Does source matter? Examining the effects of health experts, friends, and social media influencers on young adult perceptions of Instagram e-cigarette education messages

Abstract

Purpose:

Young adults’ e-cigarette use is a leading public health concern. Using messages from credible sources can help improve message acceptance, yet little research has examined the role of source credibility on young adults’ responses to e-cigarette education messages.

Methods:

We examined the impact of source on young adults’ perceptions of e-cigarette education messages and e-cigarettes. In July 2022, we conducted an experimental study using an online sample of young adults (N=459, M_age=24.6) who were randomized to one of three source conditions: expert, friend, or influencer, and viewed e-cigarette education messages. We used one-way ANOVA to estimate the association between the conditions and outcomes (perceived source credibility, message trust, curiosity, use interests, perceived message effectiveness, beliefs, harm perceptions, and intentions to refrain).

Results:

The expert condition was associated with significantly higher perceived source credibility (vs. friend, influencer; p<0.001), message trust (vs. friend, influencer; p<0.001), and curiosity (vs. influencer; p’s<0.05).

Conclusions:

Public health campaigns may leverage health experts to deliver e-cigarette education messages targeting young adults to improve effectiveness of the messages.

Introduction

In the United States (U.S.), more young adults use e-cigarettes than adults in older age groups.¹ E-cigarette use can cause cognitive and affective disorders,² lead to nicotine addiction, and initiation of cigarette smoking among young adults.^3,4 Several public health education campaigns in the U.S. have aimed to educate young adults about the health harms of using e-cigarettes to prevent, reduce, and stop them from use (e.g., The Real Cost campaign,⁵ This is Quitting⁶). However, social media has proliferated health misinformation related to tobacco products,⁷ making it especially challenging to communicate to young adults about the health harm of e-cigarette use. There needs to be a strategic effort to improve the effectiveness of e-cigarette education messaging on social media among young adults.

Source credibility is an important factor that can influence persuasiveness of tobacco education messages.⁸ Perceived expertise and trustworthiness of a source are key factors influencing individuals’ decisions to accept or reject messages.^9–11 While an expert source with credentials are perceived as trustworthy in their claims,¹² a source with shared psychosocial characteristics (e.g., status, values, interest) with the audience is perceived as trustworthy.^10,11 Thus, an expert and a peer source may differently influence young adult perceptions of e-cigarette education messages.

Current e-cigarette education campaigns employ health experts and young people in campaign advertisements. Since the 2016 report on the health harms of e-cigarette use in young people, the U.S. Surgeon General has served as the primary expert voice against e-cigarette use in young people.¹³ Public health education campaigns targeting young people include young actors with physical health harms from e-cigarette use (e.g., The Real Cost e-cigarette prevention campaign,⁵ and young adults who have quit or are trying to quit e-cigarettes (e.g., This is Quitting e-cigarette cessation campaign.⁶ Although these campaign messages are generally perceived as effective,^14,15 these campaigns have not tested for the effects of source credibility on perceived message effectiveness. The effects of source on e-cigarette risk perceptions^8,16 and influence of social media on e-cigarette use in young people¹⁷ raise the need to examine the association of source on e-cigarette social media education messages.

Tobacco use status influences the perceptions of the source of tobacco health information, such that people who ever used e-cigarettes are less likely to trust health experts (i.e., government health agencies and health organizations) than those who never use e-cigarettes.¹⁸ Similarly, a study found that individuals who smoked “heavily” were more likely to trust medical researchers, and those who smoked “lightly” were more likely to trust a government health organization.¹⁹ An expert can be an effective message source for an audience who lack strong opinions on the issue, however, using an expert source in a message that conflicts with the audience’s pre-existing beliefs may cause reactance among those who hold a strong opinion on the issue.²⁰ Conversely, a source with shared demographic characteristics or values can reduce defensiveness among those with pre-existing beliefs (i.e., people who currently smoke cigarettes) and increase positive attitudes toward counter attitudinal messages (i.e., anti-smoking campaign messages).^21,22 Considering the impact of e-cigarette use status on education message responses can help researchers and health communicators alike to better understand trusted sources of e-cigarette health information for young adults.

