This is an open-access article distributed under the terms of the Creative Commons Attribution License (
Dual use of combustible cigarettes and nicotine vapes is disproportionately high among lesbian, gay, bisexual, transgender, and queer (LGBTQ+) young adults. Mass-reach health communications may be effective at curbing dual use. Current research is exploring whether comparative risk messaging, which presents nicotine vapes as less harmful than cigarettes, reduces dual use.
This formative message testing study focused on communicating the health risks of cigarette smoking and nicotine vaping use to young adults susceptible to dual use of these products, including LGBTQ+ young adults.
Online focus groups were conducted with young adults to develop candidate messages (N=12). Interviews (N=13) qualitatively explored thematic content. An online rating survey (N=286) quantitatively assessed perceived message effectiveness (PME) of and reactance to candidate messages applying standard and comparative risk message framing, compared to adapted regulatory warnings used by the US Food and Drug Administration.
Qualitatively, interview and focus group participants found messages featuring novel information, including toxic constituents and physical harms (eg, hypertension), most effective. “Known” harms (eg, cancer) were described as effective by LGBTQ+ young adults due to the “shock value” of fear appeals. However, participants recommended pairing “known” harms with novel information; for example, addiction messaging was best received when described in the context of social or occasional use. Comparative messaging was appealing for harm reduction (ie, encouraging young adults who use cigarettes to quit smoking and use nicotine vapes), especially among LGBTQ+ participants. However, participants were concerned that comparative messages could unintentionally promote vaping among nicotine-naïve young adults. Qualitative participants preferred gain-framed efficacy messages that encouraged rather than demanded behavior change. Efficacy messages that emphasized “switching” were described as permissive for vaping, and participants were concerned that these may encourage sustained nicotine use. Some questioned whether vaping could effectively help young adults quit smoking. Survey results supported qualitative findings: messages with highest PME scores addressed toxic constituents, heart and lung disease, and cancer. Addiction messages were least effective. Among participants engaged in dual use, PME-smoking scores were higher when viewing candidate comparative messages than regulatory messages, but there were no significant differences between standard and comparative messages. The most effective candidate efficacy messages addressed quitting all smoking and vaping to reduce health risks. Messages that encouraged quitting smoking and switching to vapes were rated least effective.
Comparative messaging was associated with higher PME-smoking among young adults engaged in dual use but did not consistently outperform standard messaging. Qualitative findings suggest that comparative framing may be misinterpreted as endorsing vaping as “safe” rather than “lower harm than cigarettes.” Further research is needed to examine potential unintended consequences of comparative messaging, including sustained nicotine use among young adults who smoke or normalization of vaping among nicotine-naïve young adults.
Concurrent use of nicotine vapes and combustible cigarettes (“dual use”) is a persistent and growing problem, particularly among young adults living in the United States [
Mass-reach health communication campaigns have been effective for preventing initiation [
In 2016, the US Food and Drug Administration (FDA) proposed a new nicotine-based framework for public health, in which they described products that deliver nicotine on a continuum of risk, based on toxicity and addictiveness [
Comparative risk messages are expected to influence tobacco-related perceptions and intentions through a defined sequence of cognitive and affective responses. For example, messages that contrast the toxicant profile of cigarettes with that of nicotine vapes should heighten relative harm perceptions of cigarettes, which in turn should increase perceived message effectiveness (PME) for combustible cigarette smoking-related outcomes (PME-smoking) and strengthen intentions to quit smoking. At the same time, if comparative messages highlight the lower toxicant exposure associated with vaping, they may also elevate intentions to switch to exclusive nicotine vape use among individuals who smoke, particularly those engaged in dual use. This mechanism also creates the possibility of unintended vaping promotion among nonusers, leading to initiation. Comparative messages may also encourage uptake among mono-product users, resulting in sustained use among young adults who vape or dual use among young adults who smoke cigarettes. In contrast, standard messages that emphasize shared risks without explicit comparisons are expected to produce more uniform risk perceptions across products and fewer shifts in quit- or switch-related intentions. In this study, these hypothesized pathways informed the selection of outcomes, including PME-smoking, PME-vaping, and intentions to quit or switch.
A growing body of evidence suggests that comparative communications can effectively influence harm perceptions and behaviors among adults who smoke; however, little is known about how young adults—including those susceptible to or engaged in dual use—interpret and respond to such messages. The goal of this study was to develop candidate health communications that described either (1) shared risks of cigarettes and nicotine vapes (standard framing) or (2) comparative risks of cigarettes and nicotine vapes (comparative framing) with the aim of informing future campaigns to reduce dual use among young adults. Given documented disparities in dual use across subgroups of young adults, we assessed whether responses to standard and comparative messages differed by LGBTQ+ status. We also assessed if responses differed by current smoking and vaping status.
We conducted a mixed methods formative study to develop and test candidate messages. The study included three sequential phases: (1) focus groups to inform message development, (2) in-depth interviews to qualitatively explore perceptions of thematic content (ie, topics, populations, and tone) applied in standard and comparative candidate messages, and (3) an online rating survey to quantitatively evaluate PME (ie, beliefs about how strongly a candidate message will influence behavior) and psychological reactance (ie, motivational resistance felt in response to a perceived threat to freedom of choice).
This study complied with the protection of human subjects in accordance with the ethical standards of the institutional review board and with the 1964 Helsinki Declaration and its later amendments. The study was approved by The Ohio State University Institutional Review Board (OSU-20427). Participants provided informed consent prior to participation.
Online focus groups were conducted with young adults who were currently engaged in or susceptible to dual use of cigarettes and nicotine vapes. Evidence indicates that individuals who exclusively use cigarettes or nicotine vapes often transition to later dual use, which is more commonly observed in younger adults [
We conducted semistructured qualitative interviews with young adults (aged 18‐35 years) living in the United States who reported ever using cigarettes and nicotine vapes and who used at least 1 of those products in the past month. Eligibility criteria were informed by formative focus groups in which participants disagreed about the potential effects of comparative messaging on mono-product users (ie, concern that exposure to comparative messaging might support harm reduction among people who only smoke cigarettes but promote continued vaping among those who only vape nicotine). As such, we recruited young adults who reported ever using both products into our sample, including current mono-product users, so that we could further consider those concerns.
