This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Nutrition science is currently facing issues regarding the public’s perception of its credibility, with social media (SM) influencers increasingly becoming a key source for nutrition-related information with high engagement rates. Source credibility and, to an extent, authenticity have been widely studied in marketing and communications but have not yet been considered in the context of nutrition or health communication. Thus, an investigation into the factors that impact perceived source and message credibility and authenticity is of interest to inform health communication on SM.
This study aims to explore the factors that impact message and source credibility (which includes trustworthiness and expertise) or authenticity judgments on SM platforms to better inform nutrition science SM communication best practices.
A total of 6 databases across a variety of disciplines were searched in March 2019. The inclusion criteria were experimental studies, studies focusing on microblogs, studies focusing on healthy adult populations, and studies focusing on either source credibility or authenticity. Exclusion criteria were studies involving participants aged under 18 years and clinical populations, gray literature, blogs, WeChat conversations, web-based reviews, non-English papers, and studies not involving participants’ perceptions.
Overall, 22 eligible papers were included, giving a total of 25 research studies. Among these studies, Facebook and Twitter were the most common SM platforms investigated. The most effective communication style differed depending on the SM platform. Factors reported to impact credibility included language used online, expertise heuristics, and bandwagon heuristics. No papers were found that assessed authenticity.
Credibility and authenticity are important concepts studied extensively in the marketing and communications disciplines; however, further research is required in a health context. Instagram is a less-researched platform in comparison with Facebook and Twitter.
Science, particularly the discipline of nutrition science, is currently facing credibility issues in the eyes of the general public [
The advent of the internet and, in particular, social media (SM; see
Currently, the use of SM in health interventions has had limited effectiveness, with participant engagement rates (
Similar to celebrity endorsers, SM influencers (SMIs) or “individuals or groups of individuals who can shape attitudes and behaviours through online channels,” are arguably human brands (
In the existing posttruth era (
In the marketing literature, underlying factors such as the endorser’s expertise, trustworthiness, attractiveness, and authenticity have been shown to influence people’s behavior in traditional forms of advertising (eg, television commercials, celebrity partnerships) [
Authenticity is the concept of “being true to the self in terms of an individual’s thoughts, feelings, and behaviours reflecting their true identity” [
The source credibility model suggests that a credibility judgment is determined based on the source’s (eg, a celebrity’s) expertise, trustworthiness, and attractiveness [
Message content is assessed through cognitive processing (
Previous reviews have assessed the credibility of health information online, finding that factors such as clear website layout and professional design increased credibility [
As SM is a relatively new area of research, a scoping review was considered the most appropriate method to explore the topic area. Our research in applying social marketing techniques to the field of nutrition has led us to recognize the importance of using marketing and communication techniques, particularly on SM [
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses Scoping Review Checklist and the Joanna Briggs Institute Reviewer’s Manual were used throughout the review process [
Key databases from health, psychology, and business disciplines were used to conduct the final search in conjunction with a university librarian on March 27, 2019. Cumulative Index of Nursing and Allied Health Literature (CINAHL) Plus (422 results), Scopus (2199 results), Excerpta Medica dataBASE (EMBASE; 697 results), Ovid Medical Literature Analysis and Retrieval System Online (MEDLINE; 326 results), PsycINFO (1223 results), and Business Source Complete (1375 results) were searched for title, abstract, and keywords to identify the initial 6242 articles (example search strategy provided in
The inclusion criteria were experimental studies (ie, stimuli provided), studies focusing on microblogs (eg, Facebook, Twitter;
Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews flow diagram of the search and study selection process.
Overall, 2 investigators (ELJ and AMB) independently screened the title and abstract of the included papers against the inclusion and exclusion criteria. This process was repeated for full-text screening, with any conflicts being discussed until a joint consensus was reached. There were 22 final papers (
Data extraction was conducted independently by a researcher (ELJ) using Microsoft Excel 2019 (Microsoft Corporation) and was then cross-checked by a fellow researcher (AMB). Following data extraction, the results were collated based on various parameters such as SM platform focused on manipulation stimuli, outcome, and scales used for data collection. All papers were given a category to summarize their topic area, such as
Of the 22 papers that met the inclusion criteria, 3 papers reported on 2 separate studies, giving a total of 25 research studies [
A total of 4 microblogging platforms were used: Twitter (n=10), Facebook (n=8), Instagram (n=3), and YouTube (n=3;
All papers used a scale to assess credibility, with the most common being McCroskey and Teven’s Source Credibility Scale (n=6; α=.94) and Ohanian’s Source Credibility Scale (n=4; α>.8) [
Most papers (n=15) involved the manipulation of text (eg, in a SM post) via Facebook or Twitter feeds to assess credibility (n=11), trust (n=4), or both (n=2). Studies primarily explored the tone of voice, bandwagon cues, and expertise [
Overall, 3 papers included in the review assessed message credibility: how message characteristics impact credibility perceptions (
On Twitter, Houston et al [
Similarly, Yilmaz and Johnston [
Main findings of included papers and their effect on credibility or trust, separated by manipulated variable: number of likes, number of followers, number of retweets, source, and language.