The main goal of the study was to examine the impact of source on young adults’ perceptions of social media e-cigarette education messages. Specifically, we examined the differences in three types of sources that share or endorse health information: health experts (expert), friends (friend), and social media influencers (influencer) on perceptions of the source, message, and e-cigarettes. Finally, we examined the impact of e-cigarette use status on perceptions of the source, message, and e-cigarettes. We hypothesized that young adults would report more positive perceptions of the source,²¹ message,^15,22,23 and negative e-cigarette perceptions^15,23 from a peer source, especially influencers.²⁴ We additionally hypothesized that young adults who currently use e-cigarettes^18,25 would report more negative perceptions of the source and message and positive perceptions of e-cigarettes.

Methods

Participants

In July 2022, we recruited a sample of young adults (N=510) from Prolific (prolific.co), an online crowdsourcing survey platform for behavioral research. Participants were eligible if they were between 18 and 30 years old and lived in the U.S. We used Prolific’s prescreening measures to recruit a balanced sample of participants who use e-cigarettes (n=250) and do not use e-cigarettes (n=250). We verified participants’ e-cigarette use status with self-report responses.

Procedures

Potential participants reviewed a brief description about the study on Prolific before being directed to the Qualtrics survey platform. After providing consent, participants were asked to complete questions about their perceptions and use of e-cigarettes and other tobacco products. Then, they were randomized to one of three source conditions: expert, friend, or influencer. After random assignment, participants saw a series of Instagram-simulated posts about the health harms of using e-cigarettes and answered questions about their perceptions of the source, message, e-cigarettes, and demographics. All participants in each manipulated source condition saw the same messages. After completion, participants were thanked and compensated $3.50 via Prolific policies. All procedures were approved by the host institution’s Institutional Review Board (STUDY00000141).

Stimuli

Source Manipulation.

Informed by the source credibility theory^9–11and the e-cigarette literature,^17,26,27 we have developed three experimental conditions for the source (expert, friend, influencers). We manipulated the source by providing a text description of the source and created the Instagram account profile pictures and names to match each source condition. We additionally manipulated the number of “likes” for each source condition to match the source characteristics. Participants in the expert condition first saw a description: “Imagine that a health expert posted these messages” and saw Instagram posts of e-cigarette education messages from a health expert account that had a profile picture of a medical school’s logo and name (e.g., “Medical School”). Participants in the friend condition first saw a description: “Imagine that a friend posted these messages” and saw the same Instagram posts as the other conditions except from an account that had a profile picture of young people and a person’s name (e.g., “Alex_S”). Similarly, participants in the influencer condition saw a description: “Imagine that an Instagram influencer posted these messages” and saw the same Instagram posts as the other conditions except from an account that had a profile picture of young people and a name (e.g., Influencer: “Inspired by Sam”). For both the friend and influencer conditions, we purposefully used a stock photo consisting of both a man and woman and a non-binary name.

Instagram Posts.

We used images from the FDA’s The Real Cost campaign website (https://therealcost.betobaccofree.hhs.gov/vapes).⁵ In total, there were five Instagram posts consisting of an image and a message about the health harms and addictiveness of e-cigarettes. For example, one post stated, “Nicotine changes the way your brain works.” The image format was adapted to simulate Instagram posts and the Instagram user experience, such that participants had to scroll to view posts. All posts included the same date and other post details for all participants. Posts were identical and only manipulated to convey the condition (e.g., message source). All participants in all three source conditions saw the same five posts which were presented in a random order.

Measures

Demographics.

Participants were asked to report their age (between 18 and 30 years old), race and ethnicity (Asian-Eastern, Asian-Indian, Black/African American, Native American or Alaskan Native, Pacific Islander, White/Caucasian, Hispanic/Latino), and gender (female, male, trans female/ trans woman, trans male/trans man, genderqueer/gender non-conforming/gender expansive).

Message exposure.

Participants were asked to report if they had ever seen the images used in the experiment before with answer options including yes, no, or not sure.

E-cigarette use.

Participants were asked to report their e-cigarette use status. Participants were categorized as: currently use e-cigarettes if they used an e-cigarette in the past 30 days; ever used e-cigarettes if they had ever used an e-cigarette but not in the past 30 days; and never used e-cigarettes if they had never tried using an e-cigarette, even one or two times.²⁸

Current use of other tobacco.