Our planned analytic sample size was 8, designed to maximize the information power of qualitative research [
An experienced facilitator used semistructured questions to guide the interviews. Participants were asked about their current and historical dual use prior to viewing 3 groups of candidate messages (12 standard, 12 comparative, and 12 efficacy messages; Table S2 in
All interviews were audio-recorded, transcribed, and checked by an independent transcriptionist for subsequent analysis. We applied thematic analysis to code and analyze interview data (
Qualitative codebook for online interviews with young adults conducted between March and April 2022 examining standard risk, comparative risk, and efficacy messages regarding cigarette smoking and nicotine vaping.
| Code | Definition | Statements, n (%) | Participants, n (%) |
| Message framing | |||
| Discusses short+simple as an effective health communication tactic, either by saying a message would be better shorter or simpler or liking a message for that reason. | 17 (2.62) | 8 (61.54) | |
| Discusses facts and statistics as an effective health communication tactic. | 17 (2.62) | 7 (53.85) | |
| Discusses facts and statistics as NOT or LESS effective. | 9 (1.39) | 5 (38.46) | |
| Discusses mentioning toxic constituents (formaldehyde, etc) as effective. | 28 (4.31) | 10 (76.92) | |
| Discusses toxic constituents as NOT or LESS effective. | 9 (1.39) | 6 (46.15) | |
| Discusses “fear” as effective. May also use words like “scary” or “afraid.” | 17 (2.62) | 9 (69.23) | |
| Discusses “fear” as NOT or LESS effective. | 2 (0.31) | 2 (15.38) | |
| Discusses “shock value” as effective. Participants may discuss being “surprised” by a message, making it effective. | 4 (0.62) | 1 (7.69) | |
| Discusses “shock value” as NOT being effective. | 2 (0.31) | 1 (7.69) | |
| Discusses “novelty” (ie, a new idea, or new information, “outside the box” thinking) as effective. | 32 (4.93) | 7 (53.85) | |
| Discusses “novelty” (ie, a new idea, or new information, “outside the box” thinking) as ineffective. | 7 (1.08) | 3 (23.08) | |
| Discusses how “knowing” that something is “true” (eg, smoking causes cancer) makes the message more effective (vs something novel or new or unheard of). | 39 (6.01) | 9 (69.23) | |
| Discusses how “knowing” that something is “true” (eg, smoking causes cancer) makes the message NOT or LESS effective (vs something novel or new or unheard of). | 24 (3.70) | 9 (69.23) | |
| Discusses how referencing health harms or other impacts of smoking or vaping (eg, financial) on family or friends or bystanders is effective. | 2 (0.31) | 1 (7.69) | |
| Contextual factors | |||
| Discusses how real-life context or current events (eg, opioid epidemic and COVID-19) influence their perceptions of the message. | 4 (0.62) | 4 (30.77) | |
| Discusses how personal experience (either their own use or that of a peer/loved one) influences their perceptions of the message. | 53 (8.17) | 11 (84.62) | |
| Discusses how a history of prior quit attempts influences their perceptions of the message. | 3 (0.46) | 2 (15.38) | |
| Discusses how audience (ie, whether target audience is smokers or vapers; established or not established users; and adolescents, young, or older adults) influences how effective or acceptable the message is. | 74 (11.40) | 13 (100) | |
| Product and product use factors | |||
| Discusses the comparative risk of vaping vs smoking combustible cigarettes (or the messaging around comparative risk). | 7 (1.08) | 5 (38.46) | |
| Discusses how smoking and vaping have similar risks, or how referencing smoking or vaping in the same statement makes the risk seem equivalent. | 13 (2) | 7 (53.85) | |
| Discusses nicotine vapes or vaping as a modified risk (healthier or lower harm) product than traditional combustible tobacco or smoking. | 32 (4.93) | 10 (76.92) | |
| Discusses general health when viewing the message, or generally discusses how the message evoked a thought about addiction, whether positive or negative. | 46 (7.09) | 13 (100) | |
| Discusses general addictiveness when viewing the message, or generally discusses how the message evoked a thought about addiction, whether positive or negative. | 35 (5.39) | 12 (92.31) | |
| Discusses using nicotine or tobacco as a harm reduction strategy (ie, instead of other drug use or to manage mental health or stress). | 4 (0.62) | 3 (23.08) | |
| Discusses using “switch” language to describe vaping in lieu of smoking. May use terms including, but not limited to “turn to,” “change over,” “substitute,” “alternative,” or “use instead.” | 9 (1.39) | 7 (53.85) | |
| Discusses using information on how to decrease smoking on the path to quitting. | 14 (2.16) | 5 (38.46) | |
| Discusses how building an “off ramp” (ie, steps to quitting smoking, potentially including what to do once you know smoking is bad for you) is important content to include. | 17 (2.62) | 4 (30.77) | |
| Language | |||
| Discusses how messaging or content is permissive (ie, makes vapes and/or cigarettes seem fine to use). | 36 (5.55) | 9 (69.23) | |
| Discusses the tone of the message (eg, “preachy” and “friendly”). | 27 (4.16) | 8 (61.54) | |
| Discusses how using “lay” or “everyday” language is effective (eg, heart disease vs cardiovascular disease). | 1 (0.15) | 1 (7.69) | |
| Discusses how using “scientific” or “professional” language is effective (eg, “cardiovascular disease” vs “heart disease”). | 6 (0.92) | 4 (30.77) | |
| Discusses how using more specific language is effective (eg, “wheezing” vs “lung injury”). | 22 (3.39) | 10 (76.92) | |
| Discusses how using broader language is effective (eg, “lung injury” vs “wheezing”). | 7 (1.08) | 3 (23.08) | |
| Discusses using person-first language (eg, “people who smoke” or “using cigarettes” [implied person]) vs product-first language (eg, “smoker” or “vaper”). | 2 (0.31) | 1 (7.69) | |
| Discusses using language to appeal to a general audience | 1 (0.15) | 1 (7.69) | |
We conducted an online survey with 326 young adults (ages 18‐35 years) living in the United States. Participants were recruited through Prolific [
To ensure data integrity, we cross-verified participants’ responses. In total, 37 respondents were excluded because their self-reported tobacco use history in the survey was incongruent with initial screening responses. After excluding 3 participants due to incomplete responses on key outcome variables or covariates, the final analytic sample included 286 US young adults who reported ever smoking and vaping.
