Tobacco Advertising Exposure and Product Use among Young Adults: An Ecological Momentary Assessment Approach

Abstract

Introduction:

Tobacco-related content is prevalent on social media, yet many methods of measuring exposure are inadequate due to the personalized nature of online marketing. The purpose of this paper is to examine the association between exposure to pro-tobacco messages (both industry-sponsored and user-generated) and the use of tobacco products, as reported via ecological momentary assessment (EMA).

Methods:

Young adults (n=175) were instructed to record all sightings of marketing (both in-person and online) related to tobacco for 28 days. Tobacco product use and recall of message encounters were assessed daily using app-initiated EMA.

Results:

Participants who reported exposure to tobacco messages were significantly more likely to report using tobacco, adjusting for gender, age, race/ethnicity, baseline use of any tobacco product, and having friends who use tobacco and e-cigarettes (p <.001). For each industry-sponsored message viewed, the odds of using tobacco or e-cigarettes in a given day increased by a factor of 1.77 (95% CI =1.41, 2.23). For each user-generated message viewed, the odds of using tobacco or e-cigarettes in a given day increased by a factor of 1.52 (95% CI=1.27, 1.83).

Discussion:

To our knowledge, this is the first study to specifically examine the association between exposure to user-generated messages and daily tobacco use. The findings suggests that there is a unique element to user-generated messages that distinguishes them from both traditional marketing and from simple peer influence.

INTRODUCTION

Despite promising declines in cigarette and smokeless tobacco use in the past decade, 17.6% of young adults aged 18–24 are current tobacco users.¹ Young adulthood is a time of transition that provides many opportunities for the adoption and progression of tobacco use due to changes in environment, peer groups, and life stressors.² While first use of tobacco typically occurs in adolescence,³ the transition from experimental smoking to regular smoking is often solidified during young adulthood.^4,5 Recent studies have also suggested that the initiation of tobacco products among never-using young adults is greater than adolescents.⁶ Tobacco use among young adults is unique from other age groups in a number of ways. Among adults, those aged 18–24 have the highest rates of hookah use,⁷, e-cigarette use,¹, and polytobacco use.¹ Compared to their older peers, young adults (i.e., those aged 18–29)⁸ are more attracted to flavored products, such as flavored cigars and menthol cigarettes,⁹ and the availability of flavors of snus, hookah, and e-cigarettes contributes to an overall positive perception of these products as “fun” and “recreational” among young adults.¹⁰

As regulation and prevention programs focused on children and adolescents have helped reduce tobacco initiation in the younger age groups, young adults remain vulnerable to the tobacco industry as the “youngest legal targets” for tobacco marketing.¹¹ The relationship between tobacco marketing and initiation of tobacco use among adolescents is well-documented, with sufficient evidence to demonstrate a causal, dose-response relationship between exposure to tobacco product marketing and cigarette smoking initiation.^12,13 However, there is a relative dearth of evidence regarding the relationship between marketing and tobacco use specifically among young adults. Many studies are primarily cross-sectional in nature^14,15 and are thus unable to examine the temporal relationship between marketing exposure and use. Others examine advertising receptivity (typically defined as ownership of a promotional item or the ability to name a favorite tobacco ad)^16,17 rather than direct measures of marketing exposure. A few recent longitudinal studies have provided preliminary evidence that exposure to tobacco marketing is associated with tobacco use initiation among young adults. Mantey et al. (2019) found that self-reported exposure to smokeless tobacco marketing is associated with initiation of smokeless tobacco after 6-months.¹⁸ Loukas et al. (2019) found that recall of retail store-based e-cigarette marketing was associated with higher odds of e-cigarette initiation among youth and young adults up to 2.5 years later.¹⁹ Chen-Sankey et al. examined data from the Population Assessment of Tobacco and Health study (PATH) and found that young adults exposed to e-cigarette marketing at wave 2 were more likely to have experimented with e-cigarettes at wave 3 compared with those not exposed, yet the association did not remain significant among young adults not susceptible to e-cigarettes.²⁰ Recent trends suggest that initiation of cigarette and e-cigarette use occurs now occurs more often among young adults than among adolescents,⁶ thus, additional research on the relationship between marketing and tobacco use among young adults is warranted.