Participants were asked to report their current use of other tobacco products. They indicated whether they had used cigarettes, cigars, smokeless tobacco, and hookah, respectively, in the past 30 days (yes or no).²⁸

Perceived source credibility.

Perceived source credibility was measured by adapting four questions from the source credibility scale.²⁹ We asked the extent to which participants believed the source of the messages to be fair, accurate, told the whole story, and could be trusted on a scale of 1 (not at all) to 7 (very much so) (Cronbach’s α=0.94).

Message trust.

Message trust was measured by using a single item question³⁰ asking how much the participants trusted the messages on a scale from 1 (Not at all) to 10 (Very much).

Curiosity.

Curiosity was measured by using a single item question³⁰ asking the extent to which the messages made the participants curious about the product on a scale from 1 (Strongly disagree) to 7 (Strongly agree).

E-cigarette use interest.

Use interest was measured by using a single item question³⁰ asking the extent to which the messages made the participants want to use the product on a scale of 1 (Strongly disagree) to 7 (Strongly agree).

Perceived message effectiveness.

Perceived message effectiveness was measured by using six questions from the perceived message effectiveness scale.³¹ We asked whether the participants thought the message was powerful, informative, meaningful, convincing, worth remembering, grabbed attention on a scale from 1 (Strongly disagree) to 5 (Strongly agree) (Cronbach’s α=0.93).

Beliefs.

Beliefs were measured by using four questions from the effects perception scale.³² We asked whether the participants thought the messages made them understand the cost of vaping to their health, gave them good reasons not to vape, made vaping seem like a bad idea to them, and made vaping seem gross to them on a scale of 1 (Strongly disagree) to 5 (Strongly agree) (Cronbach’s α=0.92).

Harm perceptions.

Harm perceptions were measured by using a single item question³³ asking the extent to which e-cigarettes were harmful to their health on a scale of 0 (Strongly disagree) to 10 (Strongly agree).

Intentions to refrain.

Intentions to refrain from using e-cigarettes were measured by using two item questions from the effects perception scale³² asking whether the participants thought the messages made them not want to vape, and kept them from wanting to vape on a scale of 1 (Strongly disagree) to 5 (Strongly agree). Items were reverse coded for analysis (Cronbach’s α=0.95).

Statistical analysis

We used SPSS Statistics v.29.0.1.0 (IBM, 2023) to conduct analyses. One-way ANOVA was performed to check randomization. Randomization was successful as covariates (age, gender, race, ethnicity, e-cigarette use, and other tobacco use) were not significantly associated with the experimental conditions. Descriptive statistics were used to calculate distributions of e-cigarette use, current use of other tobacco products, perceived source credibility, message trust, curiosity, e-cigarette use interest, perceived message effectiveness, beliefs, harm perceptions, and intentions to refrain to inform the decision on appropriate statistical tests. First, we conducted one-way ANOVA tests to estimate the association between prior message exposure and outcomes, controlling for covariates, and used posthoc tests for pairwise comparisons. When the overall differences were statistically significant, posthoc tests were used for pairwise comparisons. Second, we utilized adjusted models, controlling for the covariates of interest, and used posthoc tests when overall differences were statistically significant at alpha=0.05. Lastly, we estimated the moderating effects of e-cigarette use status (never – reference group, ever, and current use) on the association between the conditions (expert – reference group, friend, and influencer) and outcomes.

Results

Participant Characteristics

Of the 510 participants who began our survey, our analytic sample included the participants (N=459) who were successfully randomized into one of the three experimental conditions and completed the survey. In the study, 20.5% of the participants reported “Yes” to previously seeing the messages, 10.9% reported “Not Sure,” and 68.6% reported “No.” The average age of participants was 24.6 years old (SD=3.4). The majority of participants self-identified as White (64.1%), with approximately half of them identifying as male (47.7%). Over half (54.9%) of participants reported past 30-day e-cigarette use, and one-third (33.6%) reported past 30-day use of other tobacco products (Table 1).

Table 1.

