Young Adults’ Visual Attention to, Unaided Recall, and Perspectives of Tobacco Public Education Messages: An Experimental Eye-Tracking Study

Abstract

Introduction

Mass-reach tobacco public education campaigns may need to be optimized for marginalized populations, including lesbian, gay, bisexual, transgender, and queer (LGBTQ+) young adults (YA). We examined attention to and unaided recall of culturally targeted (CT) tobacco education among LGBTQ+ YA in an experimental trial.

Aims and Methods

LGBTQ+ YA reporting current nicotine or tobacco (N/T) use completed a baseline survey, eye-tracking experiment (viewing 8 CT or nontargeted stimuli), and a postsurvey. Areas of interest (AOIs) were harms message, efficacy message, and image. We compared dwell time (DT) and fixation duration (FD) to AOIs by condition and modeled associations between DT/FD and message relevance and perceived message effectiveness (PME). One week postexposure, participants completed open-ended items assessing unaided recall (memory of stimuli). We coded responses using a data-driven inductive approach.

Results

There was no difference in DT/FD to AOIs by condition. Attention to harms AOIs was positively associated with message relevance (DT: Beta = 0.11; 95% CI = 0.05% to 0.17%; FD: Beta = 0.38; 95% CI = 0.18% to 0.58%) and PME (DT: Beta = 0.09; 95% CI = 0.04% to 0.15%). Attention to image AOIs was associated with increased message relevance. Participants across conditions positively recalled sociodemographic diversity in images. Experimental participants resonated with LGBTQ+ representation, while control participants recalled racial and age diversity. Participants recalled messages highlighting the tobacco industry targeting and long-term health harms. Participants expressed interest in quitting or reducing N/T use.

Conclusions

LGBTQ+ YA find CT tobacco public education memorable; however, nontargeted campaigns featuring diverse models are also effective at fostering visual attention and recall.

Implications

This study assessed how cultural targeting affects visual attention, perceptions, and unaided recall of tobacco public education among LGBTQ+ YA who use nicotine and tobacco. Regardless of condition, LGBTQ+ YA recalled stimuli featuring diverse visual representation (eg, sexual orientation, race) and messages about the tobacco industry’s manipulation of minoritized communities. Visual attention to imagery and harms messages increased message relevance and PME (harms messages only). Most participants qualitatively expressed intentions to quit or reduce use post-exposure. Tobacco public education integrating CT content, representative models, and counter-industry marketing may engage LGBTQ+ YA and motivate behavior change in this population.

Introduction

Electronic nicotine vaping and combustible cigarette use (henceforth, “N/T use”) is a significant public health concern, specifically among lesbian, gay, bisexual, transgender, and queer (LGBTQ+) young adults (YA) (terminology defined in Table S1). Between 22% and 40% of LGBTQ+ YA report dual N/T use within the past 30 days, compared to 12%–21% of non-LGBTQ+ YA.^1–3 These disparities result from social, environmental, and psychological factors.^4,5 Many LGBTQ+ individuals use N/T to cope with minority stressors (eg, homophobia, discrimination).^4,5 Moreover, the tobacco industry has normalized tobacco use with targeted advertising in LGBTQ+ spaces.⁶ Few culturally specific tobacco interventions address the experiences of this population.^7,8

Tobacco public education campaigns can influence tobacco use knowledge, attitudes, and behaviors, helping to shape public perceptions and encourage smoking cessation.^9–11 Cultural targeting (CT) may increase LGBTQ+ populations’ engagement with tobacco public education by aligning visual and thematic content with their cultural needs and preferences.^7,12 Qualitative studies suggest LGBTQ+ YA prefer antitobacco interventions that are LGBTQ+-specific, inclusive, and relatable.¹³ In prior research, LGBTQ+ YA who engaged in dual N/T use described the importance of developing “authentic” CT tobacco public education campaigns by using personal stories and gain-framed messaging that promotes tobacco cessation and harm reduction.¹⁴ Campaigns like “Every Try Counts” (ETC)¹⁵ have attempted to engage LGBTQ+ people who use tobacco by featuring LGBTQ+-specific materials, but no evaluation studies have been published.¹⁶