PME was assessed using the 3-item UNC Perceived Message Effectiveness Scale (original α=.90), which assesses effects perceptions or beliefs about a message’s potential to change behavior [
Efficacy messages were assessed by their perceived effectiveness (PE) for motivating people to quit cigarettes and nicotine vapes on a 5-point Likert scale (1=strongly disagree and 5=strongly agree) with the statement: “This message will motivate people to quit all e-cigarette and combustible cigarette products” (PE-quit). We also assessed PE for motivating people to switch from cigarettes to nicotine vapes using the same scale and the statement: “This message will motivate people who smoke cigarettes to switch completely to vaping e-cigarettes” (PE-switch).
Psychological reactance measures a person’s negative reactions to a message, which may preclude counterarguing [
To estimate message-level effects while accounting for the incomplete-block design, we used cross-classified linear mixed-effects models with random intercepts for participants and messages. This approach modeled repeated observations nested within participants and candidate messages.
Models were estimated separately for PME (smoking and vaping), PE of efficacy messages (quit and switch), and psychological reactance. For each outcome, we first estimated an unconditional (intercept-only) model and used likelihood ratio (LR) tests to evaluate the necessity of the multilevel structure. Model specification was iteratively evaluated using LR tests. Random slopes did not significantly improve fit for any model and were excluded to maintain parsimony. For the PME-smoking and vaping models, demographic covariates failed to improve fit, justifying the use of intercept-only structures. Conversely, model fit was significantly improved by adding sex-at-birth to the PE-quit and PE-switch models, and a binary age indicator (18‐25 vs 26‐35 years) to both psychological reactance models. From these final models, we derived message-specific adjusted means and 95% CIs via partial pooling to stabilize individual message estimates. Finally, to control the inflation of type I error across multiple outcomes, we applied Bonferroni corrections separately within each experiment based on the number of primary dependent variables evaluated (3 outcomes per experiment, Bonferroni-corrected α=0.0167).
We then assessed systematic differences in message framing effects by LGBTQ+ status and smoking or vaping status using linear regressions with robust SEs. To examine whether framing effects varied by group, we included product interaction terms in all models, regardless of main effect significance. If interactions were not statistically significant (
All participants (N=12) reported past 30-day nicotine vaping, and 11 reported past 30-day cigarette use. They were, on average, 23.75 (SD 4.07) years of age, and 7 identified as LGBTQ+.
Generally, participants preferred prompts that applied person-first language (eg, “people who smoke” vs “smokers”) and were less enthusiastic about prompts referring to specific subgroups of “young people” or “young adults.” Participants preferred “everyday” language to describe electronic nicotine products, including “vaping” or “vapes” (vs e-cigarettes) but noted that messages should specify “nicotine vaping / vapes” to avoid confusion with nonnicotine vape products (ie, cannabis, caffeine, and herbal).
Overall, standard and comparative message prompts describing short- and long-term physical health harms and toxic constituents were perceived as effective and memorable. Those describing nicotine addiction were perceived as least effective. Participants described harm messages that induced fear as attention-getting and effective. Emerging young adults (aged 18‐25 years) described how prior exposure to antitobacco campaigns during K-12 education (eg, Drug Abuse Resistance Education) attenuated the effectiveness of messages describing chronic tobacco-related diseases (eg, cancer and lung disease). As such, emerging young adults found more appeal in message prompts highlighting short-term health harms (eg, effects of nicotine on sleep) and lesser-known toxic constituents (eg, formaldehyde).
Participants held diverse opinions about the salience of comparative risk messages for young people susceptible to dual use. Some appreciated that comparative messages might encourage young adults using combustible cigarettes (either mono-product or dual use) to move down the tobacco harm continuum by vaping only. Others were skeptical of comparative messages and were concerned that they might encourage more frequent vape use among young people who only vape.
Participants described how health communications that solely provided information about harms were insufficient. In addition to providing quit resources for individuals interested in quitting, participants suggested that messages include information about cessation and harm reduction strategies, including encouraging statements and “how to” tips (which they described as “off-ramp” messaging). In response, we developed a series of candidate messages addressing behavioral efficacy, which we explored in qualitative interviews (Table S2 in
In total, 13 participants completed interviews, 7 identified as cisgender male, 5 were White, and 9 identified as LGBTQ+. Their average age was 24.69 (SD 3.71) years. Of participants, 12 used cigarettes, and 11 vaped nicotine during the month of their interview (
Demographic characteristics of young adult participants of online interviews conducted between March and April 2022 examining standard risk, comparative risk, and efficacy messages regarding cigarette smoking and nicotine vaping.
| Values, n (%) | |
| Gender | |
| 4 (30.77) | |
| 7 (53.85) | |
| 2 (15.38) | |
| Sexual orientation | |
| 4 (30.77) | |
| 5 (38.46) | |
| 2 (15.38) | |
| 2 (15.38) | |
| Race | |
| 5 (38.46) | |
| 5 (38.46) | |
| 3 (23.08) | |
| Age (years) | |
| 6 (46.15) | |
| 7 (53.85) | |
| Use in the last month | |
| 12 (92.31) | |
| 11 (84.62) | |
| 5 (38.46) | |
| 4 (30.77) | |
| 1 (7.69) | |
Participants, including a greater proportion of non-LGBTQ+ young adults, felt that novelty was important in candidate messages. When asked what information we should highlight, one participant responded:
Lesser-known information that I’ve never even heard before. I think that would actually get a lot of attention.
Participants considered messages describing toxic constituents and DNA damage novel and explained that, because the public is less aware of these health effects, it increases “shock value”:
It’s a consequence you don’t hear about as much about ... the metals part of it. And then also, like, arsenic and lead sound just really dangerous.
Some participants felt messages required novel information to engage their attention; otherwise, messages would be redundant with prior K-12 antitobacco education.
I feel like since I had DARE, it hasn’t come out that there’s like these new harms of smoking nicotine cigarettes that’s like “Oh, no one knew this.” We get it. It’s horrible for you in every way, so while it’s nice that the other messages, like, say specific things, I think I already know what the harm is.