With the rise of social media, the landscape of pro-tobacco and e-cigarette messaging is no longer limited to industry-sponsored advertising. In recent years, user-generated content, especially about hookah, cigars, and e-cigarettes,²¹ has been observed on a multitude of social networking sites such as YouTube, Instagram, and Facebook. Much of the content is in the form of user videos²² or imagery such as “vaping selfies,”^23,24 while websites like Reddit have multiple forums dedicated to the topics of e-cigarettes and vaping, where users share information about modifying e-cigarette devices and the best e-liquid flavors.²⁵ The presence of user-generated pro-tobacco messages on social media is well-documented. Thus far, research has been limited to either small, descriptive studies,^26,27 or studies that take a very broad approach examining overall trends rather than quantifying specific exposure.^28,29 Few studies to date have directly examined the impact of online tobacco and e-cigarette marketing on use behaviors,^30,31 but research suggests that social media depictions of tobacco use predict future smoking tendency,³² and that adolescents who use tobacco and e-cigarettes are exposed to and engaged with tobacco-related social media more than their peers.³³

Historically, measuring exposure to tobacco marketing has been limited due to an overreliance on recall, recognition, or proxy measures such as advertising receptivity. Common measures such as asking participants how frequently during the past 30 days they saw tobacco ads in various locations is subject to recall bias,³⁴ while methods like the recollection of specific tobacco ads or ownership of branded material are more likely to be reflective of an individual’s choice to seek out tobacco related items or promotions than they are of advertising exposure itself.³⁵ Moreover, current methods of measuring exposure to advertising are inadequate for internet and social-media based-ads due to the targeted and personalized nature of online marketing.³⁶ Targeted advertising dramatically changes the online experience of each individual, and it is extremely common, with 62% of adults between the ages of 18 and 29 reporting that they noticed advertisements online that were directly related to sites that they had recently visited or things for which they had recently searched.³⁷ Although some studies have attempted to summarize online tobacco and e-cigarette advertising by examining marketing expenditures and conducting broad internet searches to determine the general characteristics of ads,²⁹ this method will at best capture only those messages that are industry-sponsored rather than user-created.³⁸

In order to overcome these limitations, we utilized ecological momentary assessment (EMA). EMA is a method that utilizes the repeated collection of real-time data on participants’ behavior and experience in their natural environment.³⁹ This method is unique because it accounts for environmental characteristics of the measurement, and because it avoids the retrospective distortion of data.³⁴ EMA has been used in behavioral science for years, including in research examining the environmental and psychological antecedents of cigarette smoking⁴⁰ and smoking cessation attempts.⁴¹ Several studies have demonstrated the feasibility of EMA for capturing tobacco and alcohol marketing exposure and product use.^42–45 Exposure to pro-smoking media measured via EMA has shown to be associated with lapsing during quit attempts among adults,⁴⁶ as well as higher mean levels of future smoking risk (i.e., susceptibility to tobacco use) among college students.^47,48 Roberts et al. (2019)⁴⁹ examined the impact of tobacco marketing exposure among adolescents in a 10-day EMA study and found that recent tobacco marketing exposure was associated with recent tobacco use and increased likelihood of future tobacco use, and that youth who reported more exposure had more favorable attitudes towards the ads they saw.⁴⁹ This level of granular detail provides important insight into the mechanisms and outcomes of tobacco marketing, and is particularly useful in identifying potential targets for regulation and prevention. The purpose of this paper is to examine the association between exposure to pro-tobacco and e-cigarette messages (both industry-sponsored and user-generated) and the use of tobacco products among young adults, as reported via EMA.

METHODS

Participants

Participants in this study were young adults aged 18–29 living in Austin, Texas and its surrounding areas. Participants were recruited via printed flyers placed on the University of Texas at Austin and Austin Community College campuses, in local smoke shops near those campuses, and online ads placed on Craigslist and the University of Texas at Austin online events calendar. Participants were enrolled into the study after completing a brief eligibility questionnaire on the study website, and were required to 1) be between the ages of 18 and 29, 2) speak English, and 3) to own a smartphone capable of accessing the Internet.

Procedure

The study procedure was approved by the Institutional Review Board at [Details omitted for anonymized review]. Data collection took place between March 2015 and June 2015. At baseline, participants provided informed consent and were asked to download and install SurveySwipe by SurveyAnalytics, a free smartphone application. Participants completed a baseline survey to evaluate their demographics, use of social media and the internet, use of tobacco and alternative tobacco products, and perceptions of tobacco product advertising, and were given instructions via e-mail for participating in the 28-day study.