Studies on the effectiveness of LGBTQ+ YA CT versus nontargeted tobacco public education demonstrate mixed results.^17–20 In a single-exposure experimental study testing the efficacy of Truth© CT counter-industry advertisements among LGBTQ+ adults, CT advertisements were not more effective than nontargeted.¹⁹ However, Patterson et al. found that among YA sexual minority women, a single exposure to CT vaping education messages that integrated content on health harms effectively discouraged vaping.⁸ Similarly, Tan and colleagues found that repeated exposure to anti-smoking advertisements increased intentions to quit among YA sexual minority women who smoke cigarettes, but outcomes did not differ between CT and nontargeted conditions.²⁰ Further analysis of this data indicated that LGBTQ+ CT indirectly mediated quit intentions by influencing antitobacco industry beliefs.²¹ These findings suggest that CT messages show promise for influencing attitudes and behavioral intentions; however, content may need to be optimized to increase effectiveness.

CT may increase LGBTQ+ YA’s attention and engagement with tobacco public education by enhancing message relevancy; accordingly, understanding how LGBTQ+ YA attend to CT messages may aid content optimization. Theoretically, the Message Impact Framework posits that health messages must attract viewer attention and field a reaction to influence behaviors.²² Eye tracking is a technology that measures eye movements to analyze visual attention to stimuli.²³ Eye-tracking studies have objectively assessed visual attention to tobacco advertisements and packages^23,24 and determined features associated with message effectiveness and behaviors. Less research has applied eye-tracking to tobacco public education.²⁵ Jarman and colleagues found that adding thematic content about quitting to tobacco informational content and imagery increased recall of the quit line number among adults, suggesting that combining these elements could optimize effectiveness.²⁶ Yet Kim et al. found that visual attention to thematic content in the ETC campaign was not associated with effective message responses among YA.²⁷ No published studies have examined visual attention to LGBTQ+ CT tobacco public education.

Unaided message recall (a person’s ability to remember a stimulus without prompting) is an indicator of attention.^28,29 Unaided recall, which measures how well thematic or visual content is embedded in a viewer’s mind, is differentially associated with thematic and visual content. For example, in a study of adults who smoke, participants more often recalled tobacco public education using testimonials or graphic imagery and rated these more effective than stimuli describing quit strategies.³⁰ No published studies on LGBTQ+ CT tobacco public education have evaluated message recall. This is a critical omission as tobacco public education recall is associated with negative perceptions of tobacco products and their health consequences,^31,32 which are antecedents to behavior change.³³ To fill these gaps, our study pairs quantitative and qualitative measures of attention (ie, eye tracking and unaided recall) with measures of message effectiveness to examine what visual and thematic content attracts attention and how these features may influence LGBTQ+ YA’s attitudes. Our goal is to optimize tobacco public education for LGBTQ+ YA.

This analysis aimed to (1) quantitatively evaluate the effects of a single exposure to CT versus nontargeted tobacco public education stimuli on visual attention (assessed via eye-tracking) and self-reported message effectiveness, and (2) qualitatively evaluate unaided recall of visual and thematic content and motivations to reduce or quit N/T among LGBTQ+ YA susceptible to dual N/T use.

Materials and Methods

Study Design

This study was approved by The Ohio State University Institutional Review Board.

Participants

We used criterion-I sampling, a research technique to identify individuals with knowledge and experience of a specific topic.³⁴ We identified LGBTQ+ YA who used N/T and may be targeted by CT tobacco public education. Eligible participants (1) were 18–35 years old; (2) spoke English fluently; (3) ever used nicotine vapes and combustible cigarettes; currently used at least one product; (4) self-identified as LGBTQ+; (5) resided in the United States; (6) could access a computer with a web-camera; and (7) had no preclusions to eye-tracking (ie, glaucoma, cataracts). We recruited participants via university student organizations; flyers were distributed to campus and local businesses in Columbus, Ohio, and social media advertising via Commando, an LGBTQ+ specialized marketing agency. We posted online advertisements on Instagram, Facebook, and Sniffies, and used geo-coding to recruit in cities with high smoking prevalence³⁵ and high density of LGBTQ+ adults.³⁶

