Do tobacco industry websites target content to specific demographic groups?

Abstract

Background

Prior research has not examined whether tobacco brand websites vary content based on audience demographics. This study explored whether marketing content on tobacco brand websites varied by user ethnicity, gender or age group.

Methods

Participants (n = 32) were adult smokers, representing equal numbers of eight demographic groups: user ethnicity (Non-Hispanic White, Hispanic, African American, Asian), gender (women, men) and age (age 21–35, age 36+). This study examined 12 tobacco brand websites representing four tobacco product categories (cigarettes, cigar/cigarillos, smokeless tobacco, and e-cigarettes). From January 2016 to January 2017, participants coded websites for themes, interactive activities, and links to social media sites (n = 874 website visits). Logistic regression was used to analyze observed content by participant ethnicity, age and gender.

Results

All themes, all interactive activities and all links to social media were observed at least once for each demographic category. Male participants were more likely to observe Harm reduction themes, while female participants were more likely to observe Promotion themes. Older participants were more likely to observe website features allowing them to select music, and request coupons. Compared to Non-Hispanic White participants, African American participants were more likely to observe links to social media sites.

Conclusion.

Tobacco brand website content varied across ethnic, age and gender groups. These findings suggest that other factors, such as tobacco use behaviors, may influence marketing strategies participants recall or find appealing. The findings from this study can inform future regulatory activities and communication strategies aimed at countering pro-tobacco content online.

1.1. INTRODUCTION

Tobacco brand websites, and tobacco brand marketing on social media, remains largely unregulated by federal agencies in the United States (Bach, 2019). The lack of regulation provides an opportunity for tobacco companies to invest heavily in brand website content. It is of critical importance to determine if and how popular tobacco brand websites tailor their website content to consumers based on demographics. Such tailoring may contribute to tobacco-related behaviors and could be a point of regulatory targets in the future.

The leading cigarette companies reported spending $25 million on company websites in 2017, an increase from $2.3 million in 2007 (Federal Trade Commission, 2019a), while the leading smokeless tobacco companies reported spending $10.5 million on company websites in 2017, up from $3.1 million in 2007 (Federal Trade Commission, 2019b). Prior research demonstrated that websites for popular cigarette, cigar/cigarillo, e-cigarette and smokeless products were found to feature positive marketing themes in combination with interactive content, such as message boards or chat rooms, games, coupons, contests, music selections, and links to social media, helping to create online communities around the brand (Escobedo et al., 2018). Prior research also suggested that many popular e-cigarette brand websites contain weak age-verification gates, allowing visitors to enter the site with one click of a button (Escobedo et al., 2018).

For decades, direct to consumer marketing has allowed tobacco companies to transmit tailored messages through mail, email, and online media channels (Brock, Carlson, Moilanen, & Schillo, 2016; Lewis & Ling, 2015). Online tobacco marketing may be more effective in encouraging experimentation and deterring quit attempts than traditional marketing channels, as it allows users to engage and interact with pro-tobacco websites, games, social media content, online coupons, and other tobacco users (Calder, Malthouse & Schaedel, 2009; Ribisl, 2003; Soneji, Pierce, Choi et al., 2017). Engagement with online tobacco marketing was found to be associated with greater tobacco use initiation, more frequent tobacco use, poly-tobacco use and lower incidence of tobacco cessation among youth (Soneji, Yang, Knutzen et al., 2018). Websites with greater levels of interactivity can enhance information recall and message comprehensibility (Kim & Stout, 2010), while interactive features on tobacco brand websites have been designed to engage consumers and increase brand loyalty (Lewis, Yulis, Delnevo, & Hrywna, 2004). Furthermore, tailored messages and marketing strategies have been used by tobacco companies to target specific demographic groups, such as women, African Americans, younger consumers, and menthol smokers (Brown-Johnson, England, Glantz, & Ling, 2014; Lewis, Bover Manderski, & Delnevo, 2015; Richardson et al., 2015; Rose et al., 2018).

Tailoring messages to specific demographic groups is a type of audience segmentation, dividing a population into distinct subgroups, or segments, based on characteristics that influence a behavior or outcome of interest (Slater, 1996). In other words, segmenting strategies include the division of a population into homogenous subgroups based on demographics, geographic location, personal values, or behaviors to identify target audiences, and develop differential marketing strategies to increase the effectiveness and efficiency of communication efforts (Slater, 1996).