Participants were instructed to record all sightings of marketing or social media related to tobacco or electronic nicotine delivery systems seen during the study period. Products of interest were defined as: cigarettes, cigars, little cigars, or cigarillos, chewing tobacco, snuff, snus, or dip, hookah, dissolvables (such Camel orbs, sticks, or strips), and e-cigarettes, vape pens, or personal vaporizers. Participants were asked to report both industry-sponsored materials (defined as printed ads or flyers, billboards, coupons or promotional-email offers, online ads, and industry-sponsored social media such as an official Facebook page), as well as user-generated materials (defined as Facebook posts, Instagram photos, Vines, or Tweets, and online discussion threads).

Daily survey.

Tobacco product use and recall of advertisement sightings were evaluated every 24 hours. A push notification was delivered to participants’ phones each day at 12 P.M. to remind them that a new daily survey was available. Participants’ first daily survey was sent at the first 12 P.M. timeslot following their completion of the baseline survey. Participants who did not complete the daily survey by 9 A.M. the following morning would also receive an e-mail reminder message to complete the survey. Participants who were inactive for 7 consecutive days were sent an e-mail reminding them of the study participation requirements as well as contact information for the research team and a link to unsubscribe from the study if they desired. Participants who remained inactive after 14 days, despite the e-mail reminder, were dropped from the study.

Event-driven survey.

All encounters were submitted via a repeatable survey in the SurveySwipe application that allowed participants to upload a photograph or screenshot. Because we were interested in exploring what messages participants noticed without any kind of external influence, participants were not prompted or reminded by the mobile app to report any sighted messages.

Incentives and study completion.

After participants had completed 28 daily surveys, they were asked to complete a follow-up survey, in which they reported their message recall and tobacco product use from the past 30 days, and answered some usability questions related to the mobile app and study protocol. Participants received a gift card for every 7 daily surveys completed during the study period. Due to the possibility that participants might alter their message reporting behavior in order to increase a financial reward, the decision was made not to award incentives based on the submission of the event-driven EMA. A $10 gift card was awarded for completion of the 7 daily surveys in week 1, a $15 gift card for week 2, a $25 gift card for week 3, and a $30 gift card for week 4. In addition, 10 randomly selected participants were given a $150 gift card for participating in the full four weeks of the study.

Measures

Exposure to tobacco and alternative tobacco product messages.

Participants’ exposure to product messages was evaluated in the daily survey. Participants were asked to recall the number of messages they encountered in the past 24 hours, categorized by the source of the message: 1) industry-sponsored messages (e.g., billboards, posts, or online banner ads), and 2) user-generated messages (e.g., social media posts, discussion threads, photos, or videos).

Tobacco and alternative tobacco product use.

Participant’s tobacco use was measured in the daily survey by asking participants to check if they had used any of the following products in the past 24 hours: traditional cigarettes, cigars, cigarillos, or little cigars, smokeless tobacco (e.g., chewing tobacco, snuff, snus, or dip), e-cigarettes or vape pens, and hookah. For the purposes of analysis, we created a new variable to describe overall past 24-hour (daily) tobacco use, defined as having used any product within the last 24 hours (0=no, 1=yes). This served as the primary outcome variable. Use of cigarettes, cigars, and smokeless tobacco was also assessed at baseline by asking participants how many days in the last month they had used the product. For alternative tobacco products, including hookah, snus, dissolvables, and electronic cigarettes, participants were asked if they had used any of the products in the last month. From these items, we created a new variable to define participants past 30-day use at baseline, as 0 = nonuser, 1 = cigarettes, cigar products, hookah, or smokeless tobacco, 2 = e-cigarette only, or 3 = dual user).

Covariates.

Several covariates known to influence smoking and tobacco behaviors were measured, including age, gender (1=male, 0=female), race (coded as a series of dummy variables, including 1.White, 2. Black or African American, 3. Asian, 4. American Indian or Alaska Native, 5. Native Hawaiian or Other Pacific Islander, and 6. Other, where 1=yes, 0=no), Hispanic ethnicity (where 1=yes, 0=no), and enrollment in a 2- or a 4-year college/university. Because some race and ethnicity categories had few or no participants, we collapsed race and ethnicity into one variable (1= white, 2 = black/other/more than one, 3 = Asian, 4 = Hispanic). Finally, perceived peer use was assessed by asking participants to estimate how many of their friends use each of the following products: cigarettes, cigars, smokeless tobacco, e-cigarettes or vape pens, and hookah (0 to 3 scale, from “none” to “all”).