Outcomes/author (year) [citation] | Population group | Key significant results | |
|
|||
|
Borah and Xiao (2018) [ |
Students | In the 2 studies conducted, the number of likes did not affect source credibility overall when looking at Facebook posts (study 1: |
|
Phua and Ahn (2016) [ |
Students | Brand trust was higher when likes were high on Facebook post ( |
|
Shen et al (2019) [ |
Paid online workers | Bandwagon cues did not impact credibility when looking at images on Twitter and Facebook ( |
|
|||
|
Jin and Phua (2014) [ |
Students | A higher number of Twitter followers on the celebrity’s account increased source credibility and intention to build an online friendship with the celebrity endorser for all dimensions of source credibility: physical attraction ( |
|
Lee (2018) [ |
Students | The number of followers on Facebook made a statistically significant difference on the believability of the answer ( |
|
Westerman et al (2011) [ |
Students | Trustworthiness indicated an inverted U-shaped relationship with the number of followers on Twitter ( |
|
|||
|
Lin and Spence (2018) [ |
Students | The highest level of trust (on Twitter) was when participants viewed the post with 400 retweets, followed by 40 retweets, whereas 4000 retweets had the lowest level of trust ( |
|
Lin and Spence (2019) [ |
Students | There were significant differences in trust perceptions across varying retweet conditions ( |
|
Lin and Spence (2016) [ |
Students | Participants perceived lowest competence when viewing a peer’s Twitter page with no retweets ( |
|
|||
|
Borah and Xiao (2018) [ |
Students | In the 2 studies conducted on Facebook, the CDC and WebMD authors were seen as more credible than unknown authors (study 1: |
|
Lin and Spence (2018) [ |
Students | Participants viewing an FDA expert’s Twitter account were more likely to perceive higher trust ( |
|
Lin and Spence (2016) [ |
Students | Higher credibility was assigned to risk information from an expert compared with a peer and a stranger on Twitter ( |
|
|||
|
Borah and Xiao (2018) [ |
Students | In the 2 studies conducted, a gain-framed message was more credible than a loss-framed message on Facebook (study 1: |
|
Houston et al (2018) [ |
Paid workers | Nonopinionated tweets were perceived as more credible than opinionated tweets ( |
|
Yilmaz and Johnson (2016) [ |
Students | Personalized status updates on Facebook were seen as more competent and trustworthy than personalized tweets ( |
aFDA: Food and Drug Administration.
bCDC: Centers for Disease Control and Prevention.
Included papers, main outcomes, and the effect on either brand trust, message credibility, or source credibility (including trustworthiness, believability, and competence) as specified in their results.
Factors | Platform | Population | Outcome | Resulta | Relevant papers | |
|
||||||
|
|
|||||
|
|
Student | Message credibilityb,c | Increase | Borah and Xiao [ |
|
|
|
|||||
|
|
Student | Competence and trustworthiness | Decrease | Yilmaz and Johnson [ |
|
|
|
Student | Competence and trustworthiness | Increase | Yilmaz and Johnson [ |
|
|
|
|||||
|
|
Student | Trustworthiness | Increase | Antoci et al [ |
|
|
|
|||||
|
|
Paid worker | Message credibility | Increase | Houston et al [ |
|
|
|
YouTube | Student | Message credibility | No effect | Zimmermann and Jucks [ |
|
||||||
|
|
|||||
|
|
Student | Source credibility and trustworthiness | No effect | Borah and Xiao [ |
|
|
|
|
Student | Brand trust | Increase | Phua and Ahn [ |
|
|
|||||
|
|
Student | Believability | Increase | Lee [ |
|
|
|
Student | Source credibility | Westerman and Spence: unclear; Phua and Ahn: increase | Westerman and Spence [ |
|
|
|
|||||
|
|
Student | Competency | Increase | Westerman and Spence [ |
|
|
|
|||||
|
|
Student | Trustworthiness | Decrease | Lin and Spence [ |
|
|
|
|||||
|
|
Student | Believability and trustworthiness | Increase | Lee [ |
|
|
||||||
|
Post from expert source | Facebook and Twitter | Student | Source credibility | Increase | Borah and Xiao [ |
|
||||||
|
Interaction with followers | Student | Source credibility | Increase | Jahng and Littau [ |
|
|
High perceived privacy control | Student | Trust | Increase | Antoci et al [ |
|
|
Positive brand attitude | Paid worker | Brand credibility | Increase | De Veirman and Hudders [ |
|
|
Prosocial attitude online | Student | Source credibility | Increase | Jin and Phua [ |
|
|
Recency of updates (frequent) | Student | Source credibility | Increase | Westerman and Spence [ |
|
|
Snapshot aesthetic (vs studio aesthetic) | Paid worker | Brand credibility | Increase | Colliander and Marder [ |
|
|
Preexisting photoshop/internet skills (when looking at photoshopped images) | Twitter and Facebook | Paid worker | Source credibility | Decrease | Shen et al [ |
|
Ethos message appeal (compared with logos and pathos) | YouTube | Student | Source credibility | Increase | English et al [ |
|
Consumer-generated advertising (compared with firm-generated advertising) | YouTube | Student | Source credibility | Increase | Lee et al [ |
|
Caucasian ethnicity (compared with African American) | Student | Source credibility | Increase | Spence et al [ |
aOn the basis of reported results from studies summarized in
bFor further context, explanation, and examples of these factors, refer to
cCredibility comprises trustworthiness, expertise, and sometimes attractiveness, depending on the individual paper.
