be defined as the level of believability, accuracy and authenticity of information (Appelman & Sundar, 2016). It was measured by three items on a seven-point semantic differential scale (1 = negative valence words, 7 = positive valence words) (Appelman & Sundar, 2016; Jäger & Weber, 2020). Participants were asked to indicate the extent to which they thought the information they read was:

‘unbelievable – believable’, ‘inaccurate – accurate’ and ‘inauthentic – authentic’. A principal component analysis with varimax rotation showed that all three items load on one factor that had an eigenvalue greater than 1 (eigenvalue 2.41). The scree plot showed this with only one clear point of inflexion. The explained variance was 80.3%. The resulting scale was reliable, Cronbach’s Alpha = .88. The mean of these three items described the level of message credibility. Higher scores indicated a higher level of message credibility and lower scores lower levels of message credibility. The scale was averaged with a mean of 4.68 (SD = 1.16). The components matrix of these measurements can be found in Appendix D.

Manipulation Check. To check whether the manipulation was successful, participants rated two items on a 7-point Likert scale (1 = completely disagree, 7 = completely agree): ‘The information I read was posted by an expert’ and ‘The information I read was posted by a layperson’. The items were presented in randomized order. The manipulation was considered successful when participants scored higher on the item corresponding with the experimental condition they were assigned to and lower on the item reflecting the other condition.

Age. This quasi-experimental factor contained two levels, namely: older adults and younger adults. It was operationalized by means of an open question asking participants to fill in their age. The variable was recoded to contain two categories, namely, older adults and younger adults.

Gender. This variable was operationalized by asking the participants how they identified themselves, with the answer options ‘male’, ‘female’, ‘other’ and ‘I prefer not to say’.

Education level. This variable was operationalized by asking participants what their highest achieved education level is. The answer options were ‘Primary school/High school’, ‘MBO, ‘HBO and ‘University/Master’. Participants were required to choose one.

Results Randomization Check

A total of 177 respondents participated in the experiment. 47 older adults and 43 younger adults were assigned to the expert condition and 41 older adults and 46 younger adults were assigned to the layperson condition. In order to check whether gender and education level were equally distributed across the experimental conditions and the age groups, four Chi-square analyses were conducted. The experimental conditions did not differ in terms of gender, X2 _{(3, N = 177) = 4.00, p = 0.261, and education, X}2 _{(3, N = 177) = 2.37, p}

= 0.500. The age groups did not differ in terms of gender either, X2 _{(3, N = 177) = 2.60, p =} 0.458. However, the age groups did differ in terms of education level, X2 _{(3, N = 177) = 43.22,} p < 0.001, with a moderate to strong effect, d = -0.54. As shown in Table 2, compared to older adults (4.0%), more younger adults (23.2%) finished university. More older adults only finished high school (6.2%) or obtained an MBO degree (17.5%) compared to younger adults (0.6% high school and 5.6% MBO). These findings indicate that older adults had a lower education level than younger adults (see Appendix E for a table displaying the results). Therefore, education level was taken into account as a control variable in further analyses.

Manipulation Checks

In order to test if participants perceived the conditions to be an expert and layperson source in accordance with the experimental condition they were assigned to, a manipulation check was conducted by means of two items.

To test whether the manipulation of the source types succeeded, an independent samples t-test was conducted with source type as the grouping variable and the manipulation check items as the test variables. The analysis showed that the mean difference (of -0.69) between the experimental conditions in the extent to which participants consider the source type to be an expert was statistically significant, t (175) = 3.18, p = .002, 95% CI [1.12, -0.26] representing a medium effect, d = 0.48. Participants who saw the post with the authority cue scored higher (M = 4.58, SD = 1.52) than participants who saw the post without the authority cue (M = 3.89, SD = 1.37). Likewise, there was a statistically significant mean difference (of 1.01) between the experimental conditions in the extent to which participants consider the source type to be a layperson, t (175) = 4.43, p < .001, 95% CI [0.56, 1.46], representing a medium effect, d = .67. Participants who saw the post without the authority cue

scored higher (M = 4.38, SD = 1.43) than participants who saw the post with the authority cue (M = 3.37, SD = 1.60). Therefore, we can conclude that the manipulation succeeded.