Participant Demographics (N = 459)^a

	N (%) or mean (SD)
Age (years)	24.6 (3.4)
Gender
Female	216 (47.1%)
Male	219 (47.7%)
Non-binaryb	24 (5.2%)
Race and ethnicity
Non-Hispanic White/Caucasian	294 (64.1%)
Non-Hispanic Black/African American	24 (5.2%)
Othersc	97 (21.1%)
Hispanic/Latino	44 (9.6%)
Other tobacco use
Currentd	154 (33.6%)
E-cigarette usee
Never	107 (23.3%)
Ever	100 (21.8%)
Current	252 (54.9%)
Message exposure
Yes	94 (20.5%)
No	315 (68.6%)
Not sure	50 (10.9%)

^aParticipants were recruited from Prolific, an online crowdsourcing platform.

^bNon-binary includes trans female/trans woman, trans male/trans man, gender queer/gender non-conforming/gender expansive.

^cOthers include multiple race, Asian, Native American or Alaskan Native, or Pacific Islander.

^dCurrent other tobacco use includes using cigarettes, cigars, smokeless tobacco, and hookah at least once in the past 30 days.

^eParticipants “never” used e-cigarettes if they reported never using the product, not even a puff; “ever” used e-cigarettes if they reported using the product at least once, but not in the past 30 days; and “currently” used e-cigarettes if they reported using the product at least once in the past 30 days.

Exposure on outcomes

Prior message exposure was not significantly associated with perceived source credibility (p=0.568), message trust (p=0.616), curiosity (p=0.915), e-cigarette use interest (p=0.458), perceived message effectiveness (p=0.949), beliefs (p=0.475), harm perceptions (p=0.145), and intentions to refrain (p=0.103).

Conditions on outcomes

The overall differences in the conditions on perceived source credibility were statistically significant in the unadjusted model (p<0.001). Participants in the expert condition reported higher perceived source credibility compared to those in the friend condition (p<0.001) and to those in the influencer condition (p=0.001), but the differences between the friend condition and the influencer condition were not statistically significant (p=1.00; Table 2). The results persisted in the adjusted models, controlling for age, gender, race, ethnicity, e-cigarette use, and other tobacco use (p<0.001; friend vs. expert, p<0.001; influencer vs. expert, p<0.001; friend vs. influencer, p=1.00; Table 3).

Table 2.

One-way ANOVA estimating associations of conditions on outcomes.

Measure		M	SE	F(2, 456)	p-value
Perceived source credibility	Expert	5.00	0.11	9.128	<0.001
	Friend	4.34	0.12
	Influencer	4.40	0.12
Message trust	Expert	6.86	0.18	6.780	0.001
	Friend	5.99	0.19
	Influencer	6.03	0.19
Curiosity	Expert	3.34	0.15	2.976	0.052
	Friend	3.24	0.15
	Influencer	2.88	0.13
E-cigarette use interest	Expert	2.20	0.11	1.046	0.352
	Friend	2.42	0.13
	Influencer	2.22	0.12
Perceived message effectiveness	Expert	3.49	0.09	2.266	0.105
	Friend	3.29	0.08
	Influencer	3.25	0.08
Beliefs	Expert	3.52	0.10	0.689	0.503
	Friend	3.39	0.10
	Influencer	3.37	0.09
Harm perceptions	Expert	8.51	0.18	0.963	0.382
	Friend	8.15	0.19
	Influencer	8.32	0.18
Intentions to refrain	Expert	3.40	0.11	0.714	0.490
	Friend	3.24	0.11
	Influencer	3.23	0.11

Note: Models adjusted for multiple comparisons using a Bonferroni correction.

Table 3.

Adjusted models estimating associations of conditions on outcomes controlling for covariates.