Procedures

We collected data between June 2023 and February 2024. Potential participants completed an online screener via REDCap. Post-consent, eligible participants completed a baseline survey, an eye-tracking experiment, and a post-survey. We used iMotion’s remote eye-tracking program that measured participants’ eye movements via webcam.³⁷

We randomized participants into two conditions: control arm participants viewed eight nontargeted tobacco education stimuli and experimental arm participants viewed eight LGBTQ+ CT tobacco education stimuli (Figure 1; Figures S1–S16). Stimuli were randomized and participants viewed each stimulus for 10 s. To re-center eyes between stimuli, participants viewed “X” centered on-screen for 10 s. After viewing all stimuli, participants completed self-report survey questions.

Figure 1.

Example of nontargeted (left image) and cultural targeting (right image) stimuli.

One week post-experiment, we invited participants to a follow-up survey, which qualitatively assessed recall. Participants received a $50 online gift card for completing the baseline survey, eye-tracking activity, and post-survey, and a $20 online gift card for completing the one-week follow-up survey.

Stimuli

The research team developed the stimuli shown during the eye-tracking experiment based on four formative evaluation studies (focus groups, online rating survey, in-depth interviews, online factorial experiment). We conducted additional focus groups to determine thematic strategies and imagery for acceptable LGBTQ+ CT tobacco education.¹⁴ We developed nine CT and nontargeted stimuli and shared them with a scientific panel, including LGBTQ+ YA and experts in communications and LGBTQ+ health. Based on feedback, we adjusted visuals and thematic content, resulting in 8 stimuli per condition.

Measures

Survey

Demographic measures included age, race, ethnicity, gender, sexual orientation, education, smoking and vaping status, and quit history. We asked participants to choose a gender identity and sexual orientation that best represented them. Those endorsing a minoritized gender (agender, gender queer, non-binary, transgender woman, or transgender man) or sexual orientation (asexual or ace, bisexual, gay or lesbian, pansexual, queer) were defined as LGBTQ+.

Perceived message effectiveness (PME) (ie, beliefs that a stimulus will influence smoking and vaping attitudes and behavior) was assessed via six items adapted from a validated scale.³⁸ Participants also reported perceived message relevance.³⁹ Measures detailed in Table S2.

Eye-Tracking

For each stimulus, we defined areas of interest (AOIs): harms message, efficacy message, and image (see Figure S17). Fixation duration (FD) (ie, average duration of brief stationary visits within an AOI) and dwell time (DT) (as a measure of the depth of cognitive processing) in seconds were analyzed.⁴⁰

Qualitative

Participants were asked three open-ended questions assessing recall of thematic content and imagery: “What was the most memorable part of the images?,” “The images that you viewed included health messages. What do you remember about the health message?,” and “What do you remember most about the images?” We also asked participants if stimuli made them quit, switch, or rethink their N/T use.

Analysis

Quantitative

We used descriptive statistics to describe sample characteristics and visual attention to AOIs (StataSE 18.0). Our goals were to (1) compare DT and FD on AOIs between conditions, (2) evaluate associations between visual attention on AOIs and PME and relevance, and (3) assess if condition modified associations. As demographic covariates were balanced across conditions, we ran post-exposure linear regression models with robust standard errors to assess the main effect of condition on visual attention to AOIs via a generalized estimating equation approach. To account for minor variances in AOI size between conditions, models were adjusted for relative AOI size in squared pixels. We ran separate linear regression models to assess the main effects of visual attention to AOIs on self-reported outcomes. Type 1 error was controlled using the Holm–Bonferroni method.⁴¹ We evaluated whether condition modified the effect of visual attention on outcomes by adding condition × DT or condition × FD interaction terms to respective models.⁴² We used an alpha of 0.05 to identify statistical significance of interactions; as none met this threshold, we report main effects models without interactions.

Qualitative

The research team reviewed all data to identify descriptive categories (parent codes) pertinent to the study aims. We applied template coding (Microsoft 365 Excel). Analysis followed a data-driven inductive approach⁴³ whereby two student researchers (MS, AM) independently coded ~5% of participants’ responses to identify descriptive and interpretive subcodes (child codes). A senior researcher (AE) facilitated consensus meetings to review coding and rectify disagreements. Upon consensus, we added parent and child codes to the coding scheme. Definitions were revised (Table S3) and we iteratively coded data per the new codebook (Table 1). Reliability at initial and final coding indicated strong agreement (Cohen’s kappa = 0.83 and 0.86, respectively; StataSE 18.0). The research team reviewed coded data to identify thematic patterns and exemplar quotations.

