Study staff visited all sessions of a required course to explain study objectives and distribute parental consent and student assent forms,how to cure cannabis followed by in-class administration of the electronic survey on computers 1–2 weeks later. As a programmed feature of the survey software , students were randomized at the participant level with equal probability to one of two discrete choice experiments: e-cigarettes or moist snuff smokeless tobacco. Participants randomized to the e-cigarette experiment were presented six pairs of randomly generated hypothetical e-cigarette products under a full factorial design. The composite products differed in device type , flavor , vapor cloud , and nicotine amount . Prior to viewing the computer generated composite e-cigarettes, participants were shown an image containing the possible e-cigarette product characteristics they might see . Participants randomized to the smokeless tobacco discrete choice experiment were presented six separate, consecutive pairs of randomly generated hypothetical moist snuff products under a full factorial design. The composite products differed in brand , flavor , cut , and price . Prior to viewing the computer-generated composite moist snuff products, participants were shown an image containing the possible moist snuff product characteristics they might see . The number of displayed characteristics and their levels were necessarily constrained to avoid excessive cognitive burden. Some characteristics were product specific . Prioritizing which characteristics to retain for each product was based on existing qualitative and quantitative literature on youth tobacco-related perceptions and use motivations. In each experiment, for each pair of product composites, participants were asked which product ‘are you more curious about’, which ‘is more dangerous to health’, which ‘would be easier to use’, and which ‘would give a bigger ‘buzz’ or ‘head rush’’. These outcomes were chosen because of previous work showing associations between tobacco use and/or susceptibility with youth-reported curiosity, perceived danger, and perceived ease-of-use. The outcome ‘buzz’ was introduced to measure perceived physiological effects or potency. Participants could select either composite product within the pair or ‘neither of these options’.

Supplementary file, Figure A3 shows an example question layout. Of 1052 eligible participants, 525 took part in the e-cigarette discrete choice experiment and 522 the moist snuff discrete choice experiment . Participants providing ‘straight-line’ responses with no variation in choosing the left hand-side or right hand-side product were excluded to improve data quality, leaving 495 in the e-cigarette experiment and 508 in the moist snuff experiment.Conditional logistic regression was used to quantify the independent contribution of product attributes to participants’ choices while maintaining the matching of each pair. The position of the composite product on the screen was also included in models to account for possible ordering preference. A positive regression coefficient indicates how much the attribute level in question increased the log-odds of that composite product being chosen relative to the whereas a smaller percentage used smokeless tobacco products . In the e-cigarette experiment , tank-type and pod-type devices garnered more curiosity and were perceived as easier to use than cigalike or dripmod devices. Relative to tobacco flavor, all flavors were associated with more curiosity, less perceived danger, and greater perceived ease-of-use. On the adjusted log-odds scale, where tobacco flavor is the reference, fruit and dessert were most positively associated with curiosity, while mint and unicorn were the flavor options most negatively associated with danger. Smaller vapor cloud e-cigarettes were viewed as less dangerous, offering less buzz, and easier to use. Nicotine amount was strongly associated with perceived danger and buzz. High nicotine devices were viewed with less curiosity, as more dangerous, delivering more buzz, and less easy to use, relative to low nicotine or nicotine-free devices . In the moist snuff experiment , one brand was perceived as the most dangerous but also the easiest to use. Relative to tobacco flavor, all moist snuff flavors were associated with more curiosity, less perceived danger, and greater perceived ease-of use. On the adjusted log-odds scale, fruit and mint flavors were the characteristic levels associated with the most curiosity, while fruit flavor was also viewed as offering the least buzz .