Tobacco use varies by demographics. For example, research has shown that African American and Latino smokers were less likely to report long term quitting success compared to Non-Hispanic Whites, and nearly half of African American ever-smokers reported current daily smoking, the highest daily use rate across all ethnic groups in the study (Trinidad, Pérez-Stable, White, Emery, & Messer, 2011). Tobacco industry documents have shown that African American populations were heavily targeted by culturally tailored menthol cigarette advertisements since the 1960’s, which led to significant increases in menthol use among African Americans over the next decade (Gardiner, 2004). These marketing campaigns have had a lasting impact on tobacco use as the 2014 National survey on Drug Use and Health found that 84.6% of African American smokers use menthol cigarettes, compared to 46.9% of Latinos and 28.9% of Non-Hispanic Whites (Villanti, Mowery, Delnevo et al., 2016). While past research has examined exposure to tobacco company websites by gender and ethnicity (Soneji, Ambrose, Lee, Sargent, & Tanski, 2014), research has not explored whether content or marketing themes used on tobacco brand websites were tailored to visitor demographics.

Research has shown that brand websites were the most common method of joining a direct mailing list among young adult smokers (Lewis, Manderski & Delnevo, 2015). Many popular tobacco brand websites require that visitors verify their identity online by providing their legal name, home address, date of birth and brand preferences (e.g. favorite product) in order to enter the website, ostensibly to reduce youth access. However, this age verification system allows tobacco companies to collect personal information that can be used to refine and tailor marketing efforts to individual consumers through direct marketing (Escobedo et al., 2018; Lewis, Manderski & Delnevo, 2015). Additionally, unlike cigarette and smokeless brand websites, e-cigarette brand websites have allowed visitors to access their content without online age verification (Escobedo et al., 2018; Soneji, Gerling, Yang, & Sargent, 2016). Such weak age verification gates have been shown to give youth greater access to pro-tobacco content, products on brand websites (Nikitin, Timberlake, & Williams, 2016; Ribisl, Lee, Henriksen, & Haladjian, 2003; Williams, Derrick, & Ribisl, 2015), and allow youth to create user profiles, potentially exposing youth visitors to direct marketing strategies.

Tobacco companies may be able to infer user demographics and transmit tailored messaging and website content using online behavioral advertising, a marketing practice that involves collecting information on one’s online activities to deliver customized advertisements (Boerman, Kruikemeier, & Zuiderveen Borgesius, 2017). Online behaviors include purchases, web browsing data, search history, use of online media (e.g., videos watched, articles read), social media activity, responses to click-through advertisements, and online communication (e.g. email messages) (Boerman et al., 2017). Based on web browsing patterns, companies have been able to infer visitors’ ages (Swami, Tarte, Rakshe, Raut, & Shaikh, 2015). By examining visitor location, surname and online behaviors, companies have been able to infer visitors’ ethnicities (Lange & Coen, 2016). There is even evidence that tobacco companies tailor their marketing to appeal to specific audiences using database marketing (Lewis & Ling, 2016), and that advertisement personalization has been found to be effective by increasing perceived relevance (Bleier & Eisenbeiss, 2015).

Prior research has not examined whether tobacco brand websites content varies based on audience demographics. The current study addresses this gap by examining whether website content varied by user demographics across 12 tobacco brand websites during a one-year period. This study will examine website content by participant ethnicity (Non-Hispanic White, African American, Asian American, Hispanic), age (21–35, 36+) and gender (male, female). Findings aim to inform regulatory activities and communication strategies aimed at countering pro-tobacco content in the future.

2.1. METHODS

Twelve tobacco brand websites representing four different tobacco product categories were included in this study. Websites were selected by 1) identifying the cigarette (Camel, Marlboro, Newport, Pall Mall), cigar/cigarillo (Swisher Sweets, Black & Mild) and smokeless tobacco (Copenhagen, Grizzly) brands with the largest percentage of total sales in the U.S. (Delnevo, Giovenco, Ambrose, Corey, & Conway, 2015; Miller Lo et al., 2017; Sharma, Fix, Delnevo, Cummings, & O’Connor, 2016), and by verifying that 2) each brand had an active brand website. E-cigarette brands with large market shares and active brand websites (Blu, Vuse, Njoy, V2) were also selected (Center for Disease Control and Prevention, 2016; Dickey, 2013).