Main analyses

The PROCESS v3.5 Macro extension (Hayes, 2018) was used to test the predictions. In line with prior research (Ho & Nesbit, 2018; Khan, Yang, Shafi & Yang, 2019), this study analyzed the direct and mediation effects (H1 and H2) using Model 4 followed by a

moderated mediation analysis (for H3) using Model 7.

Direct effects of source type. The effect of source type on source credibility (H1) and the mediating role of source credibility between source type and message credibility (H2) were analyzed using Model 4. Source type was included as the independent variable and message credibility as the dependent variable. The two dimensions of source credibility, expertise and trustworthiness, were entered as the mediators. Education level was included as a covariate. Following a path analysis, the following section will start with elaborating on the effects of source type on source credibility (H1). Afterwards, to test the mediation effect (H2), the effects of expertise and trustworthiness on message credibility and the total indirect

effects are analyzed.

With regards to the relationship between source type and expertise (which is the first dimension of source credibility), the results showed that the regression model was statistically significant F (6, 170) = 4.64, p = 0.001. This means that the model could be used for

predicting expertise. However, the strength of the prediction was weak to moderate: 10% of the variation in expertise could be predicted on the basis of source type (R2 _{= .10). The effect} of source type on expertise was statistically significant with, all else held equal, b = 0.51, t = 3.34, p = .001, 95% CI [0.21, 0.81]. This means that with 95% confidence, expert sources on average score 0.51 higher on expertise than layperson sources in the population.

For trustworthiness (which is the second dimension of source credibility) the analysis showed that the regression model was statistically significant with F (6, 170) = 4.99, p = 0.001. The model could therefore be used to predict variation in trustworthiness, but the strength of the prediction was weak to moderate: 10% of the variation in trustworthiness could be predicted on the basis of source type (R2 _{= 0.10). The effect of source type on} trustworthiness was statistically significant, b = 0.41, t = 2.54, p = .01, 95% CI [0.09, 0.74], all else held equal. This means that with 95% confidence, expert sources on average score 0.41 higher on trustworthiness than layperson sources. In sum, expert sources score higher on both dimensions of source credibility (i.e. expertise and trustworthiness) than layperson sources. This suggests that expert sources have higher source credibility than layperson sources. Therefore, H1 is supported.

The mediating effect of source credibility. For testing the mediating role of source credibility in the relationship between source type and message credibility (H2), the results showed that the regression model predicting message credibility was statistically significant, F (6, 170) = 64.88, p < 0.001. The strength of the prediction was strong: 70% of the variation in message credibility could be predicted on the basis of source type, expertise and

trustworthiness (R2 _{= .70). The effect of source type on message credibility, all else held} equal, was statistically significant, b = 0.29, t = .285, p = .005, 95% CI [0.49, 0.09]. This result suggests that expert sources on average score .29 higher on message credibility than layperson sources. With regards to expertise, the first dimension of source credibility, the results showed a statistically significant effect of expertise on message credibility, b = 0.61, t = 7.79, p = < 0.05, 95% CI [0.45, 0.76]. In other words, when expertise increases with one unit, message credibility increases with 0.61. The indirect effect of source type on message credibility through expertise was statistically significant, as the confidence intervals do not

include zero, b = 0.27, 95% CI [ 0.11, 0.46]. All in all, these results show that expertise mediates the effect of source type on message credibility

The effect of trustworthiness, the second dimension of source credibility, on message credibility was also statistically significant, b = 0.36, t = 4.97, p = < 0.05, 95% CI [0.49, -0.09]. This means that when trustworthiness increases with one unit, message credibility increases with 0.36. As the confidence intervals do not include zero, the mediation as a whole is statistically significant, b = 0.13, 95% CI [ 0.03, 0.27]. In sum, these results show that expert sources have a higher source credibility (i.e. expertise and trustworthiness) than layperson sources, this in turn leads to a higher message credibility (see Appendix F for an overview of all Model 4 results). H2 is supported.