Table 1.

Codebook frequencies

Parent code	Child code	Total N = 117	Experimental N = 62	Control N = 55
Imagery		n (%)
	Models in images	70 (59.8)	44 (69.8)	26 (47.3)
	LGBTQ+ folks	29 (24.8)	28 (44.4)	1 (1.8)
	Image + content dissonance	3 (0.03)	2 (3.17)	1 (1.82)
	Colors	18 (15.4)	15 (23.8)	3 (5.5)
	Font	14 (12.0)	9 (9.5)	5 (9.1)
Diversity
	Gender	14 (12.0)	13 (20.6)	1 (1.82)
	Sexual orientation	12 (10.3)	11 (17.5)	1 (1.82)
	General diversity	10 (8.5)	7 (11.1)	3 (5.5)
	Age	10 (8.5)	5 (7.9)	5 (9.1)
	Race/skin tone	12 (10.2)	7 (11.1)	5 (9.1)
	Relatability	30 (25.6)	15 (23.8)	15 (27.3)
Health impacts
	Mental health	6 (5.1)	3 (4.8)	3 (5.5)
	Heart disease	21 (18.0)	9 (14.3)	12 (21.8)
	Cancer	16 (13.7)	8 (12.7)	8 (14.6)
	Lung disease	17 (14.5)	11 (17.5)	6 (10.9)
	Sleep	14 (12.0)	7 (11.1)	7 (12.7)
	Secondhand smoke	10 (8.5)	6 (9.5)	4 (7.3)
	Addiction	9 (7.7)	6 (9.5)	3 (5.5)
	High blood pressure	8 (6.8)	4 (6.4)	4 (7.3)
	Vaping = as bad as smoking	6 (5.1)	2 (3.17)	4 (7.3)
	Vaping = more harmful than smoking	3 (2.6)	0 (0.0)	3 (5.5)
	Vaping = less harmful than smoking	3 (2.6)	2 (3.17)	1 (1.8)
Social justice
	Disparities	20 (17.1)	14 (22.2)	6 (10.9)
	Big Tobacco	20 (17.1)	14 (22.2)	6 (10.9)
Relationships
	Community	24 (20.5)	18 (28.6)	6 (1.8)
	Couples	11 (9.4)	11 (17.5)	0 (0.0)
	Friends	9 (7.7)	7 (11.1)	2 (3.6)
Negative
	Stereotypical	4 (3.4)	2 (3.2)	2 (3.6)
	Fake	8 (6.8)	4 (6.4)	4 (7.3)
	Not new info	12 (10.3)	6 (9.5)	6 (10.9)
	Not memorable	32 (27.4)	20 (31.8)	12 (21.8)
Quitting + quit methods
	Yes, quitting	9 (7.7)	3 (4.8)	6 (10.9)
	No, not quitting	45 (38.5)	26 (41.27)	19 (34.55)
	Considering or planning to quit	59 (50.4)	34 (54.0)	25 (45.5)
	Switch to vaping	2 (0.2)	0 (0.0)	2 (3.6)
	Reducing use	20 (17.1)	11 (17.5)	9 (16.4)
	Social smoking	4 (3.4)	1 (1.6)	3 (5.5)
	Adjusting use, not because of study	7 (5.9)	2 (3.2)	5 (9.1)
	Shame/guilt	8 (6.8)	3 (4.8)	5 (9.1)
	Nonnicotine vapes	2 (1.7)	2 (3.2)	0 (0.0)
	Nicotine replacement product	2 (1.7)	2 (3.2)	0 (0.0)

Results

Sociodemographic Characteristics

Of 462 eligible participants, 321 consented to participate, and 160 completed follow-up. We excluded n = 24 participants who delayed completion of the 1-week follow-up and n = 19 participants with responses indicating fraudulent activity; specifically, qualitative responses that were mismatched (eg, participants in nontargeted conditions describing rainbow imagery), nonconforming (ie, unusual or implausible), or duplicative (ie, nearly identical sentence structure, word pairs). Final analysis included N = 117 participants (n = 55 control, n = 62 experimental; Figure S18).