In contrast, LGBTQ+ participants responded positively to messages describing scientifically established health harms (eg, cancer). Said one participant, “Heart disease is so common ... highlighting physical health risks as the danger [is effective], rather than, ‘Oh, it’s just bad for you’” (18‐25 years, LGBTQ+, current nicotine vape use). LGBTQ+ participants also described how known physical harm messages that induced fear were more attention-getting: “Cancer is one of those statement shocker words of ‘Oh, I could get this?’ I just think the word ‘cancer,’ to see it in any kind of ad ... you stop for a second and you look” (26‐35 years, LGBTQ+, current dual use).
Responses to messages on nicotine addiction were mixed. Messages stating that “nicotine is addictive” were generally considered ineffective: “There’s no point, everyone knows that” (18‐25 years, non-LGBTQ+, current dual use). However, messages presenting novel information about nicotine addiction (eg, in the context of social or occasional use) were viewed as effective. Responding to the message, “Even occasional smoking and vaping can lead to nicotine addiction,” one participant noted:
If I had heard that earlier on, with maybe some more force, I would have probably been a lot more abstinent. Because my whole thinking was like, “Well, I’m not like everyone else, I won’t get addicted. I’m too good for that.” So, it’s finding a way to like, broaden it and make it feel like ... you’re not special. You’ll get addicted. Everyone gets affected by nicotine the same way.
Comparative candidate messages were appealing for their potential for harm minimization. Several participants, including a greater proportion of LGBTQ+ young adults, remarked that comparative messages were effective as a mechanism to reduce the harm of combustible smoking:
There will be some harm reduction if people that smoke cigarettes daily can become more casual vapers. Yeah, so I like [this message].
Other participants interpreted comparative messages to mean that vaping is an effective strategy for quitting smoking:
The reason I like [the message] is, vaping is a way of trying to quit smoking, but it does not mean it’s not harmful at all. So, you have to pass across the message that it is still harmful, but it has a reduced risk from smoking. So, when you’re comfortable and try and quit, vaping will help you.
Efficacy messages emphasizing “switching” were viewed as permissive for vaping. Participants were concerned that efficacy messages directing users to “switch” to vapes might encourage continued nicotine use instead of quitting. In describing this concern, some participants implicitly equated smoking and vaping harms:
I think [this message] is another excuse to keep doing what you’ve been doing, right? You just quit smoking and just go to vaping. I think that isn’t really much different ... I think if you want to quit, you should quit.
Other participants questioned whether vaping was effective for helping smokers eventually quit all nicotine: “... is there research on—like do people that go from smoking to nicotine [vaping], do they ever end up quitting?” (18‐25 years, LGBTQ+, current nicotine vape use).
Tone was important to young adults when assessing the effectiveness of harm minimization and efficacy messages. Most participants, but especially those aged 18‐25 years, preferred when efficacy messages encouraged rather than demanded behavior change. Leading questions (eg, “Not ready to quit completely?”) and empathetic opening statements were described as effective strategies for engaging participants in behavior change: “The way [the messages are] phrased makes it seem that it’s like, it’s my choice to do it, rather than being told to do something” (18‐25 years, LGBTQ+, current nicotine vape use).
Most participants self-identified as White (186/286, 65.03%), 39.16% (112/286) were aged 18‐25 years, 54.55% (156/286) were assigned female at birth, and 51.75% (148/286) identified as LGBTQ+ (
Sample demographic characteristics for an online survey sample of US young adults (aged 18‐35 years) reporting ever using both cigarettes and nicotine vapes, conducted between February and April 2022.
| Full sample (N=286), n (%) | LGBTQ+ | Non-LGBTQ+ (n=138), n (%) | No current use (n=53), n (%) | Current vaping (n=90), n (%) | Current smoking (n=57), n (%) | Current dual use (n=86), n (%) | |||
| Age group (years) | .007 | <.001 | |||||||
| 112 (39.16) | 69 (46.62) | 43 (31.16) | 17 (32.08) | 48 (53.33) | 10 (17.54) | 37 (43.02) | |||
| 174 (60.84) | 79 (53.38) | 95 (68.84) | 36 (67.92) | 42 (46.67) | 47 (82.46) | 49 (56.98) | |||
| Sex assigned at birth | .001 | .61 | |||||||
| 156 (54.55) | 95 (64.19) | 61 (44.20) | 25 (47.17) | 52 (57.78) | 33 (57.89) | 46 (53.49) | |||
| 130 (45.55) | 53 (35.81) | 77 (55.80) | 28 (52.83) | 38 (42.22) | 24 (42.11) | 40 (46.51) | |||
| Hispanic or Latinx | 28 (9.79) | 14 (9.46) | 14 (10.14) | .85 | 7 (13.21) | 6 (6.67) | 7 (12.28) | 8 (9.30) | .55 |
| Race | .35 | .09 | |||||||
| 2 (0.70) | 2 (1.35) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 2 (2.33) | |||
| 14 (4.90) | 8 (5.41) | 6 (4.35) | 4 (7.55) | 4 (4.44) | 1 (1.75) | 5 (5.81) | |||
| 32 (11.19) | 13 (8.78) | 19 (13.77) | 6 (11.32) | 4 (4.44) | 9 (15.79) | 13 (15.12) | |||
| 186 (65.03) | 102 (68.92) | 84 (60.87) | 36 (67.92) | 58 (64.44) | 38 (66.67) | 54 (62.79) | |||
| 37 (12.94) | 16 (10.81) | 21 (15.22) | 3 (5.66) | 19 (21.11) | 5 (8.77) | 10 (11.63) | |||
| 15 (5.24) | 7 (4.73) | 8 (5.80) | 4 (7.55) | 5 (5.56) | 4 (7.02) | 2 (2.33) | |||
| Education level | .51 | .19 | |||||||
| 61 (21.33) | 32 (21.62) | 29 (21.01) | 12 (22.64) | 15 (16.67) | 15 (26.32) | 20 (22.09) | |||
| 35 (12.24) | 16 (10.81) | 19 (13.77) | 9 (16.98) | 13 (14.44) | 6 (10.53) | 7 (8.14) | |||
| 88 (30.77) | 52 (35.14) | 36 (26.09) | 11 (20.75) | 31 (34.44) | 19 (33.33) | 27 (31.40) | |||
| 89 (31.12) | 42 (28.38) | 47 (34.06) | 15 (28.30) | 30 (33.33) | 16 (28.07) | 28 (32.56) | |||
| 13 (4.55) | 6 (4.05) | 7 (5.07) | 6 (11.32) | 1 (1.11) | 1 (1.75) | 5 (5.81) | |||
| Use status | .95 | — | — | — | — | — | |||
| 53 (18.53) | 29 (19.59) | 24 (17.39) | |||||||
| 90 (31.47) | 47 (31.33) | 43 (31.16) | |||||||
| 57 (19.93) | 28 (18.92) | 29 (21.01) | |||||||
| 86 (30.07) | 44 (30.43) | 42 (30.43) | |||||||
aLGBTQ+: lesbian, gay, bisexual, transgender, and queer.
bGED: General Educational Development test.
cNot applicable.