Data Analysis

To evaluate the relationship between exposure to pro-tobacco messages and product use, we used a repeated measures generalized linear mixed model with a binary logistic response function of daily-reported product use. Fixed effects included daily-reported exposure to industry-sponsored messages, daily-reported exposure to user-generated messages, age, gender, race/ethnicity, tobacco user category (defined as past-30 day use of tobacco only, ENDS, or dual use), and number of close friends who use tobacco products. A random effect to account for individual differences across participants was also included. Because a majority (64%) of the sample were categorized as non-users at baseline, we also conducted a sensitivity analyses, limiting the model to those participants who were considered product users at baseline. All analyses were conducted using Stata 15.⁵⁰

RESULTS

Sample Characteristics

One hundred and eighty-one participants completed a baseline survey. Of those participants, 175 completed at least one daily survey, 148 completed all 4 weeks of the study, 10 dropped out, and 23 were dropped due to inactivity. Pearson’s Chi-Square and t-tests revealed no significant differences in age, gender, race/ethnicity, student status, or tobacco use status across either EMA participation or study completion status.⁵¹ Participants included both tobacco/e-cigarette users and non-users. Overall compliance with the daily survey was good, with 80% of all surveys (n=4,224) submitted on consecutive days. Of the 20% of surveys (n=814) submitted on non-consecutive days, 80% were two days apart, and 15% were submitted between 3 and 11 days apart. Because each daily survey could be completed any time between 12 PM and 11:59 AM the following day before a new daily survey was sent, it was possible for submissions to occur less than 24 hours apart. Five percent of all daily surveys were separated by less than one day. To avoid overlap, the first of any two surveys submitted by the same participant within 8 hours of each other was excluded from analysis in the regression model. A total of 4,400 daily surveys were submitted, and 4,308 were included in the analyses. Participant demographics are presented in Table 1.

Table 1.

Sample Characteristics

Characteristic		% or Mean (N or SD) N = 175
Age		20.91 (SD =2.8, range 18–29)
Female		77.7% (136)
Race/Ethnicity
White		46.9% (82)
Hispanic		13.7% (24)
Asian		28.0% (49)
Black/Other/More than One		11.4% (20)
Past 30-day Tobacco Use
Non-User		64.0% (112)
Tobacco Only		20.6% (36)
E-Cigarette or Dual User		15.4% (27)
Close friends who use...	None	Some	Most	All
Cigarettes	47.4% (83)	41.7% (73)	10.3% (18)	0.6% (1)
Cigars	59.4% (104)	34.9% (61)	5.1% (9)	0.6% (1)
Smokeless tobacco	72.0% (126)	24.0% (42)	3.4% (6)	0.6% (1)
E-cigarettes	33.7% (59)	52.0% (91)	14.3% (25)	0.0% (0)
Hookah	15.4% (27)	61.7% (108)	20.0% (35)	2.9% (5)

Note. Tobacco use only defined as use of cigarettes, cigar products, hookah, or smokeless tobacco on at least one day during the last 30 days; dual defined as the use of both e-cigarettes and one of these tobacco products during the last 30 days

Message Exposure and Prediction of Product Use

The number of industry-sponsored messages reported in one day ranged from 0 to 22 (M = 0.18, SD = .70), and the number of user-generated messages reported in one day ranged from 0 to 10 (M = 0.20, SD = .78). Detailed characteristics of the messages have been described elsewhere,^51,52 but briefly, 62.1% of marketing and messages were user-generated, and 37.9% were industry-sponsored. The majority of messages reported via Event EMA (45.0%) were seen on the Internet, followed by 32.7% seen at the retail point of sale (inside or outside a convenience store, gas station, grocery store, drug store, or smoke shop). The majority of Event EMA messages depicted traditional cigarettes (47.2%), followed by e-cigarettes (31.3%), cigars (7.3%), hookah (7.0%), and smokeless tobacco (3.4%). Product use was reported in 13% (N=573) of the total daily surveys. Although we limited our analyses to focus on pro-tobacco messages, the images submitted via our event-driven EMA survey revealed that about 5% of the submitted user-generated messages were not pro-tobacco or e-cigarettes. For example, in some instances, users shared images, text, or links to articles that either described the harms of tobacco or e-cigarettes, or depicted the use of these products in a negative light.