Participants were mostly White (59.8%) and non-Hispanic/Latinx (88.9%); however, 26.5% identified as Black or African American. Most were aged 18 to 25 years (53.8%) and had a bachelor’s degree (42.7%) or some college education (30.8%). Participants were mostly cisgender women (49.6%) and gay or lesbian (42.7%). At baseline, most reported current dual N/T use (58.1%). Over half attempted to quit smoking or vaping within the past 12 months (69.6% and 62.3%, respectively), but fewer intended to quit smoking or vaping within 30 days (25.3% and 33.9%, respectively) (Table S4).

Eye-Tracking

See Table 2 for eye-tracking metrics.

Table 2.

Associations between Eye-Tracking Metrics by Area of Interest and Self-Reported Outcomes

	Beta (95% CI)	p
PME—CIGARETTE SMOKING
Dwell time (seconds)
Harms AOI	0.09 (0.04, 0.15)	.001
Efficacy AOI	0.04 (−0.01, 0.09)	.206
Image AOI	−0.04 (−0.09, 0.01)	.084
Fixation duration (seconds)
Harms AOI	0.12 (−0.08, 0.32)	.230
Efficacy AOI	−0.001 (−0.14, 0.13)	.979
Image AOI	0.13 (−0.09, 0.36)	.250
PME—NICOTINE VAPING
Dwell time (seconds)
Harms AOI	0.09 (0.03, 0.15)	.005
Efficacy AOI	0.05 (−0.005, 0.11)	.072
Image AOI	0.01 (−0.04, 0.06)	.752
Fixation duration (seconds)
Harms AOI	0.29 (0.08, 0.50)	.007
Efficacy AOI	0.16 (0.01, 0.30)	.034
Image AOI	0.46 (0.22, 0.70)	<.001
RELEVANCE
Dwell time (seconds)
Harms AOI	0.11 (0.05, 0.17)	<.001
Efficacy AOI	0.01 (−0.04, 0.07)	.761
Image AOI	0.03 (−0.01, 0.08)	.166
Fixation duration (seconds)
Harms AOI	0.38 (0.18, 0.58)	<.001
Efficacy AOI	0.20 (0.06, 0.34)	.006
Image AOI	0.48 (0.25, 0.71)	<.001

95% CI = 95% confidence interval; PME = perceived message effectiveness; AOI = area of interest.

The Holm–Bonferroni method was used to control familywise error, accounting for multiple hypothesis testing. Bolded findings represent statistically significant results.

Visual Attention to Message Features by Condition

There were no statistically significant differences in visual attention to AOIs by condition (Figure 2). Mean DTs ranged between 1.65 s and 1.87 s on textual messages and 2.65 s to 2.69 s on images across conditions. Mean fixation time was 0.30 to 0.44 s on textual messages and 0.39 to 0.41 s on images.

Figure 2.

Visual attention to areas of interest, by condition.

Associations between Visual Attention to AOIs and PME

Regardless of condition, DT on the harms AOI was positively associated with PME for cigarette smoking (Beta = 0.09; 95% CI = 0.04% to 0.15%). PME for cigarette smoking was not associated with DT on the efficacy nor image AOIs, nor with fixation time on any AOI.

DT on AOIs was not associated with PME for nicotine vaping. Across conditions, fixation time on the image AOI was positively associated with PME for nicotine vaping (Beta = 0.46; 95% CI = 0.22% to 0.70%).

Associations between Visual Attention to AOIs and Message Relevance

Regardless of condition, DT (Beta = 0.11; 95% CI = 0.05% to 0.17%) and FD (Beta = 0.38; 95% CI = 0.18% to 0.58%) on the harms AOI were positively associated with message relevance. FD on the image AOI was also positively associated with message relevance (Beta = 0.48; 95% CI = 0.25% to 0.71%) across conditions.

Recall of Visual Content

See Table 1 for codebook frequencies and Table S5 for more exemplar quotations.