Measure		M	SE	F(6, 452)	p-value
Perceived source credibility	Expert	5.01	0.11	11.282	<0.001
	Friend	4.32	0.11
	Influencer	4.42	0.11
Message trust	Expert	6.89	0.18	8.297	<0.001
	Friend	5.94	0.18
	Influencer	6.05	0.18
Curiosity	Expert	3.35	0.14	3.12	0.045
	Friend	3.24	0.14
	Influencer	2.88	0.14
E-cigarette use interest	Expert	2.20	0.11	1.430	0.24
	Friend	2.44	0.11
	Influencer	2.21	0.11
Perceived message effectiveness	Expert	3.49	0.08	2.607	0.075
	Friend	3.26	0.08
	Influencer	3.26	0.08
Beliefs	Expert	3.53	0.09	1.068	0.344
	Friend	3.36	0.09
	Influencer	3.39	0.09
Harm perceptions	Expert	8.54	0.17	1.880	0.154
	Friend	8.08	0.17
	Influencer	8.35	0.17
Intentions to refrain	Expert	3.41	0.10	1.218	0.297
	Friend	3.21	0.10
	Influencer	3.26	0.10

Note. One-way ANOVA test was performed to adjust for age, gender, race, and ethnicity, e-cigarette use status, and other tobacco use. Models adjusted for multiple comparisons using a Bonferroni correction.

The overall differences in the conditions on message trust were statistically significant in the unadjusted model (p=0.001). Participants in the expert condition reported higher message trust compared to those in the friend condition (p=0.004) and to those in the influencer condition (p=0.006), respectively. The differences between the friend condition and the influencer condition were not statistically significant (p=1.00; Table 2). The results persisted in the adjusted models (friend vs. expert, p<0.001; influencer vs. expert, p=0.003; friend vs. influencer, p=1.00; Table 3).

The overall differences in the conditions on curiosity (p=0.052), perceived message effectiveness (p=0.105), e-cigarette use interest (p=0.352), beliefs (p=0.503), harm perceptions (p=0.382), and intentions to refrain (p=.490) were not statistically significant in the unadjusted models (Table 2). In the adjusted models, results remained not significant for e-cigarette use interest (p=0.240), perceived message effectiveness (p=0.075), beliefs (p=0.344), harm perceptions (p=0.154), and intentions to refrain (p=0.297). The adjusted model predicting curiosity (p=0.045) was significant, although the differences across conditions were not statistically significant (Table 3).

The interaction effects between condition and e-cigarette use status on any of the outcomes were not statistically significant.

Discussion

We examined the impact of health information sources on young adult perceptions of Instagram e-cigarette education posts. Our findings suggest that young adult perceptions of the message source, message trust, and curiosity in e-cigarette information differed based on the source type. Specifically, young adults who viewed Instagram e-cigarette education posts from a health expert reported greater perceived source credibility, message trust, and curiosity toward the posts. Findings from our study present opportunities for future research and implementation of these strategies to improve e-cigarette education posts on social media.

In the current sample, we found that young adults perceived a health expert (i.e., medical school) as a more credible source of Instagram e-cigarette education posts than a friend or social media influencers. Our findings are consistent with the existing evidence that young adults perceived a health expert (i.e., the FDA) as a more credible source of e-cigarette health information than a young adult peer.³⁴ Furthermore, other research shows that the majority of young adults trust scientists and health experts,³⁵ which potentially explains our findings. Empirical evidence on other health topics (i.e., cancer) suggests that people have the highest trust in government health organizations and lowest trust in people without a scientific background on social media.³⁶

Additionally, we found that young adults rated Instagram e-cigarette education posts from a health expert higher on trustworthiness compared to posts from friends and influencers. Based on cognitive dissonance theory,³⁷ which posits that individuals seek to maintain their preexisting attitudes to avoid psychological discomfort, it is possible that participants showed more positive responses to a source that was congruent with their expectations. Typically, health experts provide health information, whereas friends and social media influencers tend to endorse risk behaviors, such as using e-cigarettes.²⁴ Similarly, past empirical work has demonstrated that young people respond more positively to social media e-cigarette advertisements featuring models who endorsed pleasurable lifestyles compared to those featuring models who endorsed healthy lifestyles because the models matched their expectations.³⁸