Figure 1. Illustration of the main and mediator effects (N = 177), * p < .05 ** p <.01 ***p <.001

The moderating effect of age. This section uses Model 7 (Hayes, 2018) to test whether the effect of source type on message credibility through source credibility is moderated by age. In the model, source type was included as the independent variable, the source credibility dimensions expertise and trustworthiness as the mediators, age group as the moderator, education level as a covariate and message credibility as the dependent variable. Source Type Expertise Trustworthiness Message Credibility .51** _.61*** .29** .41* .36***

Indirect effect through expertise, b = 0.27, 95% CI [ .11, .46]. Indirect effect through trustworthiness, b = 0.13, 95% CI [ .03, .27].

With regards to expertise, the results showed that age does not have a statistically significant effect on expertise, b = -0.18, t = -.75, p = .45, 95% CI [-0.65, 0.29], all else held equal. Additionally, there was no statistically significant interaction effect between source type and age group on expertise, b = -0.14, t = .44, p = .66, 95% CI [-.47, .74]. Lastly, the results indicated that there is no moderated mediation effect, index = .08, 95% CI [-0.27, 0.48].

For the dimension of trustworthiness, the results also indicated no statistically significant effect of age on trustworthiness, b = -0.45, t = -1.78, p = .078, 95% CI [-0.95, 0.05]. Moreover, there was no statistically significant interaction effect between source type and age group on trustworthiness b = 0.41, t = 1.26, p = .209, 95% CI [-0.23, 1.05]. Finally, the results showed no statistically significant moderated mediation effect, index = .15, 95% CI [-0.07, 0.43]. These results show that age does not moderate the mediated relationship of source type and message credibility through source credibility (see Appendix G for an overview of all Model 7 results). H3 is rejected.

Figure 2. Illustration of the moderated mediation effects (N = 177), * p < .05 ** p <.01 ***p <.001 Message Credibility Source Type Expertise Trustworthiness Age .14 .41

Moderated mediation through expertise, index = .08, 95% CI [-.27, .48] Moderated mediation through trustworthiness, index = .15, 95% CI [-.07, .43]

Discussion

The objective of the current study was to uncover how expert and layperson sources affect source and message credibility, and whether older and younger adults differ in how they evaluate the credibility of expert and layperson sources online. In the context of nutrition information provided on online health forums, this research shows that expert sources have more source credibility than layperson sources and that higher source credibility results in higher message credibility. The study does not show a difference between older and younger adults in how they evaluate the credibility of expert versus layperson sources.

In line with the expectations, this study shows the importance of authority cues for conveying credibility in health nutrition information (Flanagin & Metzger, 2013; Lee & Sundar, 2013; Ramachandran et al., 2018; Sillence & Briggs, 2007). Specifically, adding the label ‘general practitioner’, which was used to indicate an expert source, resulted in different credibility evaluations of the same information. This means that participants must have relied on this authority cue to judge the source and message credibility. From the perspective of ELM, this indicates that internet users adopt a peripheral processing style, thereby relying on heuristics for assessing the credibility of online sources and health information (Wilson, 2007).

The current study also shows that source credibility mediates the effect of expert versus layperson sources on message credibility (Hu & Sundar, 2009; Saleh, 2016). This is in line with prior studies, which showed that source credibility is important for determining message credibility when information is unclear (Go et al., 2014; Saleh, 2016). However, the relationships between credibility concepts are highly dependent on the context in which they are analyzed. In this regard, Saleh (2016) explains that source and message credibility are inherently related and that characteristics in one credibility dimension (i.e., the message) may influence credibility evaluations in another dimension (Hocevar, Metzger & Flanagin, 2017).

Therefore, the findings should be interpreted in terms of correlations between the concepts rather than proof for a causal relationship.

Contrary to expectations, older and younger adults do not differ in how they assess the source credibility of expert and layperson sources. This suggests that, while older adults grew up to be passive recipients of health information (Brody, 1980; Eliassen, 2015; Makoul, 1998; Xie, 2009) and younger adults are more autonomous in their search for health information (Hu & Sundar, 2009; Xie et al., 2012), both age groups rely on the authority cue ‘general practitioner’ to the same extent. An explanation is that the internet reduces the socializing effects of the doctor-patient relationship that older adults grew up with. In this sense, Broom (2005) showed that the internet as an information source liberates people from having to rely on health professionals completely and reduces the inhibitions people could experience in face-to-face interactions with health professionals. This suggests that the social patterns that characterized the traditional doctor-patient relationship do not influence how older adults evaluate online expert sources.