2.2. Sampling & Participants

Participant coders (n=32) visited twelve tobacco brand websites from January 2016 to January 2017. To comply with CASRO Social Media Research Guidelines, and to avoid exposing nonusers to pro-tobacco marketing content, only adults 21 years of age or older who reported smoking cigarettes at least some days were recruited to code the tobacco websites (CASRO, 2011). Eligible participants needed to reside within the U.S. and have access to a computer or tablet. Participants were recruited using online advertisements placed on Craigslist and Facebook and recruitment fliers distributed at nearby universities. Adult smokers in two age groups (21–35 and 36+) were recruited with equal numbers of men and women from four ethnic groups (African American, Asian American, Hispanic, Non-Hispanic White) in each of those two age groups. The younger age group cut off was set at 35 to represent the end of early adulthood (Levinson, 1986). All participants provided informed consent and received a monthly incentive of $40. The University Institutional Review Board approved all procedures.

2.3. Procedure

Participants completed individual training sessions with the help of investigators using online conferencing software, and were instructed on how to navigate websites and answer survey questions while using the Webcoder application (Escobedo et al., 2018). Members of the research team explained survey items and assisted participants during practice coding sessions. Participants who discontinued the study, or were dismissed due to inactivity, were replaced by participants with the same demographic characteristics. Unique user profiles were created by all participants on websites requiring online registration.

The WebCoder application allowed participants to explore websites using a browser-like window with an interface displaying questions about website characteristics and themes. The WebCoder application was designed to only display the assigned tobacco brand websites. The assigned websites were rotated each month to ensure coders visited each of the 12 websites every three to four months. Participants were also required to log into the WebCoder application for a minimum of two hours each month and respond to survey questions while navigating through each website. The number of webpages per website varied, therefore participants were instructed to code at least ten different webpages per website visit. The Webcoder application captured screenshots of each webpage visited while coders used the application. Members of the research team reviewed responses from randomly selected participants to ensure they matched the website content captured by Webcoder application. Retraining was provided as needed. Among participants who coded at least once during the study period, about 68% (n = 22) completed monthly coding requirements for 12 of the 13 months of data collection. Among participants who coded at least once during the study period, about 40% (n = 13) of participants were dismissed due to inactivity. The turnover rate did not differ by demographic factors.

2.4. Measures

Participants filled out surveys on the Webcoder application to indicate whether they observed any of 53 marketing themes (e.g. adventure, flavors, sex appeal, health, technology) for each webpage within the site. Participants indicated whether they observed any of eight interactive activities: 1) chat rooms/message boards, 2) ability to “friend” other visitors, 3) create a user profile, 4) post pictures, 5) access to event information, 6) contests, 7) ability to request coupons [through email, mobile phone or printed from the website], or 8) the ability to select music, while navigating each website. Participants also indicated if the website provided links to one of five external social media sites (Facebook, Google+, Instagram, Twitter, YouTube) at any time during a monthly website visit. Factor scores, as well as interactive features and social media links, were dichotomized with a “1” indicating that the theme, interactive feature or social media link was observed by a participant at least once per website visit.

2.5. Analysis Plan

This analysis reports on 13 consecutive months of coding from January 1, 2016 to January 31, 2017. To improve interpretability of identified marketing themes, data from participant website visits (n = 874) were analyzed using principal axis factoring (PAF) extraction and promax rotation methods. Using the scree plot test, simple structure loading, and primary factor loading of .5 or above, a five-factor solution was preferred. The five factors included: 1) User experience (10 items: social, hip, pleasure, indulgence, cool, independence, adventure, fun, leisure, freedom, 11.72 [eigenvalue]), 2) Harm reduction (8 items: less harmful, safer for others around you, alternative to quitting, use in places you can’t smoke, discreet, reduce anxiety, sadness or depression, longer lasting, light/low tar, 4.18), 3) Socializing (7 items: party lifestyle, nightlife, music, holiday, celebrities, hashtag, 1.94), 4) Design (3 items: price, thin, technology/science, I can see smoke/vapor, 1.75), and 5) Promotions (2 items: discounts, other, 1.56).