Bandwagon cues such as the number of followers, number of retweets, and number of likes were a way in which student participants (mean age range 19-22.9 years) assessed source credibility across 30% (8/22) of the included papers; however, the findings were inconsistent among studies [
Borah and Xiao [
On Twitter, Westerman et al [
Manipulating the source of the Facebook or Twitter posts was found to impact source credibility in 3 studies with student participants [
Similarly, expert sources (eg, FDA) were considered more trustworthy and competent and, thus, overall more credible than strangers or peers on Twitter (
Surprisingly, there were no studies in the field of health and nutrition research included in this scoping review; however, there are some important learnings that could be utilized for nutrition and health communication. There were many different factors that affected the perceived credibility of a message and source on SM, such as language usage, expertise heuristics, and bandwagon heuristics. However, no information was found on the factors affecting perceived authenticity in this context. The scales used as well as the different models and theories to inform various fields of research are reported.
The results of the included studies indicated that the language used in tweets affected message credibility. Personalization was more effective on Facebook, whereas depersonalization was more effective on Twitter [
Each study in the review that compared an expert to a stranger or peer found that the expert was more credible [
A key focus of the included papers was bandwagon heuristics, which relate to the number of likes, followers, or retweets assigned to information on SM [
The most commonly used scale in the research papers was McCroskey and Teven’s Source Credibility Scale (n=6), used to assess the credibility of a source of information (ie, a person). In contrast, message credibility was assessed by examining the characteristics of a message such as the structure, perceived accuracy, and language used [
Communications and psychology were the predominant fields of research within the included papers. The theories and models used to underpin research overlapped among disciplines, with the source credibility model being the most common within communications, information research, psychology, and business. The MAIN model (a digital extension of source credibility model) was also used frequently in communications, information research, computer science, and psychology. Previously, source credibility and its impact have been investigated more broadly in the health discipline to assess consumers’ perceptions of health information online, but this was limited to websites and did not include SM [
On the basis of the search strategy and the inclusion and exclusion criteria, the concept of authenticity has not been explored in this specific context. In addition, there was a lack of research regarding SMIs or nutrition professionals as the source of SM posts. Most studies focused on organizations or an unknown fictional author as the source. As SMIs are key to digital marketing and health communication, research into consumer perceptions of the source credibility and authenticity of SMIs would be beneficial to further understand how they communicate effectively with their followers. Results from these future studies would be beneficial for informing the delivery of health communication in various digital formats.
The strength of the papers included in the scoping review was the large sample size that was achieved (range of the number of participants from included studies was 85-3476). As most papers had a student cohort participating for university credit, large numbers of participants were recruited. In addition, many of the papers completed a manipulation check during the pilot of the survey to ensure that the mock scenario they were creating was robust, for example, testing if the
As most papers (n=21) used convenience sampling, selection biases were inherent. Furthermore, generalizability was limited to the geographical area in which the research was conducted, making it difficult to draw conclusions without conducting further research. Student samples were predominant (n=17), further limiting the variability and generalizability of results as student samples (referred to as western, educated, industrialized, rich, and democratic [WEIRD]) [
Fostering credibility online should be considered by corporate and human health brands to create a stronger relationship with their audience. This scoping review highlighted that message and source credibility can be affected by language usage, expertise heuristics, and bandwagon cues. Gaps in the literature were identified, highlighting the need for further research on SM platforms, as Instagram and YouTube were studied less than Facebook and Twitter. The main field of research identified from the included papers was communications, with no papers from health or nutrition science. Currently, there is a limited understanding of the use of SMIs and science experts to relay health messages. Further research needs to be undertaken to apply information from communications (on source and message credibility and authenticity) in a health context and in populations other than students.
Glossary of terms.
Search strategy (Scopus database).
Research studies assessing trust and credibility on Facebook.
Research studies assessing trust and credibility on Twitter.
Research studies assessing credibility on both Facebook and Twitter.
Research studies assessing credibility on Instagram.
Research studies assessing credibility on YouTube.
Centers for Disease Control and Prevention
elaboration likelihood model
Food and Drug Administration
modality, agency, interactivity, and navigability
self-determination theory
social media
social media influencer
The authors thank Monash University librarian, Anne Young, for assistance when developing the search strategy.
Communicating Health is funded by the National Health and Medical Research Council (grant number GNT1115496).
ELJ and AMB completed the screening and data extraction of the included papers. The manuscript was drafted by ELJ, with critical revisions suggested by TAM and JI. All authors read and approved the final manuscript.
None declared.