Visual Attention to the Use of #ad versus #sponsored on e-Cigarette Influencer Posts on Social Media: A Randomized Experiment

Abstract

Youth and young adults are the largest consumers of social media content. Individuals with large followers are paid to share social media content using specific products for compensation. This type of activity is considered commercial sponsorship and requires a disclosure in order to comply with Federal Trade Commission regulations. Between July and August 2019, youth and young adult (ages 16–24; n = 200) participants were recruited into an eye-tracking laboratory to view their native Instagram feed on a mobile device where a set of posts from e-cigarette influencers were inserted with one of the two potential labeling strategies: #ad and #sponsored. Participants spent an average of 6.6 seconds viewing e-cigarette influencer posts. Youth and young adults spent 3.1 seconds on the area labeled #ad, compared to 2.2 seconds on the area of interest labeled #sponsored (p = .03). After accounting for age, current tobacco use, and dependence, #ad drew 0.93 seconds more than #sponsored on influencer posts (p = .02). Both labeling strategies drew visual attention to Instagram e-cigarette influencer posts, with nearly 1 second more attention paid to the presence of #ad. Labeling commercially sponsored content on social media is a viable strategy to attract the attention of youth and young adults.

In the United States (U.S.), the vast majority of the public (69%) uses social media (Pew Research Institute, 2018; “Social media and young adults,” 2010); over 90% of U.S. young adults actively use one or more types of social media daily (Perrin, 2015), predominantly YouTube, Instagram, Snapchat, and Twitter (“Teens fact sheet,” 2012). Social media play an important role in adolescents’ lives, connecting them with friends, acquaintances, popular culture, along with traditional and influencer product marketing. These connections can shape both attitudes and behaviors, especially among adolescents who learn from observing others (Bandura & Walters, 1977; Cho, Li, Shen, & Cannon, 2019).

Exposure to content that can influence unhealthy behavior, such as vaping, is prevalent among youth. In fact, the portrayal of “vaping lifestyle” (Ritter, 2015) imagery is prevalent on social media, and over half of youth in a 2016 survey reported exposure to information about electronic cigarettes on social media (Hébert et al., 2017; Pokhrel et al., 2018; Sawdey, Hancock, Messner, & Prom-Wormley, 2017). Such exposure is a robust predictor of youth and young adults’ e-cigarette use and susceptibility to tobacco use (Barrington-Trimis et al., 2015; Pokhrel et al., 2018).

Aspirational images featuring attractive, affluent, and popular young people using e-cigarette products are utilized to attract the attention of young people and influence their behavior (Strasser, Tang, Tuller, & Cappella, 2008; Vandewater et al., 2018). For example, a content analysis of e-cigarette-related hashtags on Instagram found a preponderance of product shots, promotion offers, images of participants at tobacco company-sponsored events, and “vape-selfies” (Cortese et al., 2018; Ritter, 2015). Many of these posts are not organic (Allem, Ferrara, Uppu, Cruz, & Unger, 2017; Chu, Allem, Cruz, & Unger, 2017), and these individuals can be well compensated for promoting products. In a recent investigative report on vaping influencers, one individual reported receiving U.S. 1,000 USD for a blog and Instagram post about JUUL (Nedelman & Selig, 2018). Consumer behavior research has demonstrated that social media users are much more likely to be influenced by user-generated content compared to content known to be sponsored by businesses or marketers (Goh, Heng, & Lin, 2013; Liljander, Gummerus, & Söderlund, 2015).

Paid promotional posts by social media influencer are considered commercial sponsorship and require disclosure in order to comply with U.S. Federal Trade Commission (FTC) regulations which protect consumers from deceptive and unfair business practices (Federal Trade Commission, 2013). A collection of leading public health groups petitioned the FTC to stop tobacco and e-cigarette companies from using influencer advertising without required hashtags such as #ad or #sponsored (Campaign for Tobacco-Free Kids, 2018). But to date, no study has examined whether these hashtags attract the attention of social media users (and, if so, which one attracts more attention). The goal of the present study was to examine visual attention paid by youth and young adults to two disclosure labeling strategies on social media posts by e-cigarette influencers.

Methods

A convenience sample of participants (n = 200) ages 16–24-years-old living in one Midwest and one Southern state were recruited to complete a face-to-face randomized experiment. All recruitment efforts were framed as an in-lab evaluation of video content, conducted through an existing panel of youth and adults, maintained by MediaScience Labs (www.mediascience.com). For the recruitment of minors ages 16–17, recruitment e-mails were sent to panelists aged 30+ to determine if eligible teens were living at home. Phone consent was collected from the parent, then eligibility was verified over the phone with the teen during the same call. Eligibility criteria included being between the ages of 16–24 years-old, possessing use of both eyes without reading glasses, the ability to read a screen without difficulty, having an Instagram account, and being willing to provide consent to participate. All participants provided informed consent and parental assent when required. The study protocol was approved by the Advarra Institutional Review Board.