The unit of analysis was the participant. Frequencies and percentages were calculated for observed themes, interactive features and social media links per demographic category. Each theme from the factor analysis, interactive feature, and social media link was regressed on participant demographic group (2 age groups, 2 gender groups, 4 racial/ethnic groups) in separate logistic regression models. Dichotomized data from all website visits (n = 874) were analyzed in each model. Non-Hispanic White (NHW), female, and the younger age group (21–35) served as the reference group in their respective demographic categories in all models. To reduce Type 1 errors, a Bonferroni correction was used to determine statistical significance; statistical significance was evaluated at the corrected p values e.g., themes, p < 0.01 (.05/5); interactive features, p < 0.006 (0.05/8), and social media links, p < 0.01 (.05/5). Statistical analysis was performed using SAS (version 9.4, SAS Institute Inc, Cary, North Carolina).

3.1. RESULTS

3.2. Participants

The average age of participants was 36.8 (SD = 10.11), and most participants reported smoking everyday (71%). All participants reported using the internet at least once a day, and nearly all participants (79%) visited one social media site per day. Participants spent an average of 81 minutes on each website, visited an average of 10 webpages per website, and spent an average of 10.1 minutes on each webpage.

3.3. Website Marketing Themes

All themes from the factor analysis were observed at least once for each demographic category during the study period. Compared to NHW participants, Hispanic participants were less likely to observe themes related to Harm Reduction (Table 1). Compared to NHW participants, Asian American participants were nearly twice as likely to observe themes related to Socializing (Table 1). Compared to NHW participants, African American and Asian American participants, were less likely to observe themes related to Promotions (Table 1). Compared to participants between the ages of 21–35, participants over the age of 36 were twice as likely to observe themes related to Promotions (Table 1). Compared to female participants, male participants were nearly twice as likely to observe themes related to Harm Reduction, however female participants were more likely to observe themes related to Promotions (Table 1).

Table 1.

Odds ratio and 95% confidence intervals by participant demographics representing the likelihood that themes from factor analysis were observed during website visits (N = 874).

		User Experience			Harm Reduction			Socializing			Design			Promotions
Group		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)
Ethnicity
White		28.06	-		27.05	-		27.29	-		28.39	-		35.12	-
AA		24.72	1.32 (0.88 – 1.99)		28.60	1.44 (0.99 – 2.10)		23.20	0.93 (0.64 – 1.36)		24.05	0.91 (0.62 – 1.34)		21.39	0.34* (0.23 – 0.52)
Asian		21.76	1.00 (0.64 – 1.55)		24.17	1.43 (0.96 – 2.11)		24.76	1.69* (1.13 – 2.53)		22.60	1.20 (0.80 – 1.81)		16.93	0.28* (0.18 – 0.44)
Hispanic		25.46	0.87 (0.57 – 1.33)		20.18	0.71* (0.49 – 1.03)		24.76	1.04 (0.72 – 1.50)		24.95	0.96 (0.66 – 1.40)		26.56	0.53 (0.35 – 0.81)
Age
21–35		46.85	-		48.34	-		45.03	-		45.21	-		40.29	-
36+		53.15	0.92 (0.67 – 1.25)		51.66	0.86 (0.65 – 1.14)		54.97	1.14 (0.86 – 1.50)		54.79	1.14 (0.86 – 1.51)		59.71	2.01* (1.50 – 2.70)
Gender
Female		51.05	-		44.12	-		48.54	-		48.82	-		57.04	-
Male		48.95	1.36 (1.08 – 1.85)		55.88	1.90* (1.45 – 2.50)		51.46	1.36 (1.04 – 1.79)		51.18	1.40 (1.06 – 1.85)		42.96	0.55* (0.41 – 0.74)

Note: AA = African American, Non-Hispanic White, female, and age 21–35 served as reference groups,

^*p< 0.01 with Bonferroni correction

3.4. Interactive Activities

All interactive activities were observed at least once for each demographic category during the study period. Compared to NHW participants, African American participants were more likely to observe website features allowing them to create user profiles, and website features allowing them to friend others, but were less likely to observe event information, and coupon information (Table 2). Compared to NHW participants, Asian American participants were more likely to observe event information but were less likely to observe website features allowing them to create user profiles, website features allowing them to friend others, and website features allowing them to post pictures (Table 2). Compared to NHW participants, Hispanic participants were less likely to observe website features allowing them to create user profiles (Table 2).