This finding may also be due to the relatively young sample of older adults that participated in this study (M = 58.51, SD = 5.87, range 50 – 79). The current study included older adults that were born before 1970, because the traditional doctor-patient relationship started to change from the 1970s onwards (Xie, 2009). However, other studies highlight that baby-boomers, those people born between 1945 and 1965, exhibit the passive response

patterns (Eliassen, 2015). This would suggest that at least part of the group of older adults that participated in this study grew up when the doctor-patient relationship was already changing, which would make the sample used potentially too young for exhibiting the expected

Limitations and future research

The results of the current study should be interpreted in light of its limitations. Expert and layperson sources were manipulated in terms of the presence and absence of an authority cue, namely the label ‘general practitioner’. However, in real health forums these source types vary in many more ways, such as the language use and the information that these sources provide. Expert sources generally are more knowledge-oriented and prescriptive (Hatzler & Pratt, 2011) whereas layperson sources are more personal, containing a subjective narrative focused on providing affective support (Koepfler & Fleischmann, 2011; Lederman et al., 2013; Sanders et al., 2020). To have a more ecologically valid research, future studies should therefore consider taking these elements into account. A better operationalization of, for example the experiential expertise that characterizes layperson sources (Hu & Sundar, 2009), could reduce the gap between the credibility effects of expert and layperson sources.

Secondly, this study is limited to only one health topic, namely nutrition, and one authority cue, namely ‘general practitioner’. The cue ‘general practitioner’ was decisive for the evaluation of credibility, as this was the only element that differed between the conditions, indicating peripheral processing. However, the effect of this cue on credibility could be weaker when information is centrally processed. According to Hu and Sundar (2009), when a health message is not relevant to people, they rely more on peripheral cues for judging the credibility. Conversely, when a topic is relevant, people are more motivated to think about the credibility of the source based on message contents (Hu & Sundar, 2009), which could reduce the effect of authority cues on credibility. Future studies should therefore analyze what the impact of the label ‘general practitioner’ is on credibility in different contexts.

Lastly, the participants were recruited by means of social media, which makes the older participants more digitally skilled than the majority of Dutch older adults is. According to Statistics Netherlands, between 18% and 48% of older adults have basic digital skills

(2020). The current sample likely does not represent those older adults with weaker digital skills, which limits the external validity of the findings.

Practical recommendations

The current study shows that people rely on traditional markers of credibility (namely the cue ‘general practitioner’) to assess online information and sources and older and younger adults do so to the same extent. Based on these findings, communication professionals are advised to incorporate authority cues in the communication of health information, and particularly when it concerns a health topic about which there is a lot of confusion. This would increase the chance that people of different age groups will consider the message to be credible.

These authority cues would assist people for making credibility evaluations of health information, which may benefit health behavior (Fan et al., 2014). On the other hand, these cues can be unjustly attributed to, for example, non-expert sources that communicate incorrect information. When people rely on these cues to the extent that they do not critically reflect on the message arguments, this has the potential to result in adverse health effects. Authority cues should thus be employed with caution and internet users are advised to reflect on the credibility of health messages beyond these cues. It has yet to be seen what the influence is of authority cues on informed decision making in the domain of health.

References

Appelman, A., & Sundar, S. S. (2015). Measuring Message Credibility. Journalism & Mass Communication Quarterly, 93(1), 59–79. https://doi.org/10.1177/1077699015606057 Bol, N., van Weert, J. C. M., Loos, E. F., Romano Bergstrom, J. C., Bolle, S., & Smets, E. M.