Models in Images

Participants across conditions found the featured models memorable and relatable due to their perceived sociodemographic identities, including sexual orientation, gender identity, age, and race.

Sexual Orientation and Gender Identity

“The most memorable part of the images is that they all showed photos of LGBT+ individuals. I am trans, and I felt the message had an impact because it also showed trans individuals” (Transgender woman, queer, 26–35, CT condition).

LGBTQ+ status was signaled to experimental participants through inclusion of pronouns, model-specific visual elements (eg, fashion), and perceived relationships between models: “I remember there being pretty diverse representation in the images, including a variety of queer relationships aside from just same sex cis couples” (Cisgender man, gay, 26–35, CT condition). Some participants questioned using stock models: “Stock image queer folk are annoying and unmemorable” (Cisgender man, gay, 18–25, CT condition).

General Diversity

“The most memorable image had an African American woman in it. I saw myself in her. It was relatable” (Cisgender woman, bisexual, 26-35, control condition).

While control participants also recalled models’ perceived identities as memorable, their observations focused on general sociodemographic diversity. Race was a salient factor for participants identifying as Black, Indigenous, and People of Color (BIPOC+). Although we did not intentionally target control stimuli toward LGBTQ+ participants, some found sexual orientation and gender identity salient. One participant recalled “…different representations of individuals. Whether it was people of color, LGBTQIA+ individuals, or different gender representations” (Cisgender man, bisexual, 18–25, control condition). Nonetheless, a few control participants found the visual stimuli inauthentic: “The attempts at making the actors look ‘like me’ and relatable…comes off funny.” (Cisgender man, pansexual, 26–35, control condition).

Like control participants, experimental participants discussed the age and race of models. Some emphasized the difference between traditional antitobacco interventions and study stimuli:

“I still think about the people in the images and how they represent more than just older white men and women. I feel like a lot of anti-smoking ads are older people struggling and can be hard to relate to, but it did strike a nerve seeing real people who identify and look like me.” (Cisgender woman, lesbian, 18–25, CT condition)

Unaided Recall of Thematic Content

Health Impacts

Regardless of condition, participants recalled thematic content about long-term physical health harms, including heart disease and cancer. One participant reflected on messages about health risks for bystanders: “The second-hand effects and heart disease/blood pressure ones stuck in my mind best. Been thinking more about the second-hand effects and the influence on others” (Non-binary, pansexual, 26–35, CT condition). Fewer participants recalled short-term health harms (eg, high blood pressure, sleep, lung damage) as memorable: “As someone who suffers from insomnia, the message about nicotine affecting sleep schedules was particularly memorable” (Cisgender man, bisexual, 18–25, control condition).

Social Justice

“I'm still thinking about the one that said vapes and cigarettes are being marketed towards LGBT people, and how quitting was an act of resistance…every time I hit my vape I thought of that one” (Cisgender woman, gay or lesbian, 18–25, CT condition).

Participants in both arms resonated with thematic content about the tobacco industry targeting minoritized LGBTQ+ (CT condition) and Black communities (control condition):

“The most memorable [image] discussed how Black people are disproportionately impacted by the harmful effects of smoking. I find images related to social issues more impactful than images related to my own personal health.” (Non-binary, gay or lesbian, 18–25, control condition)

In response to social justice themes, participants in both conditions discussed how group resiliency influenced quit motivations: “It was very empowering and makes me want to quit vaping to fight the power” (Cisgender woman, gay or lesbian, 18–25, CT condition).

Motivations to Quit N/T Use

Considering or Planning to Quit

“[The images] have added pressure to the already nagging feeling of wanting to quit” (Cisgender man, gay, 26–35, control condition).

Postexposure, participants from both conditions reported that they were considering or planning to quit. Experimental participants discussed how the stimuli made them rethink N/T use: “I’ve been trying to quit for a while but continue to be unsuccessful but having these in the back of my mind might make a shift” (Cisgender woman, gay or lesbian 18–25, CT condition). Others discussed alternative quit methods such as non-nicotine vapes. Control participants also shared how stimuli made them reconsider N/T use: “[The stimuli] definitely made me think more about…what I want to do about my cig consumption going forward, but I haven’t committed to quitting yet” (Cisgender woman, gay or lesbian, 18–25, control condition).