Table 2.

Odds ratio and 95% confidence intervals by participant demographics representing the likelihood that interactive website features were observed during website visits (N = 874).

		User profile			Friend others			Chat room/message boards			Post pictures
Group		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)
Ethnicity
White		30.30	-		29.85	-		27.27	-		28.10	-
AA		32.66	1.59* (1.08 – 2.33)		56.72	2.16* (1.19 – 3.92)		29.87	1.32 (0.88 – 1.99)		30.95	0.66 (0.41 – 1.05)
Asian		17.51	0.69* (0.45 – 1.06)		4.48	0.15* (0.04 – 0.54)		20.78	1.00 (0.64 – 1.55)		15.24	0.35* (0.19 – 0.62)
Hispanic		19.53	0.65* (0.43 – 0.97)		8.96	0.29 (0.11 – 0.75)		22.08	0.87 (0.57 – 1.33)		25.71	0.78 (0.50 – 1.21)
Age
21–35		37.71	-		59.70	-		48.48	-		41.03	-
36+		62.29	1.85* (1.38 – 2.49)		40.30	0.54* (0.31 – 0.94)		51.52	0.92 (0.67 −1.25)		58.97	1.28 (0.89 – 1.83)
Gender
Female		46.46	-		29.85	-		46.32	-		46.67	-
Male		53.54	1.37* (1.03 – 1.83)		70.15	3.21* (1.82 – 5.65)		53.68	1.36 (1.00 – 1.85)		53.33	1.00 (0.70 – 1.42)
		Events			Select music			Contests			Request Coupons
Group		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)
Ethnicity
White		28.24	-		28.71	-		33.46	-		33.49	-
AA		17.18	0.62* (0.40 – 0.95)		27.72	1.25 (0.71 – 2.21)		21.19	0.59 (0.40 – 0.87)		17.34	0.40* (0.27 – 0.59)
Asian		26.72	1.42* (0.94 – 2.13)		13.86	0.66 (0.33 – 1.29)		20.45	0.67 (0.45 – 0.99)		23.04	0.86 (0.59 – 1.27)
Hispanic		28.24	1.18 (0.79 – 1.75)		29.70	1.17 (0.67 – 2.04)		24.91	0.69 (0.48 – 1.00)		26.13	0.75 (0.52 – 1.08)
Age
21–35		42.75	-		34.65	-		41.64	-		41.33	-
36+		57.25	1.17 (0.87 – 1.58)		65.35	1.90* (1.22 – 2.96)		58.36	1.20 (0.91 – 1.59		58.67	1.39* (1.06 – 1.83)
Gender
Female		44.66	-		66.34	-		47.58	-		52.97	-
Male		55.34	1.51* (1.12 – 2.03)		33.66	0.50* (0.32 – 0.78)		52.42	1.10 (0.84 – 1.45)		47.03	0.91 (0.69 – 1.83)

Note: AA = African American, Non-Hispanic White, female, and age 21–35 served as reference groups,

^*p< 0.006 based on Bonferroni correction

Compared to younger participants, older participants were more likely to observe website features allowing them to create user profiles, website features allowing them to select music, and coupon information, but were less likely to observe website features allowing them to friend others (Table 2). Compared to female participants, male participants were more likely to observe website features allowing them to create user profiles, website features allowing them to friend others and event information but were less likely to observe website features allowing them to select music (Table 2).

3.5. Social Media

All external links to social media were observed at least once for each demographic category during the study period. Compared to NHW participants, African American participants were more likely to observe links to all five social media sites on tobacco brand websites (Table 3). Compared to NHW participants, Hispanic participants were less likely to observe links to Facebook, Instagram, Twitter and YouTube (Table 3). Compared to younger participants, older participants were less likely to observe links to Facebook, Google+ and YouTube (Table 3.) Compared to female participants, male participants were more likely to observe links to Google+ (Table 3).

Table 3.

Odds ratio and 95% confidence intervals by participant demographics representing the likelihood that links to external social media sites were observed during website visits (N = 874).