Our findings potentially indicate that young adults may require different strategies than adolescents between 12 and 17 years old (as targeted by the FDA’s The Real Cost campaign) when designing e-cigarette education messages. Most federal-level tobacco prevention work has focused on adolescents, yet our results indicate that the same campaign strategies may not be equally effective for young adults, as our results showed higher perceived source credibility and message trust when messages were from health experts (compared to friends and influencers). This finding also algins with national survey data, which demonstrated high trust in health experts and scientists among young adults.³⁵ Collectively, our data suggest that health experts may be a more appropriate source to share or endorse e-cigarette education messages for young adults. The association between the condition (expert, friend, influencer) and e-cigarette beliefs, use interests, curiosity, harm perceptions, and intentions to refrain were not significant, as a one-time exposure experiment is unlikely to change these hard-to-shift beliefs and intentions.³⁹

Contrary to our hypothesis, e-cigarette use status did not moderate the association between the source and outcomes. This finding contradicts the existing evidence that tobacco use status can influence perceptions and trust toward tobacco health information.^18,40 Specifically, past work has shown that people who currently use tobacco are more likely to accept anti-tobacco messages from someone with similar attributes (i.e., peer) than those without shared attributes.²¹ Future research can incorporate these measures to comprehensively understand the role of e-cigarette use status on the association between the source type and perceptions of e-cigarette education messages.

Limitations and future directions

Our study has several limitations. First, we used a convenience sample, although prior tobacco research studies found that results from convenience samples show similar patterns to studies using representative samples.⁴¹ Thus, results from our study provide preliminary evidence to inform the development of a large trial using a nationally representative sample. Additionally, we did not assess participants’ social media use and e-cigarette knowledge, which could provide more insight into how young adults respond to e-cigarette education messages on social media.

Furthermore, it is possible that our participants did not perceive the friend and influencer conditions as relatable as they would for their actual friends or influencers that they follow, which is a limitation of a controlled experimental study. Providing more specific instructions, for example, “influencer you currently follow,” and “a friend that you hang out with regularly,” could help increase relational proximity and better capture the nuances of the peer conditions.

It is also possible that participants who formerly used e-cigarettes but have quit may respond differently to the conditions, compared to those who have never used or currently use e-cigarettes. Future studies should examine the potential effects of participants’ e-cigarette quitting history on the association of social media e-cigarette education messages on outcomes.

Approximately 20 percent of our participants reported previously seeing The Real Cost campaign messages. Prior message exposure was not associated with any of the outcomes (perceived source credibility, message trust, curiosity, e-cigarette use interest, perceived message effectiveness, beliefs, harm perceptions, and intentions to refrain). Additionally, we conceptualized the message source as a social media account user in our study, but conceptualizing the source as advertising models may have led to different results. More work is needed to explore the nuances around different points at which a source can be incorporated into message development. Finally, we manipulated the number of “likes” for each condition to match the source characteristics (e.g., higher likes for influences than friends). Although the number of “likes” did not impact our findings, this is a limitation of the study.

Despite these limitations, our study is among the first to provide insight into how different types of sources – health experts, friends, and influencers – can influence young adults’ perceptions of Instagram e-cigarette education posts. The messages used in this study were intended for an adolescent audience,⁵ but we did not test them in the adolescent audience. More research is needed to investigate the differences between adolescents and young adults and trusted sources of health information to effectively reach each age group. It is critical to consider credible message sources to increase the effectiveness of existing campaign messages, and our formative work provides evidence to support such need.

Conclusions

In this study, we examined how young adults’ perceptions of Instagram e-cigarette education posts may be influenced by three different source types: a health expert, friend, and influencer. We found that source can influence perceived message effectiveness of e-cigarette education messages among young adults, with young adults perceiving a health expert as a more credible source of e-cigarette education messages than a friend or a social media influencer, respectively. Results from our study underscore the importance of understanding who a credible source of e-cigarette education messages is to young adults, which can increase source credibility, message effectiveness, trust, and curiosity in e-cigarette information among young adults. Taken together, our findings suggest that public health officials may be able to leverage health experts, such as medical school institutions, to improve delivery and effectiveness of e-cigarette education messages on social media.

Highlights.

• Using messages from credible sources can help improve message acceptance, yet little is known about its impact on young adults’ responses to e-cigarette education messages.

• Young adults in a health expert condition reported greater trust in a message source and message than those in a friend or an influencer condition after seeing Instagram-style e-cigarette education messages.

• Health experts may be an effective message source of e-cigarette education messages targeting young adults.