For the experimental stimuli, six organic posts from unique influencers were selected through an expert review of known e-cigarette influencers on Instagram; influencers were defined as individuals with roughly 10–50,000 followers and shared content regarding e-cigarette brands at least several times per week. While both male and female influencers are sponsored to promote e-cigarette products, many of the current vaping lifestyle influencers are females who share imagery that is highly sexualized (“Sex sells, Social Media influences | Social Media Today,” n.d.). The stimuli used in this study reflected that reality, with all of the stimuli featuring females, and one (Mia Khalifa) including sexually suggestive content.

All posts were formatted as either static (single image and text) or dynamic (video and text) to mimic the current landscape on Instagram (see Table 1). Two labeling conditions were assigned to a series of four e-cigarette posts: 1) #ad or 2) #sponsored; these conditions are referred to as “labeled posts.” These two conditions were selected because both the FTC and the public health groups proposed them as methods for complying with the disclosure obligation (Goh et al., 2013). Each study post was manipulated to have the appropriate hashtags placed below the photo and caption with white space surrounding it, placing it separately from other hashtags (see Figure 1). Two control posts were displayed without disclosure, referred to here as “unlabeled posts.”

Table 1. Experimental social media posts by type, format, and frequency

Type of post	Format	Number
Labeled (#ad or #sponsored) E-cigarette endorsement/demonstration	Video	2
Labeled (#ad or #sponsored) E-cigarette endorsement/demonstration	Static image	2
Unlabeled (control exposure) E-cigarette endorsement/demonstration	Video	1
Unlabeled (control exposure) E-cigarette endorsement/demonstration	Static image	1

Figure 1. Areas of interest illustration for social media posts by condition

Once participants arrived at the study laboratory, informed consent procedures were conducted, followed by calibration of the eye-tracking equipment. Once successfully calibrated to the equipment, participants went into a private room where they logged in to their personal Instagram account on a standardized cell phone equipped with a remote eye-tracking system (Tobii Pro X3-120) which used an infrared camera to precisely capture visual attention duration and patterns on areas of interest (AOIs) drawn in the static or dynamic content (Duchwoski, 2007). Participants were asked to browse content naturally as they normally would on their own phone. The sequence of posts was standardized so that the initial display drew were from a participant’s own Instagram feed, and then the stimuli posts were interjected using MediaScience Labs proprietary content insertion software into the users’ natural feed using a fixed rotation. The cell phone was placed in a stand to keep it stationary, reduce user movement, and maximize usable eye-tracking data. Once all experimental posts were seen or the fifteen-minute time-period elapsed, a computer-based survey was taken.

Measures

Two AOIs were drawn on each social media post: 1) entire post including imagery (static or dynamic) and text, and 2) experimental hashtag and surrounding white space (see Figure 1.)

Post-experiment survey measures included demographics of age (in years), any current tobacco use (e-cigarettes, cigarettes, cigars/cigarillos, hookah, and nicotine replacement; yes/no); time to first tobacco use within five minutes of waking was used as a single-item proxy for nicotine dependence (Carpenter, Baker, Gray, & Upadhyaya, 2010).

Analysis

The primary outcome of interest in the present study was visual attention (measured in cumulative milliseconds of dwell time) to labeled hashtag AOI; results are reported in seconds for ease of interpretation. Descriptive statistics include the mean total dwell time on the entire post (experimental and control) and the experimental hashtags. The main study outcome was the average difference in dwell time on hashtag AOI which was derived by summing dwell time for all four experimental posts and for each post separately. The primary predictor was the study condition (#ad vs. #sponsored). The relationship between study condition used and attention paid to the hashtag AOI was examined with generalized estimating equation regression to account for differences in visual attention due to dynamic format (video content versus imagery only) and source (influencer). To account for any attentional bias to tobacco cues among current tobacco users, the final multivariate regression analysis included covariates to account for current tobacco use, dependence, and age (in years). Tests of association were two-sided (p < .05). All analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC).

Results

A total of 200 participants (mean age of 20.6) completed the study procedures (shown in Table 2). Approximately 51% of participants were male, 65% were White, and nearly half of the sample reported current use of any type of tobacco product (47%). There were no significant demographic differences between the study conditions (p > .05).