		Facebook			Google+			Instagram			Twitter			YouTube
Group		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)		Yes (%)	OR (95% CI)
Ethnicity
White		24.79	-		20.62	-		26.09	-		28.06	-		30.22	-
AA		35.10	2.22* (1.52 – 3.24)		40.21	2.66* (1.71 – 4.15)		32.61	1.54* (1.02 – 2.32)		32.90	1.54* (1.05 – 2.24)		30.67	1.14* (0.76 – 1.70)
Asian		18.66	0.98 (0.65 – 1.46)		15.98	0.98 (0.58 – 1.64)		21.30	1.10 (0.71 – 1.72)		18.06	0.79 (0.53 – 1.20)		20.44	0.85 (0.54 – 1.31)
Hispanic		21.45	0.90* (0.61 – 1.32)		16.13	1.28 (0.79 – 2.06)		20.00	0.79* (0.51 – 1.23)		20.97	0.74* (0.50 – 1.10)		18.67	0.59* (0.38 – 0.92)
Age
21–35		54.87	-		56.70	-		52.61	-		51.29	-		52.89	-
36+		45.13	0.61* (0.46 – 0.80)		43.30	0.64* (0.46 – 0.89)		47.39	0.75 (0.55 – 1.02)		48.71	0.79 (0.59 – 1.05)		47.11	0.71* (0.52 – 0.97)
Gender
Female		50.70	-		44.33	-		50.43	-		52.58	-		48.89	-
Male		49.30	1.12 (0.85 – 1.48)		55.67	1.57* (1.13 – 2.19)		49.57	1.09 (0.80 – 1.48)		47.42	0.97 (0.73 – 1.29)		51.11	1.19 (0.88 – 1.62)

Note: AA = African American, Non-Hispanic White, female, and age 21–35 served as reference groups,

^*p< 0.01 with Bonferroni correction

4.1. DISCUSSION

The present study examined how marketing content across 12 tobacco brand websites varied by visitor demographics. The likelihood of observing various marketing themes, interactive activities, and links to external social media sites on tobacco websites varied by participant ethnicity, age, and gender groups. Understanding how website visitors from diverse backgrounds experience pro-tobacco website content is an important public health priority given that engagement with online tobacco marketing has been associated with greater tobacco use and fewer cessation attempts (Soneji, Yang, Knutzen et al., 2018).

In line with previous research (Lewis, Manderski & Delnevo, 2015), this study found that female participants were more likely than male participants to observe website themes related to Promotions. However, in contrast to prior research (Brown-Johnson, England, Glantz, & Ling, 2014; Lewis, Bover Manderski, & Delnevo, 2015; Richardson et al., 2015; Rose et al., 2018) this study found that NHW participants and older participants, were more likely to observe marketing themes focusing on Promotions compared to African American and Asian American participants, and younger participants, respectively. These discrepancies may be explained, in part, by tobacco company market segmentation strategies. The tobacco industry has been known to segment audiences by consumer behaviors (e.g. users, non-users, purchasing behaviors, product or brand preferences), attitudes about smoking (e.g. health concerns, social acceptability, motivations for use, price sensitivity), lifestyle (e.g. leisure activities, occupation, media use), and personality (e.g. goals, political views, values) (Ling & Glantz, 2002).

Tobacco company marketing messages have been shown to be tailored to different market segments based on behaviors and needs, which can vary within demographic groups (Ling & Glantz, 2002). For some consumers, product use is associated with freedom and independence, while for others, use is associated with social interaction and belonging, demonstrating how smoking can be marketed using seemingly antithetical messages (Ling & Glantz, 2002). The type of marketing message has also been shown to vary by tobacco brand, depending on the intended market segment (Ling & Glantz, 2002). For example, using emotional (experiential) versus rational (utilitarian) marketing messages is often driven by market segment behaviors and motivations for use (Albers-Miller & Royce, 1999). The findings from this study highlight the importance of including factors such as consumer behaviors, attitudes about smoking, and personality along with demographic traits when designing interventions to reduce tobacco use, and tobacco control policies (Ling & Glantz, 2002).