Reducing Use

“Yes, they made me rethink and seriously cut down. I only smoke maybe half a pack a week though I do still vape more frequently than I'd like.” (Cisgender woman, bisexual, 26–35, CT condition)

Fewer participants across conditions shared how exposure to the stimuli affected their plans to reduce N/T use: “It reminded me that I should’ve never smoked that first cigarette. And even though vaping is a ‘healthier’ alternative, it’s still not a great thing to do. I will likely switch to a lower nicotine vape juice until I can eventually quit altogether” (Cisgender man, gay, 18–25, CT condition). Some discussed reducing social smoking and vaping: “I feel more strongly about not using other people’s vapes but will continue to occasionally smoke cigarettes” (Cisgender woman, bisexual, 18–25, control condition).

Not Quitting

“Health messages about smoking generally don't affect me. I already know it's bad for my health. What I need to know is how to cope with the feeling of relief from life, concentration, etc. it provides” (Cisgender woman, bisexual, 26–35, control condition).

Participants from both conditions expressed that they were not planning to quit post-exposure. While most acknowledged the harms of N/T use, many felt they did not have sufficient self-efficacy to quit given sociocultural stressors. One participant said: “I know I should stop and want to eventually but don’t feel I’m in a place to do so” (Transgender woman, queer, 18–25, CT condition). Another stated: “I want to quit but life is hard right now so it’s hard to make an adjustment. The pictures and message [were] impactful for when I am at a better place mentally to quit” (Cisgender woman, pansexual, 26–35, control condition). Others felt that the stimuli needed more information about alternative coping strategies for nicotine addiction and stress management.

Discussion

Tobacco public education campaigns demonstrate effectiveness for preventing and reducing tobacco use among the general public,⁴⁴ and a growing body of research is examining strategies to optimize messaging for populations experiencing N/T disparities,⁴⁴ including cultural targeting.^17,20 Our study extends this literature by identifying specific visual and thematic content that captures LGBTQ+ YA’s attention, aiding in message recall and perceptions that may influence behavior.

LGBTQ+ YA responded positively to CT tobacco public education but still resonated with nontargeted advertisements. Regardless of condition, greater visual attention to the image AOI was associated with greater message relevance. In our formative research,¹⁴ YA described how CT tobacco public education that incorporated subtle LGBTQ+ iconography and model diversity (ie, gender-diverse people) increased effectiveness. In this study, these visual elements were associated with unaided recall of tobacco public education messages over time. Participants also recalled and responded positively to stimuli that were not CT but included demographically diverse models (eg, by race, age). This is important, as it suggests that LGBTQ+ YA may respond well to inclusive campaigns like “ETC” in which LGBTQ+ CT and nontargeted advertisements are integrated.

Responses to imagery were not universally positive. Participants qualitatively critiqued CT stimuli that felt forced due to stock photographs. This echoes prior findings in which LGBTQ+ YA were skeptical of LGBTQ+ CT tobacco public education¹⁴ and targeted commercial marketing,⁴⁵ which they did not perceive as authentic. LGBTQ+ audiences may question the motivations behind CT marketing such that symbolic representation of LGBTQ+ communities (eg, rainbow colors, stereotypical models) is not sufficient to cultivate trust.

Across conditions, greater visual attention to harms messaging was positively associated with message relevance and higher PME scores for cigarette smoking, suggesting that thematic content about health harms may resonate personally with LGBTQ+ YA and, thus, could influence cigarette smoking attitudes and behaviors. Participants qualitatively recalled various health impacts and social justice issues as memorable. Participants described stimuli addressing industry exploitation of minoritized communities as novel and impactful. Our findings extend those by Schillo and colleagues, in which young people were generally unaware that Big Tobacco targets the LGBTQ+ community and communities of color.⁴⁶ In the general population, tobacco public education, exposing intentional targeting by the tobacco industry (ie, counter-industry marketing), is effective at evoking anger toward Big Tobacco.^19,47 In their study of YA SMW, Zulkiewicz et al. found that anti-industry beliefs mediated the effect between exposure to CT tobacco public education and quit intentions.²¹ Together, these studies suggest that including information about tobacco industry manipulation may be impactful, not only for message engagement and recall, but for attitudinal and behavior change among LGBTQ+ YA who use N/T.