Risk message-specific marginal means are presented in
Cross-classified linear mixed models for standard and comparative risk candidate messages about cigarette smoking and nicotine vaping tested in an online survey of US young adults between February and April 2022 (N=286)
| Parameter | PME | PME-smoking, estimate (95% CI) | Reactance, estimate (95% CI) |
| Fixed effects | |||
| 2.87 (2.62-3.13) | 3.57 (3.38-3.77) | 2.05 (1.86-2.24) | |
| — | — | 0.24 (0.05-0.43) | |
| Random effects (variance) | |||
| 0.83 (0.68-1.02) | 0.79 (0.65-0.96) | 0.52 (0.42-0.64) | |
| 0.32 (0.18-0.58) | 0.16 (0.08-0.29) | 0.09 (0.05-0.17) | |
| 0.82 (0.75-0.89) | 0.72 (0.66-0.78) | 0.66 (0.61-0.72) | |
aA total of 1430 observations across all models. Fixed effects display unstandardized coefficients. Random effects display variance components. A Bonferroni correction was applied for 3 primary outcomes (α=.0167).
bModel fit indices: PME-vaping: log-likelihood=−2181.43; likelihood ratio (LR) test X2 vs ordinary least squares (OLS)=633.64 (
cPME: perceived message effectiveness.
d
eNot available.
f
Mean perceived message effectiveness and psychological reactance scores for standard and comparative risk candidate messages about cigarette smoking and nicotine vaping tested in an online survey of US young adults between February and April 2022 (N=286).
| ID# | Type | Message | PME | PME-vaping, mean (95% CI) | Reactance, mean (95% CI) |
| 9 | Candidate standard | Smoking cigarettes and vaping nicotine increases your risk for heart disease. | 4.06 | 3.64 (3.39-3.90) | 1.78 (1.56-1.99) |
| 11 | Candidate standard | Wheezing lately? Smoking cigarettes and vaping nicotine increases your risk for lung problems. | 4.00 | 3.49 (3.23-3.75) | 1.86 (1.65-2.08) |
| 3 | Candidate standard | Cigarettes and some nicotine vapes contain formaldehyde, a chemical that causes cancer. | 3.99 (3.76-4.22) | 3.66 (3.41-3.91) | 1.84 (1.63-2.05) |
| 8 | Candidate standard | Smoking cigarettes and vaping nicotine damages your DNA and increases your risk for cancer. | 3.79 (3.55-4.02) | 3.49 (3.23-3.74) | 2.05 (1.84-2.27) |
| 5 | Candidate standard | Using cigarettes and nicotine vapes exposes you to heavy metals—like arsenic and lead. | 3.76 (3.52-3.99) | 3.57 (3.31-3.82) | 1.90 (1.68-2.11) |
| 12 | Candidate standard | Can’t find your keys? Nicotine in cigarettes and vapes can harm your brain and lead to memory loss. | 3.56 (3.33-3.79) | 3.25 (3.00-3.50) | 2.47 (2.26-2.68) |
| 10 | Candidate standard | Can’t sleep? Nicotine in cigarettes and vapes can lead to sleep problems, including insomnia. | 3.49 (3.23-3.74) | 3.36 (3.08-3.64) | 1.95 (1.72-2.18) |
| 2 | Candidate standard | Most people think smoking cigarettes and vaping nicotine are harmful to your health. They’re right. | 3.44 (3.20-3.67) | 3.26 (3.00-3.52) | 1.95 (1.74-2.17) |
| 4 | Candidate standard | Nicotine doesn’t cause cancer, but toxic chemicals in e-liquid and tobacco can. | 3.42 (3.19-3.65) | 3.15 (2.89-3.40) | 2.10 (1.88-2.31) |
| 6 | Candidate standard | Even occasional smoking and vaping can lead to nicotine addiction. | 3.30 (3.06-3.54) | 2.91 (2.64-3.17) | 1.86 (1.65-2.08) |
| 7 | Regulatory | Cigarettes and vapes contain nicotine. Nicotine is an addictive chemical. | 3.10 (2.88-3.32) | 2.96 (2.72-3.19) | 1.60 (1.40-1.80) |
| 15 | Candidate | Cigarettes contain more cancer-causing chemicals—like formaldehyde—than nicotine vapes. | 4.05 (3.81-4.30) | 2.25 (1.98-2.52) | 1.91 (1.69-2.14) |
| 16 | Candidate | Many people think nicotine in vapes and cigarettes causes cancer. Truth is, toxic chemicals in burning cigarettes are more likely to cause cancer. | 3.90 (3.66-4.14) | 2.47 (2.20-2.73) | 1.88 (1.66-2.10) |
| 21 | Candidate | People who smoke cigarettes and vape nicotine have a higher risk of developing heart disease than those who only vape. | 3.89 (3.66-4.13) | 2.79 (2.53-3.05) | 1.93 (1.72-2.15) |
| 23 | Candidate | Wheezing much? People who smoke cigarettes and vape nicotine report more lung problems than those who only vape. | 3.84 (3.59-4.09) | 2.93 (2.66-3.20) | 2.12 (1.89-2.35) |
| 17 | Candidate | Using cigarettes and nicotine vapes exposes you to more heavy metals—like arsenic and lead—than if you only vape. | 3.76 (3.53-4.00) | 3.00 (2.74-3.26) | 2.03 (1.81-2.24) |
| 20 | Candidate | Smoking cigarettes can increase your risk for lung, head, and neck cancer. Using nicotine vapes only may decrease your cancer risk. | 3.74 (3.50-3.99) | 2.20 (1.93-2.47) | 2.31 (2.09-2.54) |
| 13 | Candidate | Vaping nicotine is harmful to your health, but it is less harmful than smoking. | 3.62 (3.39-3.85) | 2.33 (2.08-2.58) | 2.14 (1.93-2.34) |
| 14 | Candidate | Many people think vaping nicotine is just as harmful as smoking cigarettes. Truth is, nicotine vapes are less harmful than cigarettes. | 3.45 (3.21-3.69) | 1.87 (1.61-2.13) | 2.13 (1.91-2.35) |
| 19 | Candidate | Nicotine vapes are addictive, but people who vape and smoke tobacco are 7 times more likely to be addicted to nicotine than those who only vape. | 3.44 (3.20-3.67) | 2.66 (2.41-2.92) | 1.93 (1.72-2.14) |
| 22 | Candidate | Can’t sleep? People who vape and smoke report more sleep problems than those who only vape. | 3.42 (3.19-3.65) | 2.44 (2.18-2.69) | 2.27 (2.06-2.48) |
| 18 | Candidate | Vaping heats nicotine, resulting in lower levels of harmful chemicals than burned tobacco in cigarettes. | 3.41 (3.17-3.65) | 2.12 (1.86-2.38) | 1.98 (1.76-2.19) |
| 1 | Regulatory comparative | Nicotine vapes are not a safe alternative to cigarettes. | 2.69 (2.45-2.93) | 3.26 (3.00-3.52) | 2.08 (1.86-2.30) |
| 24 | Candidate | Can’t find your keys? Research suggests that vaping nicotine can improve attention and memory during smoking abstinence. | 2.60 (2.36-2.83) | 1.92 (1.67-2.18) | 3.06 (2.84-3.27) |
aPME: perceived message effectiveness.