Predict Recall of Information in Younger and Older Adults. Journal of Health Communication, 21(4), 387–396. https://doi.org/10.1080/10810730.2015.1080327

Borah, P., & Xiao, X. (2018). The Importance of ‘Likes’: The Interplay of Message Framing, Source, and Social Endorsement on Credibility Perceptions of Health Information on Facebook. Journal of Health Communication, 23(4), 399–411.

https://doi.org/10.1080/10810730.2018.1455770

Brody, D. S. (1980). The Patient’s Role in Clinical Decision-Making. Annals of Internal Medicine, 93(5), 718. https://doi.org/10.7326/0003-4819-93-5-718

Broom, A. (2005). Virtually He@lthy: The Impact of internet Use on Disease Experience and the Doctor-Patient Relationship. Qualitative Health Research, 15(3), 325–345.

https://doi.org/10.1177/1049732304272916

Carstensen, L. L., Isaacowitz, D. M., & Charles, S. T. (1999). Taking time seriously: A theory of socioemotional selectivity. American Psychologist, 54(3), 165–181.

doi:10.1037/0003-066X.54.3.165

Chung, J. E., Park, N., Wang, H., Fulk, J., & McLaughlin, M. (2010). Age differences in perceptions of online community participation among non-users: An extension of the Technology Acceptance Model. Computers in Human Behavior, 26(6), 1674–1684.

https://doi.org/10.1016/j.chb.2010.06.016

Clark, D., Nagler, R. H., & Niederdeppe, J. (2019). Confusion and nutritional backlash from news media exposure to contradictory information about carbohydrates and dietary fats. Public Health Nutrition, 22(18), 3336–3348.

https://doi.org/10.1017/s1368980019002866

Colditz, J. B., Woods, M. S., & Primack, B. A. (2018). Adolescents Seeking Online Health Information: Topics, Approaches, and Challenges. Technology and Adolescent Mental Health, 21–35. https://doi.org/10.1007/978-3-319-69638-6_2

Dong, Z. (2015). How to persuade adolescents to use nutrition labels: effects of health consciousness, argument quality, and source credibility. Asian Journal of

Communication, 25(1), 84–101. https://doi.org/10.1080/01292986.2014.989241 Dutta-Bergman, M. J. (2004). The Impact of Completeness and Web Use Motivation on the

Credibility of e-Health Information. Journal of Communication, 54(2), 253–269.

https://doi.org/10.1111/j.1460-2466.2004.tb02627.x

Eastin, M. S. (2006). Credibility Assessments of Online Health Information: The Effects of Source Expertise and Knowledge of Content. Journal of Computer-Mediated Communication, 6(4), 0. https://doi.org/10.1111/j.1083-6101.2001.tb00126.x Eliassen, A. H. (2015). Power Relations and Health Care Communication in Older

Adulthood: Educating Recipients and Providers. The Gerontologist, 56(6), 990–996.

https://doi.org/10.1093/geront/gnv095

Fan, H., Lederman, R., Smith, S. P., & Chang, S. (2014). How Trust Is Formed in Online Health Communities: A Process Perspective. Communications of the Association for Information Systems, 34. https://doi.org/10.17705/1cais.03428

Flanagin, A. J., & Metzger, M. J. (2013). Trusting expert- versus user-generated ratings online: The role of information volume, valence, and consumer

characteristics. Computers in Human Behavior, 29(4), 1626–1634.

https://doi.org/10.1016/j.chb.2013.02.001

Go, E., Jung, E. H., & Wu, M. (2014). The effects of source cues on online news perception. Computers in Human Behavior, 38, 358–367.

https://doi.org/10.1016/j.chb.2014.05.044

Gray, N. J., Klein, J. D., Noyce, P. R., Sesselberg, T. S., & Cantrill, J. A. (2005). Health information-seeking behaviour in adolescence: the place of the internet. Social

Science & Medicine, 60(7), 1467–1478.

https://doi.org/10.1016/j.socscimed.2004.08.010

Greiner, A., Clegg Smith, K., & Guallar, E. (2010). Something fishy? News media

presentation of complex health issues related to fish consumption guidelines. Public Health Nutrition, 13(11), 1786–1794. https://doi.org/10.1017/s1368980010000923 Hartzler, A., & Pratt, W. (2011). Managing the Personal Side of Health: How Patient

Expertise Differs from the Expertise of Clinicians. Journal of Medical internet Research, 13(3), e62. https://doi.org/10.2196/jmir.1728