Table 2 presents the results of the generalized linear mixed model predicting daily tobacco and e-cigarette use from pro-tobacco message exposure. Participants who reported exposure to tobacco and e-cigarette messages were significantly more likely to report using tobacco and e-cigarettes, adjusting for gender, age, race/ethnicity, baseline product use, and having friends who use tobacco and e-cigarettes (p <.001). For each additional industry-sponsored message viewed, the odds of using tobacco or e-cigarettes in a given day increased by a factor of 1.77 (95% CI =1.41, 2.23). For each additional user-generated message viewed, the odds of using tobacco or e-cigarettes in a given day increased by a factor of 1.52 (95% CI=1.27, 1.83). In the sensitivity analysis of the same model restricted to participants who were categorized as product users at baseline, a total of 63 participants were included with 1,442 observations. Results in this model were similar; the odds of using an e-cigarette or tobacco product within a given day increased by a factor of 1.70 (95% CI=1.29, 2.24) for each additional industry sponsored message viewed, and by a factor of 1.44 (95% CI=1.17, 1.77) for each additional user-generated message viewed.

Table 2:

Generalized Linear Mixed Model Predicting Tobacco and E-Cigarette Use from Pro-tobacco Message Exposure

Fixed Effects
Model Term	Coefficient	SE	z	p	95% CI	AOR	AOR 95% CI
Intercept	−12.62	1.94	−6.49	0.00	(−16.43, −8.81)	0.00	(0, 0)
Number of Daily Industry Messages	0.57	0.12	4.86	0.00	(0.34, 0.8)	1.77	(1.41, 2.23)
Number of Daily User Messages	0.42	0.09	4.46	0.00	(0.24, 0.61)	1.52	(1.27, 1.83)
Male	1.16	0.54	2.16	0.03	(0.11, 2.22)	3.20	(1.11, 9.19)
Age	0.24	0.08	2.97	0.00	(0.08, 0.4)	1.27	(1.08, 1.49)
Race/Ethnicitya
Hispanic	1.99	0.64	3.10	0.00	(0.73, 3.25)	7.32	(2.08, 25.78)
Black/Other	0.94	0.74	1.27	0.21	(−0.51, 2.39)	2.55	(0.6, 10.86)
Asian	0.34	0.60	0.56	0.57	(−0.84, 1.51)	1.40	(0.43, 4.54)
Baseline Tobacco Useb
Tobacco Only	3.86	0.60	6.43	0.00	(2.68, 5.03)	47.38	(14.62, 153.55)
E-Cigs or Dual	4.65	0.68	6.89	0.00	(3.33, 5.98)	105.07	(27.96, 394.81)
Close friends who usec
Cigarettes	0.76	0.37	2.03	0.04	(0.03, 1.49)	2.14	(1.03, 4.44)
Cigars	−0.11	0.39	−0.27	0.78	(−0.87, 0.66)	0.90	(0.42, 1.93)
Smokeless tobacco	−1.33	0.43	−3.10	0.00	(−2.17, −0.49)	0.27	(0.11, 0.61)
E-cigarettes	0.95	0.43	2.19	0.03	(0.1, 1.8)	2.59	(1.11, 6.07)
Hookah	−0.32	0.36	−0.89	0.37	(−1.01, 0.38)	0.73	(0.36, 1.46)
Random Effect
Random Effect Covariance	Estimate	SE			95% Confidence Interval
Variance	4.62	1.18			(2.81, 7.61)

^aWhite = reference

^bNon-user = reference

^cNo close friends who use = reference

DISCUSSION

The purpose of this study was to use EMA to measure exposure to e-cigarette and/or tobacco messaging and examine the relationship between pro-tobacco and e-cigarette message exposure and product use among young adults. Although the extant literature suggests a strong, dose-response relationship between tobacco marketing exposure and tobacco use among youth,¹² few studies have focused on young adults, and the majority of existing measures of marketing exposure have relied on recall and recognition. Our results confirmed a significant relationship between daily exposure to tobacco and e-cigarette messages and use of these products.

To our knowledge, this is the first study to specifically examine the association between exposure to user-generated messages and daily tobacco and e-cigarette use. Most of the user-generated images shared by participants in the event-driven EMA were personal photos or videos on social media that featured a tobacco product or e-cigarette, but were not necessarily explicitly promoting their use. For example, “selfies” in which the subject was holding a cigarette were common. These messages, although not directly encouraging tobacco use, serve to normalize the use of tobacco and e-cigarettes among the participants’ social networks, which has been shown to predict future tobacco use. However, many of the messages were more explicit in their promotion of tobacco and e-cigarettes by clearly featuring a specific brand or product or including positive imagery or text related to product use.