Table 2. Social media participant demographics (n = 200)

Demographics	Total (n = 200)	#ad (n = 103)	#sponsored (n = 97)
Mean age in years (SD)	20.6 (2.6)	20.3 (2.6)	20.9 (2.5)
Gender identity Female  Male  Other	47.0% 50.5% 2.5%	49.5% 47.6% 2.9%	44.3% 53.6% 2.1%
Race*  White  African American/Black  American Indian  Asian  Other race	65.5% 17.0% 1.0% 9.0% 13.0%	65.0% 15.5% 2.0% 12.6% 13.6%	64.9% 18.6% 0.0% 5.2% 12.4%
Hispanic/Latino	26.0%	23.3%	28.9%
Educational attainment  In high school or less  High school graduate  Some college/2 year degree  4 year degree or higher	18.7% 12.8% 47.3% 20.2%	22.9% 11.4% 48.6% 17.1%	16.3% 14.3% 45.9% 23.5%
Tobacco use characteristics
Any current tobacco use	47.0%	48.6%	43.9%
Cigarettes every day/some days	17.0%	17.5%	16.5%
Cigar every day/some days	12.5%	14.6%	10.3%
E-cigarettes every day/some days	35.5%	40.8%	29.9%
Chew every day/some days	3.0%	3.8%	2.1%
Hookah every day/some days	21.0%	18.5%	23.7%
Nicotine replacement every day/some days	2.0%	2.9%	1.0%
Smoked ≥ 100 cigarettes in lifetime	16.5%	18.4%	14.4%
Use tobacco ≤5 min of waking	2.5%	2.0%	3.1%
Definitely try cigarettes in the next year	14.0%	17.5%	10.3%
Definitely try e-cigarettes in the next year	22.5%	27.2%	17.5%
Definitely try hookah in the next year	18.0%	17.5%	18.6%

*Totals may exceed 100% due to multiracial category selection.

Descriptive statistics on viewing patterns are shown in Table 3. There were no significant differences in viewing time by demographics (p > .05). Participants spent an average of 6.6 seconds viewing the influencer posts, compared to 8.4 seconds on the unlabeled posts. Dwell time ranged between 4.1 and 9.5 seconds across the individual influencer posts. For mean dwell time by study condition, participants spent 2.6 seconds overall on either hashtag, with a mean of 3.1 seconds on #ad and 2.2 seconds on #sponsored.

Table 3. Eye-tracking descriptive statistics for e-cigarette influencer posts on Instagram (n = 200)

Eye-tracking metrics	Entire post	Either hashtag	#ad	#sponsored
Dwell time in seconds (SD)
Labeled influencer posts	6.7 (6.2)	2.6 (2.7)	3.1 (3.3)	2.2 (1.9)
Courtney Vapes	8.8 (10.8)	1.5 (2.3)	1.7 (2.7)	1.3 (1.7)
Buffalo Juice Girl	4.8 (4.1)	2.7 (2.8)	2.7 (3.1)	2.7 (2.4)
Mia Khalifa	9.5 (12.0)	4.2 (6.6)	5.7 (8.0)	2.7 (4.4)
Pandora Blue	4.1 (3.2)	2.8 (2.5)	3.0 (2.4)	2.6 (2.6)
Unlabeled influencer posts
Justpeachyy	11.4 (11.1)	–	–	–
Vapeporn	4.2 (3.1)	–	–	–

In Table 4, results are presented for dwell time differences by study condition after accounting for individual influencers. Participants in the #ad condition spent 0.9 seconds more viewing the hashtag area of interest compared to #sponsored condition (p = .03). In the final multivariate model after accounting for current tobacco use and age, the #ad condition attracted 0.93 seconds more visual attention compared to #sponsored condition (p = .02).

Table 4. Bivariate and multivariate regression analysis differences in visual attention by condition (n = 200)

Variables	Dwell time (sec)*
Dwell time on #ad (vs #sponsored)	0.90 (0.40)
Courtney Vapes	0.35 (0.43)
Buffalo Juice Girl	0.05 (0.50)
Mia Khalifa	3.03 (1.14)
Pandora Blue	0.43 (0.44)
Multivariate model
Dwell time on #ad (vs #sponsored)	0.93 (0.41)
Current tobacco use (any type)	0.05 (0.39)
Age (in years)	0.12 (0.08)

*Bold type indicates statistical significance <0.05.

Discussion

Findings indicate that youth and young adults paid significantly greater visual attention to the labeling strategy of #ad compared to #sponsored for e-cigarette-related posts, and results did not differ by tobacco use. Overall visual attention to any labeled post was lower compared to unlabeled posts (6.6 versus 8.4 seconds). These results demonstrate that although the #ad garnered a small but significantly greater amount of visual attention than #sponsored, either labeling strategy attracts attention from the target audience of youth and young adults.