Hart, A., Henwood, F., & Wyatt, S. (2004). The Role of the internet in Patient-Practitioner Relationships: Findings from a Qualitative Research Study. Journal of Medical internet Research, 6(3), e36. https://doi.org/10.2196/jmir.6.3.e36

Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation, moderation, and conditional process modeling [White paper]. Retrieved from http://www.afhayes.com/ public/process2012.pdf

Heuberger, R. A., & Ivanitskaya, L. (2011). Preferred Sources of Nutrition Information: Contrasts Between Younger and Older Adults. Journal of Intergenerational Relationships, 9(2), 176–190. https://doi.org/10.1080/15350770.2011.567919 Hilligoss, B., & Rieh, S. Y. (2008). Developing a unifying framework of credibility

assessment: Construct, heuristics, and interaction in context. Information Processing & Management, 44(4), 1467–1484. https://doi.org/10.1016/j.ipm.2007.10.001

Hirvonen, N., Tirroniemi, A., & Kortelainen, T. (2019). The cognitive authority of user generated health information in an online forum for girls and young women. Journal of Documentation, 75(1), 78–98. https://doi.org/10.1108/jd-05-2018-0083

Trust in Health and Risk Messaging. Oxford Research Encyclopedia of Communication. https://doi.org/10.1093/acrefore/9780190228613.013.287

Ho, J., & Nesbit, P. (2018). Personality and work outcomes: A moderated mediation model of self-leadership and gender. International Journal of Management Excellence, 10(2), 1292-1304.

Huisman, M., Joye, S., & Biltereyst, D. (2019). Searching for Health: Doctor Google and the Shifting Dynamics of the Middle-Aged and Older Adult Patient–Physician

Relationship and Interaction. Journal of Aging and Health, 089826431987380.

https://doi.org/10.1177/0898264319873809

Hu, Y., & Sundar, S. (2009). Effects of Online Health Sources on Credibility and Behavioral Intentions. Communication Research, 37(1), 105–132.

https://doi.org/10.1177/0093650209351512

Jäger, A.-K., & Weber, A. (2020). Can you believe it? The effects of benefit type versus construal level on advertisement credibility and purchase intention for organic food. Journal of Cleaner Production, 257, 120543.

https://doi.org/10.1016/j.jclepro.2020.120543

Jung, E. H., Walsh-Childers, K., & Kim, H.-S. (2016). Factors influencing the perceived credibility of diet-nutrition information web sites. Computers in Human Behavior, 58, 37–47. https://doi.org/10.1016/j.chb.2015.11.044

Kamiński, M., Skonieczna-Żydecka, K., Nowak, J. K., & Stachowska, E. (2020). Global and local diet popularity rankings, their secular trends and seasonal variation in Google Trends data. Nutrition, 110759. https://doi.org/10.1016/j.nut.2020.110759

Kanthawala, S., Vermeesch, A., Given, B., & Huh, J. (2016). Answers to Health Questions: internet Search Results Versus Online Health Community Responses. Journal of Medical internet Research, 18(4), e95. https://doi.org/10.2196/jmir.5369

Khan, Yang, Shafi, & Yang. (2019). Role of Social Media Marketing Activities (SMMAs) in Apparel Brands Customer Response: A Moderated Mediation

Analysis. Sustainability, 11(19), 5167. https://doi.org/10.3390/su11195167

Kim, H.-S., & Sundar, S. S. (2011). Using interface cues in online health community boards to change impressions and encourage user contribution. Proceedings of the 2011 annual conference on Human factors in computing systems - CHI ’11, 599–608.

https://doi.org/10.1145/1978942.1979028

Koepfler, J. A., & Fleischmann, K. R. (2011). Classifying values in informal communication: Adapting the meta-inventory of human values for tweets. Proceedings of the

American Society for Information Science and Technology, 48(1), 1–4.

https://doi.org/10.1002/meet.2011.14504801116

Lederman, R., Fan, H., Smith, S., & Chang, S. (2013). Who can you trust? Credibility assessment in online health forums. Health Policy and Technology, 3(1), 13–25.

https://doi.org/10.1016/j.hlpt.2013.11.003

Lee, J. Y., & Sundar, S. S. (2013). To Tweet or to Retweet? That Is the Question for Health Professionals on Twitter. Health Communication, 28(5), 509–524.