Surprisingly, visual attention to efficacy AOIs was not associated with message effectiveness. Our qualitative findings suggest that stimuli may need to include content that highlights realistic quitting strategies to increase effectiveness. This aligns with prior studies wherein participants responded positively to antitobacco communications that described cessation and harm minimization strategies.¹⁴ Across conditions, half of the participants qualitatively expressed that they were considering or planning to quit N/T use after the study. However, almost one-third of the sample stated that they were not planning to quit. These participants described how contextual stressors were barriers to quitting and that messages may need to include alternative quit strategies and resources to support self-efficacy.

Our convenience sample limits generalizability to the US LGBTQ+ YA population, but robust representation of Black and African American participants in the sample is a strength. Most participants were cisgender women; future studies should include more participants of diverse genders. This study is the first to experimentally assess how exposure to LGBTQ+ CT (vs. nontargeted) tobacco public education influences visual attention, recall of thematic content and imagery, and perceptions. Single exposure to stimuli may not be sufficient for attitudinal and behavioral change; however, many participants expressed that exposure increased their motivations to quit or reduce N/T use. Future studies with repeated exposures to CT tobacco public education are needed. In the online eye-tracking experiment, participants were asked to hold their heads still and view each stimulus for 10 s before proceeding. This is different than how YA would interact with tobacco public education in a social media setting, where they could skip advertisements deemed irrelevant. This study employed remote eye-tracking via participants’ webcams, which is less accurate than lab-based eye-tracking equipment. However, studies have validated iMotions’ web eye-tracking software,³⁷ and conducting the study remotely allowed us to recruit a diverse sample while reducing the burden. The one-week follow-up period does not capture long-term recall, so an analysis of recall, attitudes, and behaviors over a prolonged period could be beneficial. Future studies should examine intragroup variance in response to health communications within LGBTQ+ subpopulations (eg, by gender or race).

Although there were few differences between conditions in recall of thematic content, we learned that visual attention to harms messaging may increase message relevance and effectiveness among LGBTQ+ YA. Qualitatively, participants highlighted harms messages that referenced industry targeting messaging as novel. Thus, combining this thematic content may increase LGBTQ YA’s attention and positive response to tobacco public education. Visual attention to imagery was associated with message relevance, and LGBTQ+ YA positively recalled visuals they perceived as authentic and representing their diverse identities, including and beyond LGBTQ+ status. Meaningfully integrating LGBTQ+ people and stories into broad and diverse tobacco public education campaigns may be a viable strategy for reaching LGBTQ+ YA who use N/T.

Author Contributions

Maxwell Schoen (Formal analysis [equal], Writing—original draft [lead], Writing—review & editing [equal]), Sydney Galusha (Writing—original draft, Writing—review & editing [equal]), Alysha C. Ennis (Data curation, Formal analysis [supporting], Project administration [lead], Writing—original draft [supporting], Writing—review & editing [lead]), Elle Elson (Methodology, Project administration, Writing—review & editing [supporting]), Emma Jankowski (Methodology, Writing—review & editing [supporting]), Ashley Meadows (Data curation [lead], Formal analysis [equal], Writing—review & editing [supporting]), Monica Stanwick (Writing—review & editing [equal]), Hayley Curran (Project administration [equal], Writing—review & editing [supporting]), Elizabeth G. Klein (Formal analysis, Methodology, Writing—review & editing [supporting]), and Joanne G. Patterson (Conceptualization [lead], Formal analysis [supporting], Funding acquisition, Investigation, Methodology [lead], Project administration [supporting], Supervision [lead], Writing—original draft [equal], Writing—review & editing [lead])

Funding

Research reported in this publication was funded by the National Cancer Institute (NCI) of the National Institutes of Health (NIH) and the U.S. Food and Drug Administration (FDA) Center for Tobacco Products under Award Number K99CA260718 and R00CA260718 (PI: JGP). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the FDA. Research reported in this publication was supported by The Ohio State University Comprehensive Cancer Center and the OSU College of Public Health.

Declaration of Interests

None declared.

Data Availability

Data and materials available upon request.