Most standard candidate messages scored PMEs above the midpoint score (>3) for smoking and vaping. Those with the highest PMEs addressed the risk of heart disease (ID 9), lung disease (ID 11), cancer (IDs 3 and 8), and toxic constituents, including formaldehyde (ID 3) and heavy metals (ID 5). Two standard candidate messages had lower PMEs for smoking and vaping; both messages addressed the addictive potential of nicotine (IDs 6 and 7).
Like standard candidate messages, most comparative candidate messages scored PMEs for smoking behavior >3. The comparative candidate messages with the highest PME-smoking scores addressed how smoking increases cancer risk more than vaping (IDs 15 and 16), heart disease (ID 21), lung disease (ID 23), and heavy metals (ID 17). With respect to PME-vaping scores, comparative candidate messages were less effective. Only one message scored above 3 for PME-vaping (ID 1).
Almost all standard and comparative candidate messages scored below the midpoint for psychological reactance, indicating that participants did not respond negatively to the messages (
Effect of risk message framing on perceived message effectiveness (PME) and reactance: findings from an online survey of US young adults reporting ever using both cigarettes and nicotine vapes conducted between February and April 2022 (N=286)
| Mean (95% CI) | ||||
| Interaction effects of message framing on PME-smoking, by current use status | ||||
| Regulatory standard message (reference) | 2.85 (2.39-3.32) | — | — | — |
| Regulatory comparative message | 1.81 (1.41-2.20) | — | — | |
| Standard candidate messages | 3.38 (3.13-3.63) | .22 | — | |
| Comparative candidate messages | 3.56 (3.35-3.77) | .43 | ||
| Regulatory standard message (reference) | 2.99 (2.66-3.32) | — | — | — |
| Regulatory comparative message | 3.38 (3.05-3.72) | .43 | — | — |
| Standard candidate messages | 3.39 (3.25-3.54) | .10 | >.99 | — |
| Comparative candidate messages | 2.39 (2.26-2.51) | |||
| Regulatory standard message (reference) | 1.68 (1.48-1.87) | — | — | — |
| Regulatory comparative message | 2.10 (1.82-2.39) | .08 | — | — |
| Standard candidate messages | 2.12 (2.00-2.23) | >.99 | — | |
| Comparative candidate messages | 2.32 (2.20-2.44) | .63 | ||
aA total of 286 young adults viewed 5 randomly selected textual health messages (of N=24 messages) with differing message framing. Linear regression models with generalized estimating equations were used to model differences in outcome variable means by message risk framing (regulatory standard, regulatory comparative, standard candidate, comparative candidate). All models controlled for baseline differences. When no interactions were identified, statistically significant main effects models without interaction terms are reported.
b
cSignificant interactions between message type and nicotine vape or cigarette use status were identified and decomposed via simple main effects within each level of the moderator. To minimize family-wise error at the decomposition stage, we evaluated simple main effects against a Bonferroni-corrected α=0.0125 (adjusting for the 4 moderator levels); only strata reaching this significance threshold are displayed. Post-hoc pairwise comparisons were conducted using Sidak adjustment. All participants reported ever using both cigarettes and nicotine vapes; participants were categorized by self-reported current use of cigarettes and nicotine vapes in the past 30 days as follows: current nonuse of cigarettes or nicotine vapes, current nicotine vape vaping only, current cigarette smoking only, or current dual use of cigarettes and nicotine vapes.
dNot available.
eSignificant differences indicated in italics format (
We observed a significant interaction between message framing and current nicotine or tobacco use status (
We found no significant interaction effects for PME-vaping. Main effects models revealed that message framing was associated with PME-vaping (
We found no significant interaction effects for reactance. Main effects models revealed that message framing was associated with reactance (
Cross-classified linear mixed models for standard and comparative risk candidate messages about cigarette smoking and nicotine vaping tested in an online survey of US young adults between February and April 2022 (N=286)
| Parameter | PE | PE-switch, estimate (95% CI) | Reactance, estimate (95% CI) |
| Fixed effects | |||
| 2.59 (2.25 to 2.94) | 2.94 (2.57 to 3.32) | 1.83 (1.64 to 2.02) | |
| −0.24 (−0.44 to −0.03) | −0.22 (−0.43 to −0.02) | — | |
| — | — | 0.30 (0.08 to 0.52) | |
| Random effects (variance) | |||
| Participant σ2 | 0.46 (0.34 to 0.61) | 0.44 (0.33 to 0.60) | 0.64 (0.51 to 0.80) |
| Message σ2 | 0.30 (0.13 to 0.71) | 0.37 (0.16 to 0.85) | 0.02 (0.01 to 0.07) |
| Residual σ2 | 0.98 (0.87 to 1.10) | 0.98 (0.88 to 1.11) | 0.63 (0.56 to 0.71) |
aA total of 858 observations across all models. Fixed effects display unstandardized coefficients. Random effects display variance components. A Bonferroni correction was applied for 3 primary outcomes (α=.0167).