In line with prior research (Escobedo et al., 2018), the current study found that tobacco brand websites engage users by allowing visitors to create user profiles, friend other users, join chat rooms, post photos, select music, and request coupons. This is concerning as websites with higher levels of interactivity were associated with greater levels of message comprehensibility and information recall, which may be the result of users engaging in more active information processing (Kim & Stout, 2010). The current study also found that observations of interactive activities varied across ethnicity, gender and age. Studies by the Pew Research Center show that the level of internet use by ethnicity and gender is similar (Pew Research Center, 2017), however the reasons for internet use and how individuals use website features can differ. Earlier research found that when men and women visit popular brand websites, they preferred to use different types of interactive features, and time spent on interactive features differed (McMahan, Hovland & McMillan, 2009). In addition, use behaviors, personal attitudes about use, and lifestyle may influence the type of interactive activities observed, as a more price sensitive tobacco user may seek out interactive coupons and contests, while other visitors use the website to discuss product features. There is a lack of research focusing on how different groups perceive the appeal and effectiveness of interactive activities on tobacco brand websites. Future research should examine the appeal and effectiveness of interactive features on tobacco brand websites among vulnerable groups defined by demographic characteristics, behaviors, and lifestyle. The number of interactive features on tobacco brand websites may be an important regulatory target in the future. Local and state tobacco control agencies may benefit from increasing the level of interactivity of their own websites to promote engagement with tobacco prevention material.

In line with prior research (Escobedo et al., 2018), results from this study indicated that tobacco brand websites feature links to popular social media sites like Facebook, Instagram and Twitter, where tobacco industry sponsored coethntent can be accessed, often without age-verification. While use of social media is similar across ethnic groups (Pew Research Center, 2019) the reason for social media use has been reported to differ by demographic groups. Recent research found that half of African American respondents (compared to only a third of NHW respondents) reported that social media was an important platform for becoming familiar with social issues (Pew Research, 2018). In this study, African American participants were more likely to observe links to all five social media sites compared to NHW participants. Taken all together, social media may be an important platform for reaching African American populations and should be used by public health and tobacco control programmers to disseminate tobacco prevention material. Additionally, participants in this study aged 21 to 35 were most likely to observe nearly all social media sites, corroborating earlier findings that younger adults (compared to older adults) report greater overall social media use (Pew Research Center, 2018). This is concerning as Escobedo et al (2018) found that tobacco brand websites encourage visitors to disseminate promotional messages through social media, bypassing age verification and regulatory systems. Future research should examine which users click on social media links embedded on tobacco brand websites, and who is more likely to share (retweet or repost) tobacco company marketing material on their own social media platforms. Understanding how users disseminate tobacco marketing messages is important, as tobacco use exposure on social media has been associated with subsequent smoking (Depue, Southwell, Betzner, 2015; Unger et al., 2018).

4.2. Limitations

While this study examined 12 of the leading tobacco brand websites, findings may not generalize to all brands or companies. The selection criterion of e-cigarette brands for study inclusion was based on sales data prior to 2016, therefore e-cigarette brands that currently have a larger share of market sales (e.g. Juul) were not included. Data collection relied on self-reported measures. This study could not determine if marketing themes were tailored to different demographics or if certain themes were more appealing to specific demographic groups and is an area of future research. This study focused on group differences and did not track individual differences over time. Participants in this study were only asked about cigarette use during the recruitment interview. As a result, we do not know about participants’ poly-tobacco use.

4.3. Conclusions

The observed marketing themes, interactive activities and social media links on tobacco websites varied across ethnic, age and gender groups. These findings and tobacco company documents suggest that other factors, such as tobacco use behaviors and lifestyle, may influence the kind of marketing strategies participants recall or find appealing. Furthermore, tobacco brand websites direct visitors to external social media sites, providing additional opportunities for dissemination of positive messages about tobacco products, often without age restrictions. The restriction or prohibition of tobacco brand websites and pro-tobacco advertising online as a part of comprehensive tobacco control strategies may reduce the burden of tobacco-related diseases.

HIGHLIGHTS.

• Interactive activities and social media links were observed by all demographic categories

• Male participants (compared to females) were more likely to observe Harm reduction themes

• Female participants (compared to males) were more likely to observe Promotion themes

• Older participants (compared to younger) were more likely to observe website coupons

• African-American participants (compared to Whites) were more likely to observe social media content