https://doi.org/10.1080/10410236.2012.700391

Lee, Y., Huang, M.-C., Zhang, X., & Xu, W. (2015). FridgeNet: A Nutrition and Social Activity Promotion Platform for Aging Populations. IEEE Intelligent Systems, 30(4), 23–30. https://doi.org/10.1109/mis.2015.55

Linn, A. J., van Weert, J. C. M., Gebeyehu, B. G., Sanders, R., Diviani, N., Smit, E. G., & van Dijk, L. (2018). Patients’ Online Information-Seeking Behavior Throughout

Treatment: The Impact on Medication Beliefs and Medication Adherence. Health Communication, 34(12), 1461–1468. https://doi.org/10.1080/10410236.2018.1500430

information. Proceedings of the 2nd ACM SIGHIT symposium on International health informatics - IHI ’12. https://doi.org/10.1145/2110363.2110404

Ma, T. (Jennifer), & Atkin, D. (2017). User generated content and credibility evaluation of online health information: A meta analytic study. Telematics and Informatics, 34(5), 472–486. https://doi.org/10.1016/j.tele.2016.09.009

Machackova, H., & Smahel, D. (2018). The perceived importance of credibility cues for the assessment of the trustworthiness of online information by visitors of health-related websites: The role of individual factors. Telematics and Informatics, 35(5), 1534 -1541. https://doi.org/10.1016/j.tele.2018.03.021

Makoul, G. (1998). Perpetuating Passivity: Reliance and Reciprocal Determinism in Physician-Patient Interaction. Journal of Health Communication, 3(3), 233–259.

https://doi.org/10.1080/108107398127355

McGinnies, E., & Ward, C. D. (1980). Better Liked than Right. Personality and Social Psychology Bulletin, 6(3), 467–472. https://doi.org/10.1177/014616728063023 Mendes, Á., Abreu, L., Vilar-Correia, M. R., & Borlido-Santos, J. (2016). “That Should Be

Left to Doctors, That’s What They are There For!”—Exploring the Reflexivity and Trust of Young Adults When Seeking Health Information. Health

Communication, 32(9), 1076–1081. https://doi.org/10.1080/10410236.2016.1199081

Metzger, M. J., Flanagin, A. J., & Medders, R. B. (2010). Social and Heuristic Approaches to Credibility Evaluation Online. Journal of Communication, 60(3), 413–439.

https://doi.org/10.1111/j.1460-2466.2010.01488.x

Minich, D. M., & Bland, J. S. (2013). Personalized Lifestyle Medicine: Relevance for Nutrition and Lifestyle Recommendations. The Scientific World Journal, 2013, 1–14.

https://doi.org/10.1155/2013/129841

information environment (AAI3429172). Retrieved from

https://repository.upenn.edu/dissertations/AAI3429172/

National Institute for Public Health and the Environment. (2011). Dutch National Food Consumption Survey 2007-2010. Diet of children and adults aged 7 to 69 years. Retrieved from https://www.rivm.nl/bibliotheek/rapporten/350050006.pdf

O’Connor, A. (2009). Always Follow the Elephants. New York, New York: Henry Holt and Company.

Ohanian, R. (1990). Construction and Validation of a Scale to Measure Celebrity Endorsers’ Perceived Expertise, Trustworthiness, and Attractiveness. Journal of

Advertising, 19(3), 39–52. https://doi.org/10.1080/00913367.1990.10673191

O’Keefe, D., 2002. Persuasion: Theory and Research. Sage Publications, Inc.. Retrieved from

http://books.google.com/books?hl=enlr=id=e3V6Zen0UGwCoi=

fndpg=PR13dq=o%27keefe+persuasion+theory+and+research+ots=ieGtwk 2f5sig=cQ81JpI2fly9_OjJVQzjOBAYNTc.