bModel fit indices: PE-quit: log-likelihood=−1350.9; likelihood ratio (LR) test X2 vs ordinary least squares (OLS)=197.03 (
cPE: perceived effectiveness.
d
eNot available.
f
Efficacy message-specific marginal means are presented in
Mean perceived effectiveness of candidate smoking and vaping messages to motivate quitting and switching behaviors and psychological reactance scores for candidate messages tested in an online rating survey of US young adults between February and April 2022 (N=286).
| Message | Effectiveness for motivating quitting smoking, mean (95% CI) | Effectiveness for motivating switching to vapes only, mean (95% CI) | Psychological reactance, mean (95% CI) | |
| Q2 | Quit all smoking and vaping to reduce your cancer risk. | 3.51 (3.25-3.78) | 2.21 (1.94-2.47) | 1.74 (1.57-1.91) |
| Q3 | Quit smoking and vaping completely to reduce your exposure to toxic chemicals. | 3.28 (3.03-3.53) | 2.34 (2.09-2.59) | 1.77 (1.60-1.93) |
| Q1 | Quit all smoking and vaping to reduce serious risks to your health. | 3.06 (2.78-3.33) | 2.26 (1.98-2.54) | 1.86 (1.69-2.04) |
| Q4 | Ready to quit smoking and vaping completely? Manage your cravings with nicotine replacement therapy. | 3.04 (2.76-3.32) | 2.34 (2.06-2.63) | 1.77 (1.59-1.95) |
| C4 | If you’re not ready to quit all nicotine, using nicotine replacement therapy can help you manage cravings and quit smoking. | 2.92 (2.68-3.16) | 2.61 (2.37-2.86) | 1.64 (1.48-1.81) |
| S4 | Manage your smoking cravings with nicotine replacement therapy. | 2.77 (2.53-3.01) | 2.54 (2.30-2.78) | 1.72 (1.56-1.88) |
| S1 | Switching completely from smoking to vaping nicotine can reduce serious risks to your health. | 2.29 (2.04-2.54) | 3.59 (3.34-3.84) | 2.02 (1.86-2.19) |
| C2 | If you’re not ready to quit all nicotine, switching completely to vaping may reduce your cancer risk. | 2.15 (1.89-2.41) | 3.89 | 1.83 (1.66-2.00) |
| S3 | Switching from smoking to vaping only can reduce your exposure to toxic chemicals. | 2.12 (1.86-2.38) | 3.07 (2.81-3.33) | 1.89 (1.72-2.06) |
| S2 | Switching completely from smoking to vaping nicotine can reduce your cancer risk. | 2.10 (1.85-2.35) | 3.51 (3.26-3.76) | 1.91 (1.74-2.07) |
| C3 | Start quitting smoking by switching to nicotine vapes only. | 1.96 (1.72-2.20) | 3.47 (3.23-3.70) | 1.97 (1.81-2.13) |
| C1 | Not ready to quit completely? Quit smoking all tobacco, and use nicotine vapes only to reduce serious risks to your health. | 1.94 (1.70-2.19) | 3.46 (3.22-3.71) | 1.85 (1.68-2.01) |
All efficacy messages scored below the midpoint for psychological reactance (
Effect of efficacy message framing on perceived effectiveness (PE) and reactance: findings from an online survey of US young adults reporting ever using both cigarettes and nicotine vapes conducted between February and April 2022 (N=286)
| Mean (95% CI) | |||
| Main effects of message framing on PE-quit | |||
| 3.11 (2.93-3.29) | — | — | |
| 2.23 (2.08-2.39) | — | ||
| 2.08 (1.94-2.23) | .33 | ||
| Main effects of message framing on PE-switch | |||
| 2.14 (1.98-2.29) | — | — | |
| 3.07 (2.92-3.23) | — | ||
| 3.22 (3.06-3.38) | .38 | ||
| Interaction effects of message framing on reactance, by current use status | |||
| 1.77 (1.47-2.08) | — | — | |
| 2.43 (2.07-2.78) | — | ||
| 2.22 (1.82-2.62) | .15 | .77 | |
aA total of 286 young adults viewed 3 randomly selected textual health messages (of N=12 messages) with differing message framing. Linear regression models with generalized estimating equations were used to model differences in outcome variable means by message framing (“Quit all smoking/vaping”, “Switch to vaping or nicotine replacement therapy (NRT)”, and “Quit smoking and switch to vaping or NRT”). All models controlled for baseline differences. When no interactions were identified, statistically significant main effects models without interaction terms are reported.
b
cNot applicable.
dNRT: nicotine replacement therapy.
eSignificant interactions between message type and nicotine/tobacco use status were identified and decomposed via simple main effects within each level of the moderator. To minimize family-wise error at the decomposition stage, we evaluated simple main effects against a Bonferroni-corrected α of 0.0125 (adjusting for the 4 moderator levels); only strata reaching this significance threshold are displayed. Post-hoc pairwise comparisons were conducted using Sidak adjustment, with significant differences indicated in italics format. All participants reported ever using both cigarettes and nicotine vapes; participants were categorized by self-reported current use of cigarettes and nicotine vapes in the past 30 days as follows: current non-use of cigarettes or nicotine vapes, current nicotine vape vaping only, current cigarette smoking only, or current dual use of cigarettes and nicotine vapes.
We found no significant interaction effects. Main effects models revealed that message framing was associated with PE-quit (
We found no significant interaction effects. Main effects models revealed that message framing was associated with PE-switch (
We observed a significant interaction between message framing and current nicotine or tobacco use status (
This mixed methods formative study examined young adults’ responses to standard and comparative candidate messages about cigarette smoking and nicotine vaping, including variation by LGBTQ+ identity and product use status. Overall, both standard and comparative candidate messages were perceived as more effective than regulatory referent messages for influencing smoking-related perceptions, with standard messages generally performing best across outcomes. Psychological reactance was low across message types, suggesting limited counterarguing in response to these text-only stimuli [
Importantly, evidence supporting the relative effectiveness of comparative risk messaging was limited and outcome-specific. While comparative messages were associated with higher PE for influencing smoking among young adults reporting current dual use, comparative messages did not demonstrate greater effectiveness than standard messages overall and were less effective for influencing vaping-related perceptions. Our findings suggest that comparative messaging may influence smoking-related perceptions among specific subgroups but do not provide clear evidence of consistent advantages over standard risk messaging.