The finding that user-generated messages had a comparable association with product use as industry-sponsored messages, which was also independent of having close friends who use tobacco or e-cigarettes, suggests that there is a unique element to user-generated messages that distinguishes them from both traditional marketing and from simple peer influence. Advertising research has shown that consumers identify and interpret advertising as attempts to persuade, and that this knowledge affects their perceptions and intentions towards the ad, often causing them to be skeptical or suspicious of its claims.⁵³ However, there is evidence to suggest that ads seen in non-traditional settings may not be perceived as advertising, and thus may not trigger the same suspicion.⁵⁴ For example, Colliander and Dahlén (2011) found that brand marketing on popular blogs was associated with higher attitudes and purchase intentions compared to marketing in online magazines, and that this difference could be explained in part by the blogs’ engagement with its users.⁵⁵ Furthermore, user-generated messages are viewed within the context of one’s online social network, adding to the perception that the messages are more relevant and more acceptable to the viewer.⁵⁶ The presence of friends or an external source of social identity clearly affects the response to and intentions associated with a pro-tobacco message, as Setodji et al. (2013) found that pro-smoking media exposure was associated with stronger smoking intentions and lower smoking refusal self-efficacy when college students were in the presence of friends.⁵⁷ As user-generated messages contain elements of both pro-tobacco marketing and social identity, more research is needed to fully understand their impact on tobacco use intentions and behaviors. This finding is especially important given the rise in use of paid “influencers,” or individuals with a high following on social media, to promote tobacco products.^58–60 Although many social media websites such as Facebook and Twitter have policies that prevent the direct promotion of tobacco products by tobacco manufacturers,^61,62 tobacco companies have managed to subvert this by utilizing interactive material that encourages users to share content and spread brand preferences to their peers,⁶³ as well as using influencers to promote their products. While the U.S. Food and Drug Administration (FDA) does not have formal guidelines directed at influencers or endorsers themselves, existing guidance for industry on fulfilling regulatory requirements for postmarketing submissions for drugs makes clear that FDA’s regulation of prescription drug promotion “extends to both promotional activities that are carried about by the firm itself, and to promotion conducted on the firm’s behalf.”⁶⁴ More evidence on the potential impact of user-generated and influencer marketing on tobacco use behaviors is therefore needed to inform future regulatory strategies, countermarketing campaigns, and prevention efforts.

This study has several limitations. First, our sample was limited to a relatively homogenous group of college students who were primarily white and female. Thus, the generalizability of our results may be limited. While college students remain an important target for tobacco research, future studies would benefit from the recruitment of a larger, more diverse sample with a greater proportion of tobacco users. Due to the small proportion of tobacco and e-cigarette users in our sample, we evaluated overall product use (both tobacco and e-cigarettes) as our outcome of interest; however, it would be valuable to examine the effects of message exposure on specific products such as hookah and e-cigarettes. Second, while the EMA approach reduces recall bias by shortening the overall time period over which participants have to remember message encounters, there is still the possibility that the messages reported do not accurately represent what participants actually experienced. Third, the EMA protocol specifically asked participants to estimate marketing exposure over the last 24 hours, and allowed participants to submit their response from 12 pm on one day to 12 pm on the following day. Although data quality procedures were implemented to examine potential overlap in reporting from one day to the next, the protocol would have been improved by either restricting the survey availability to a smaller window or specifying a narrower timeframe for marketing exposure recall. Finally, these data were collected in 2015, and the tobacco product market, as well as product advertising has changed since then, with the addition of new products like JUUL and the stark rise in e-cigarette use among youth and young adults. Thus, these results may not be generalizable to the current landscape. However, these results highlight the potential of this methodology to examine current advertising strategies such as social media influencers and online marketing.

Despite these limitations, our results demonstrate the utility of using EMA to measure the role of online messaging and social media in the changing landscape of tobacco marketing. While the presence of user-generated pro-tobacco messages on social media is well-documented, research has thus far been unable to quantify specific exposure, instead relying on content analyses or describing overall trends.^28,29 This type of broad-level analyses ignores some of the common social media venues such as Snapchat, Instagram, and Facebook, which have a substantial amount of content hidden behind friends-only privacy settings. With the use of EMA, it is possible to measure exposure and tobacco use behaviors at the individual level in real-time, thus allowing a more nuanced and clearer understanding of the many factors affecting tobacco product initiation and use, such as message location, source, content, and viewer characteristics.

• Young adults are exposed to pro-tobacco messaging from both the industry and peers.

• Exposure to tobacco marketing is associated with same-day product use.

• Both user-generated and industry-sponsored messages are associated with tobacco use.