Associations of modest magnitude suggested that fine cut products were perceived as less dangerous and offering less buzz. Higher price products were viewed with more curiosity, as more dangerous, offering a greater buzz, and being less easy to use . Among all responses, the probability of choosing ‘neither of these options’ rather than selecting one of the two composite products varied by product and the question being asked. In the e-cigarette experiment, participants indicated ‘neither’ most often when asked about which of the two products they were more curious . ‘Neither’ was less often selected when asked about ease of use , buzz , and danger . Similarly,cure cannabis in the smokeless tobacco experiment, ‘neither’ was indicated most often when asked about curiosity , followed by ease-of-use , buzz , and danger . There was no statistically significant interaction by gender in the e-cigarette experiment . In contrast, having ever used an e-cigarette was associated with differences in all four perception outcomes . E-cigarette ever users held stronger perceptions about device types, viewing tank-type and pod-type devices with more curiosity relative to cigalike devices than did never users. Both e-cigarette ever and never users perceived flavored products with more curiosity and as easier to use compared to tobacco flavored products, but only never users believed that flavored products delivered less buzz than tobacco flavored e-cigarettes. Likewise, only never users were less curious about higher nicotine content e-cigarettes . In the moist snuff experiment, the direction and magnitude of associations were similar by gender, but there was nominally statistically significant interaction for the outcomes curiosity and danger, as male participants indicated more curiosity about higher price products . Only 44 smokeless tobacco ever users completed the moist snuff experiment, limiting statistical power to detect differences in association by product use. Generally, brand perceptions were stronger among smokeless tobacco ever users. Only never users viewed flavored products as offering less buzz and as easier to use, whereas only ever users associated pouched products as easier to use .This study provides quantitative evidence that specific characteristics of non-cigarette tobacco products independently shape how youth perceive these products. As hypothesized, for both e-cigarettes and moist snuff smokeless tobacco, flavored products were viewed with more curiosity and as being less dangerous, less potent and easier to use compared to non-flavored products. Associations of flavors with greater curiosity and ease of use and less perceived danger held for all non-tobacco flavors, including mint and wintergreen. Thus, evidence from this cross sectional study population suggests that mint varieties should be included alongside fruit and dessert in flavor restrictions intended to reduce youth tobacco use9 . Other product characteristics, such as e-cigarette device type, vapor amount and moist snuff price, also appear to shape product perceptions, which could inform tobacco control policy.

In January 2020, citing concern over growing youth e-cigarette use, the FDA announced a policy to prioritize enforcement of premarket authorization requirements for some types of e-cigarettes, but exempted mint and menthol flavors18. The present results suggest that youth perceive the properties of mint and wintergreen flavored tobacco similarly to fruit, dessert, and other flavors, which could undermine the effectiveness of the FDA policy. No such enforcement policy exists for conventional smokeless tobacco, but the present results demonstrate similar flavor association for moist snuff as observed for e-cigarettes. This finding is consistent with tobacco industry documents suggesting that flavored, lower priced, lower nicotine ‘starter products’ are used to target novice users before later ‘graduation’ to established use through a series of higher nicotine products. Use of flavored tobacco, including menthol, is more common among youth than adults8 and is the predominant way youth and young adults consumed tobacco across all tobacco products. A review of qualitative studies reported that flavored tobacco is viewed favorably by consumers, who associate flavors with less danger and often report that flavors contributed to their own tobacco experimentation and initiation10. Given the evidence that flavors contribute to youth use for all tobacco products8,9, current policies should consider not only restricting all non-tobacco flavors in cigarettes and e-cigarettes but in all forms of tobacco. Pod-type e-cigarettes, such as market-leading brand JUUL, have become the most popular type of e-cigarette among US youth21. Independent of nicotine content and flavor, participants in the present study viewed pod devices with more curiosity and as easier to use than other device types but not necessarily as less dangerous. However, participants also associated low nicotine content e-cigarettes with less danger, more curiosity, and greater ease of use. Research suggests that youth may not recognize that pod-type e-cigarettes contain nicotine at high concentrations; a misperception potentially reinforced by the nicotine amount printed on JUUL product packaging. Prior applications of discrete choice methods in tobacco control have focused on adults and/or recruited participants through online panels. Previously reported findings include a preference for non-tobacco flavors both in e-cigarettes and water pipe tobacco, as well as identifying warning labels as a factor reducing product interest. Consistent with the present study, Shang et al.reported that adolescents least prefer tobacco-flavored e-cigarettes and cigalike closed-system devices. The present study expands such work to moist snuff products, finds e-cigarette vapor cloud size and podtype devices as independent contributors to youth perceptions, and assesses the additional outcomes of perceived danger and ease of use. Furthermore, the present study shows no gender difference regarding e-cigarette perceptions, although beliefs associated with certain moist snuff product attributes were stronger among male participants, likely reflecting higher use and male-targeted marketing. Not all perceived qualities observed in the present study aligned with actual product properties. Participants correctly associated higher nicotine content with stronger physical effects . Other perceptions, such as flavored tobacco or small vapor cloud e-cigarettes being less dangerous, are not supported by scientific consensus. This discordance between perceived and actual effects represents a possible area for corrective public messaging or for greater regulatory vigilance for potentially misleading marketing practices. Of note, and not unexpectedly, product ever users held stronger perceptions about device type and brands than did product naïve participants. In a cross sectional setting, it cannot be distinguished to what extent experience shaped perceptions or that existing attention to product attributes contributed to use initiation.For example, danger could refer to either long-term or short term health effects. Meanwhile, ease of use could refer to concealability, access, adverse reactions, or social acceptance.