Qualitative findings provide important context for interpreting these patterns. Participants expressed mixed reactions to comparative messaging. Some viewed these messages as useful for harm minimization, particularly in encouraging reduced cigarette use or substitution of cigarettes with vapes. However, others raised concerns that comparative framing may unintentionally normalize vaping, sustain nicotine use, or be misinterpreted as endorsing vaping as “safe” rather than “lower harm than cigarettes.” These concerns highlight the importance of developing health messages that clearly distinguish between “lower harm” and “no harm” and underscore the need for careful testing and implementation of comparative risk messages [
Participants also highlighted the importance of message novelty and content specificity. While messages describing well-established harms (ie, “known truths”) performed well in quantitative analyses, qualitative data suggested that repeated exposure to familiar health risks may reduce engagement among young adults. Messages presenting less familiar information (eg, hypertension and toxic constituents) were perceived as more engaging and credible. Information about nonrespiratory health risks shared by smoking and vaping was particularly “shocking” to participants, especially among LGBTQ+ young adults. This finding was surprising; while fear appeals have commonly been used in tobacco public education campaigns and found to be effective [
Efficacy messaging also emerged as a critical influence in shaping message receptivity. Qualitatively, participants preferred messages that acknowledged their ambivalence about quitting all nicotine and provided actionable guidance in a nondirective tone. Encouraging, autonomy-supportive language was viewed as more acceptable than directive or prescriptive messaging, consistent with prior research [
Our study also highlights the importance of considering developmental differences within the young adult population. Both qualitative findings and quantitative analyses indicated meaningful differences across age groups, with emerging young adults (age 18‐25 years) demonstrating distinct preferences for message tone (encouraging) and content (novel and short-term harms and “off-ramp” messaging). Additionally, younger participants exhibited higher levels of psychological reactance, suggesting that developmental stage may shape how young adults interpret and respond to tobacco risk messages. Prior evidence suggests that emerging young adults may respond to health messages and warnings differently than older, more established young adults [
A key strength of this formative research was the systematic development and testing of candidate messages across qualitative and quantitative phases, allowing triangulation of findings. Our focus on LGBTQ+ young adults supports equity-oriented tobacco research by identifying messaging strategies that may be uniquely salient for this group. Although quantitative analyses did not find evidence of differences by LGBTQ+ status, qualitative data revealed unique thematic patterns. This divergence both underscores the value of qualitative methods for capturing identity-specific nuances and highlights considerations for further investigation. For example, it is possible that standard and comparative messaging approaches resonate broadly across young adults (regardless of LGBTQ+ identity), which could be experimentally tested with larger samples that support subgroup analyses.
Limitations should be considered when interpreting findings. First, convenience sampling through online platforms may have resulted in a young adult sample that is more engaged with or receptive to health research, potentially elevating PME and attenuating reactance relative to broader or higher-risk populations. However, consistency of themes across qualitative and quantitative phases suggests that underlying mechanisms (ie, how young adults process messages) are robust. Second, although mixed-effects models accounted for the incomplete-block design, message-level estimates are constrained by the number of observations and formative nature of the study. Third, our sample included young adults who reported ever smoking and vaping; future studies should use representative samples and focus on at-risk user groups (eg, nicotine naïve users) to better assess population-level effects, including unintended consequences of exposure to comparative messages. Given documented increases in dual use among young people [
This mixed methods formative study contributes to understanding how young adults perceive risk and efficacy messages about cigarette smoking and nicotine vaping. Findings suggest that standard risk messaging performs consistently well across outcomes, while evidence supporting comparative risk messaging is more limited and context-specific. Comparative messages may influence smoking-related perceptions among young adults engaged in dual use, but may also introduce unintended consequences, including normalization of vaping or sustained nicotine use. Regardless of framing, participants responded most positively to messages that conveyed credible and specific health harms, especially novel or less familiar information (eg, heart disease and consequences of occasional use). At the same time, qualitative findings highlight the importance of pairing risk information with clear and actionable efficacy messaging that acknowledges ambivalence about quitting and specific behavioral recommendations. Taken together, findings suggest that the health communications targeting young adults at risk for dual use should prioritize clarity, accuracy, and careful framing of harm information. Campaigns should clearly distinguish “lower harm” from “no harm,” while supporting young adults’ autonomy through informed decision-making. As noncombusted nicotine products continue to enter the tobacco marketplace, future experimental research is needed to assess the behavioral impact of comparative risk messaging, particularly among nicotine-naïve young adult populations, and to evaluate how message effects vary across developmental stages and user groups.
Thank you to Elle Elson and Sydney Galusha for your important contributions to project coding.
The authors declare the use of generative AI (GAI). According to the GAIDeT taxonomy (2025), the following tasks were delegated to GAI tools under full human supervision: proofreading and editing. The GAI tool used was Microsoft Copilot. Copilot was used at the final stage of revision to proofread the manuscript using the following prompt: “Proofread the attached file for formatting and grammatical errors and generate recommendations.” JGP evaluated the large language model–generated recommendations before reviewing and editing written content for formatting, spelling, and grammar. Responsibility for the final manuscript lies entirely with the authors. GAI tools are not listed as authors and do not bear responsibility for the final outcomes.
Research reported in this publication was funded by the National Cancer Institute of the National Institutes of Health (NIH) and the US Food and Drug Administration (FDA) Center for Tobacco Products (awards K99CA260718 and R00CA260718; principal investigator: 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 Ohio State University College of Public Health.
The datasets generated or analyzed during this study are available from the corresponding author on reasonable request. Custom code is available upon request.
Conceptualization: JGP, AKF
Formal analysis: EEJ, GT, ACE, HC, JGP
Funding acquisition: JGP
Investigation: JGP
Writing—original draft: EEJ, GT, ACE, JGP
Writing—reviewing and editing: EEJ, GT, ACE, HC, DM, PAM, MDS, TW, AKF, JGP
None declared.
Food and Drug Administration
lesbian, gay, bisexual, transgender, and queer
likelihood ratio
perceived effectiveness
perceived message effectiveness
Messages shown to focus group participants and messages developed after focus groups and tested in interviews and online survey.