Evidence suggests that sponsorship disclosures are effective in improving consumer recognition of content as advertising (Ohlhausen, Chairman, & Mcsweeny, 2017). Further, a growing body of research suggests that consumers, and young people, in particular, are less likely to purchase a brand when influencers disclose brand sponsorship in blog or social media posts (Goh et al., 2013; Liljander et al., 2015; Tucker, 2012; van Reijmersdal et al., 2016). Recognizing sponsorship disclosures on social media content impacts perceptions of content as an advertisement and may reduce the credibility of those influencers (Vogel, Guillory, & Ling, 2020). Our findings, in light of the robust theoretical and empirical evidence for the influence of social media on young people (Cho et al., 2019), support mandating some type of disclosure on tobacco influencer posts. In particular, results suggest that the relatively simple use of a hashtag can assist young consumers in assessing the credibility of posts, supporting the potential efficacy and effectiveness of this approach in the visually busy social media environment.

Since the implementation of this study Instagram and Facebook has banned paid influencer posts that promote vaping and other tobacco use on their platforms although the mechanism of enforcement for this policy has yet to be detailed (“Instagram to ban influencers from promoting vaping and guns,” 2019). These voluntary actions may be insufficient at preventing exposure to paid tobacco-related influencer content, particularly given the lack of penalty associated with posting such content; the FTC has taken action against individual influencers found in violation of deceptive practices regarding commercial sponsorship (“CSGO Lotto Owners Settle FTC’s First-Ever Complaint Against Individual Social Media Influencers | Federal Trade Commission,” n.d.). If the FTC adopted a standard hashtag-based approach to identify paid influencer content, then the penalty associated with noncompliance with this rule may be enough to deter companies and individual influencers from engaging in this marketing strategy.

As advertisements on social media actively work to replicate organic content, evidence persists that such messages are perceived as trustworthy and persuasive (Johnson, Potocki, & Veldhuis, 2019; Laestadius, Penndorf, Seidl, & Cho, 2019). Thus, commercially sponsored posts, even with disclosure, are likely to be more persuasive to youth than formal ads. It is relevant to consider whether the U.S. Food and Drug Administration (FDA) deems influencer posts advertising under the Tobacco Control Act. If such posts are considered advertising by the FDA, they must also include the mandatory warning, “WARNING: This product contains nicotine. Nicotine is an addictive chemical” which can additionally encourage consumer recognition of influencer content as advertising (“Covered” Tobacco Products and Roll-Your-Own/Cigarette Tobacco Labeling and Warning Statement Requirements | FDA, n.d.). Experimental evidence suggests the inclusion of this warning message within tweets resulted in more negative brand perceptions (Guillory et al., 2020), suggesting users recognize and consider the warning information. Importantly, the lack of a disclosure on influencer posts may also be considered a misleading form of advertising, which would constitute a separate violation of the Tobacco Control Act (“Required warnings for cigarette packages and advertisements. Final rule,” 2011) in addition to a violation of any adopted standard by the FTC.

There are important limitations of the current research to consider. Instagram currently uses the label of “paid partnership with” above paid posts, which may have drawn more attention to the #ad and #sponsored as novel terms for this platform versus #sponsorship which may be similar in character length and concept to partnership. Second, Instagram was selected based on the popularity of this format among the target audience but testing for effectiveness of a consistent labeled strategy is needed across all social media platforms. Our placement of the experimental hashtags was consistent by study condition, but future research should examine whether placement in the bio, at the start of a caption per FTC recommendations, in line with other hashtags, etc. impacts visual attention. Future research should also take into account that optimal placement likely varies by the social media platform. Participants may have paid more attention to the experimental stimuli because of sexualized imagery of women, and the potential novelty of posts from content outside of their normal feed; there was no significant effect for gender identity on visual attention to the hashtags or the entire posts (data not shown). Examination of a potential difference by gender identity may be warranted. The present study focused on e-cigarettes, so attention to other commercial sponsorships may vary by the product type. Since the sampling frame for this study focused on youth and young adults ages 16–24, these results may not generalize to older or younger age groups.

The use of hashtags to indicate a paid influencer post appears to successfully attract visual attention from youth and young adult users within a naturalistic social media environment. Establishing that youth and young adults actually pay attention to such indications of commercial intentions on social media is an important first step in formulating practicable and effective approaches to regulating commercial content on social media. Further work is necessary to understand whether and how such indicators affect message credibility, attitudes about the product, and intentions to use. Given the importance of this medium of influencing youth and young adults and the limits of relying only on self-regulatory actions from social media companies or the tobacco industry, regulatory agencies must establish and enforce standards for commercially sponsored e-cigarette content on social media.

Ethics Approval

Advarra institutional review board approved this study protocol.