Petracci, M., Schwarz, P. K. N., Sánchez Antelo, V. I. Ma., & Mendes Diz, A. M. (2017). Doctor–patient relationships amid changes in contemporary society: a view from the health communication field. Health Sociology Review, 26(3), 266–279.

https://doi.org/10.1080/14461242.2017.1373031

Ramachandran, D., Kite, J., Vassallo, A. J., Chau, J. Y., Partridge, S. R., Freeman, B., & Gill, T. (2018). Food Trends and Popular Nutrition Advice Online – Implications for Public Health. Online Journal of Public Health Informatics, 10(2).

https://doi.org/10.5210/ojphi.v10i2.9306

Saleh, H. F. (2016). Developing New Media Credibility Scale: A Multidimensional Perspective. International Scholarly and Scientific Research & Innovation, 10(4),

1347–1360. Retrieved from

https://pdfs.semanticscholar.org/f082/9d691f75b1170c67c361d67528f55e5c1c02.pdf Sanders, R., Linn, A. J., Araujo, T. B., Vliegenthart, R., van Eenbergen, M. C., & van Weert,

J. C. M. (2020). Different platforms for different patients’ needs: Automatic content analysis of different online health information platforms. International Journal of Human-Computer Studies, 137, 102386. https://doi.org/10.1016/j.ijhcs.2019.102386 Sanders, R., & Linn, A. J. (2018). A Mixed Method Study Investigating the Impact of Talking

about Patients’ internet Use on Patient-Reported Outcomes. Journal of Health Communication, 23(9), 815–823. https://doi.org/10.1080/10810730.2018.1514443 Sbaffi, L., & Rowley, J. (2017). Trust and Credibility in Web-Based Health Information: A

Review and Agenda for Future Research. Journal of Medical internet Research, 19(6), e218. https://doi.org/10.2196/jmir.7579

Shan, Y. (2016). How credible are online product reviews? The effects of self-generated and system-generated cues on source credibility evaluation. Computers in Human

Behavior, 55, 633–641. https://doi.org/10.1016/j.chb.2015.10.013

Sillence, E., & Briggs, P. (2007). Please advise: using the internet for health and financial advice. Computers in Human Behavior, 23(1), 727–748.

https://doi.org/10.1016/j.chb.2004.11.006

Simpson, S. (2004). Explaining the Commercialization of the internet. Information, Communication & Society, 7(1), 50–68.

https://doi.org/10.1080/1369118042000208898

Sundar, S., Knobloch-Westerwick, S., & Hastall, M. (2007). News cues: Do indicators of newsworthiness by newsbots affect our perception of news stories? A cross-cultural study in Germany, the Netherlands, and the U.S. JASIST Journal of the American

Society of Information Science and Technology, 58, 366–378. Retrieved from

https://ucom.osu.edu/for-media/faculty-experts/a-m/knobloch-westerwick-silvia.html Statistics Netherlands. (2020, February 14). Nederlanders in Europese kopgroep digitale

vaardigheden. Retrieved from

https://www.cbs.nl/nl-nl/nieuws/2020/07/nederlanders-in-europese-kopgroep-digitale

-vaardigheden

Statistics Netherlands (2019, December 15). Internet; toegang, gebruik en faciliteiten. (2020, June 25). Retrieved from https://www.cbs.nl/nl-nl/cijfers/detail/83429NED?dl=13A7B Thon, F. M., & Jucks, R. (2016). Believing in Expertise: How Authors’ Credentials and

Language Use Influence the Credibility of Online Health Information. Health Communication, 32(7), 828–836. https://doi.org/10.1080/10410236.2016.1172296 Tseng, S., & Fogg, B. J. (1999). Credibility and computing technology. Communications of

the ACM, 42(5), 39–44. https://doi.org/10.1145/301353.301402 PUT IN TEXT van der Goot, M. J., Bol, N., & van Weert, J. C. M. (2019). Translating socioemotional

selectivity theory Into persuasive communication: Conceptualizing and operationalizing emotionally-meaningful versus knowledgerelated appeals. International Journal of Communication : IJoC, 13, 1416-1437.

Vennik, F. D., Adams, S. A., Faber, M. J., & Putters, K. (2014). Expert and experiential knowledge in the same place: Patients’ experiences with online communities connecting patients and health professionals. Patient Education and

Counseling, 95(2), 265–270. https://doi.org/10.1016/j.pec.2014.02.003 Voedingscentrum. (2020). Homepage. Retrieved from

https://www.voedingscentrum.nl/nl.aspx