Participant characteristics
Based on initial recruitment criteria, 1,472 participants qualified for the online survey, but complete data was only available from 620 individuals. Most completers reported vaping at least once per week, with 4.9% using eCigs less than once per week. Individuals who vaped less than once per week were excluded from further analyses, resulting in a final n of 587, that was 64.8% white and 47.5% male. Additional demographics are provided in Table 1. The average PSECDI score across all participants was 8 ± 2.83, which falls in the low dependence category16. The means for the various WISDM subscales are shown in Table 1. These means are roughly similar to those reported elsewhere for light and heavy daily smokers18, except for the weight control subscale, which was roughly 1 point higher. Still, this apparent difference was not reflected in the conjoint data reported below.
Top line results across the entire cohort (n = 587)
Overall, flavor was the most important attribute when choosing an eCig product, as shown in Fig. 1. Utility scores for individual flavors are depicted in Fig. 2. A larger, positive utility score indicates that specific element had a larger, positive impact on preference. Likewise, a larger, negative utility score indicates that specific element had a larger, negative impact on preference. Confectionery (Apple Pie, Birthday Cake) and fruit (Cherry, Mango) flavors had the highest utility scores, whereas classic menthol and tobacco flavors were the lowest. Notably, the utility score for Menthol was close to zero, and Cucumber Mint, Cuban Cigar, and Classic Tobacco all had negative utility scores for this cohort of young adults.
Utility scores of individual elements across all participants, ordered vertically by importance of each category. Elements with positive scores are to the right, and negative scores are to the left.
Relative utility scores for product messaging deviated substantially from our original hypotheses. We had expected appetite suppression and satisfaction of food cravings would be major drivers of product selection. However, as shown in Fig. 2, these were among the lowest utility scores, along with “produces big vapor clouds.” Conversely, “relaxes you” and “reduces stress” had the highest utility scores, which is generally consistent with data from the nicotine level and health warning categories that are discussed below. Collectively, these data suggest user beliefs about expected outcomes (accurate or not) play a major role in electronic cigarette desirability, consistent with prior work on other tobacco products.
Contrary to expectations, the overall importance of nicotine level across the entire cohort was surprisingly small, at least relative to flavor (see Fig. 1). Relative utility scores for individual nicotine levels are shown in Fig. 2. Notably, the strongly negative score for 0 mg/ml nicotine indicates that, as a whole, the young adults studied here do not vape simply for flavor, and some amount of nicotine is desired. Notably, the highest nicotine levels also show negative utility scores, presumably due to harshness that occurs with high amounts of nicotine (e.g.,8,19,20).
Results from the health warning category are consistent with data above for 0 mg/ml products. That is, as Fig. 2 shows, the utility score for the health warning “This product contains nicotine, an addictive chemical” has a positive utility score. This suggests young adults who vape regularly want nicotine in the products they choose. While intuitive, a positive utility score of a message nominally intended as a health warning is a critical insight—it implies that a well-meaning message intended to inform buyers of the deleterious health impact of eCigs is actually viewed as a positive product attribute. Conversely, the “Not for sale to minors” warning has a negative utility score, despite our sample consisting of individuals aged 18 to 30.
Regarding device style, the magnitude of utility scores was relatively small in either direction. We had expected the highest utility scores for “Compact, does not look like a cigarette (boxy, flat design)”—that is, devices like a Juul. Instead, the option “Compact, looks like a cigarette (round, long design with glowing tip)” had the highest score (+ 17.8 vs. − 7.2 for flat boxy designs). The other two device styles also had negative utility scores (see Fig. 2). Still, the overall attribute importance for device style is quite small (4.4% in Fig. 1) in 18-to-30-year old adults and should not be overinterpreted. Whether device style may be more important in other age groups remains to be determined.
Last, across all 587 young adults, the least important driver of product selection was purchase location (see Fig. 1). As might be expected, the relative utility scores for these individual elements were quite small (see Fig. 2), although the scores did go in opposite directions: “Available in convenience stores” was positive (+ 13.75), while “Available in smoke shops” (− 4.62) and “Available online” (− 9.13) were negative.
Segmentation of participants by gender
Since we originally hypothesized women would give higher utility scores to products that may facilitate weight control, we separated the cohort by gender and re-ran the logit model in the Sawtooth Discover platform. We focused on the top three product attributes: flavor, product messaging, and nicotine level (see Fig. 1). In the overall cohort, flavor was most important (48.1%), followed by product messaging (21.0%), then nicotine level (15.3%); for the women in our sample, these attributes’ importances were rated in the same order: flavor (55.5%), messaging (19.4%), and nicotine level (12.2%). Interestingly, men rated nicotine level and flavor as roughly equally important (27.5% vs. 26.7%, respectively), with product messaging being third (19.8%). These data suggest women and men may have different motivations for selecting eCig products.
Women and men were fairly similar in their flavor preferences, though a few key differences became apparent. Women found specific flavors more polarizing, with utility scores ranging from -216.24 for Cuban Cigar to + 116.62 for Birthday Cake; conversely, men did not feel as strongly about flavor, with utility scores ranging from − 97.16 for Cuban Cigar to + 62.95 for Birthday Cake (see Supplemental Fig. 2, left panel). Still, both men and women gave negative ratings to Cuban Cigar, Classic Tobacco, and Cucumber Mint; Menthol was slightly negatively rated by women and slightly positively rated by men. Cherry, Apple Pie, Mango, and Birthday Cake were all positively rated, and women rated the latter three as more positive than did men. Because the magnitude of women’s utility scores for flavor were larger than for men, it appears women are more motivated by flavor.
Next, we observed gender differences in product messaging. Both women and men were interested in the messaging “Relaxes you” and “Reduces stress,” though men valued the latter as more important than women (+ 53.86 vs. + 36.47). Men had a positive score (+ 44.26) for “Helps you focus,” while this was scored negatively by women (− 9.99). In our cohort, neither women nor men were interested in appetite suppression, satisfaction of food cravings, or production of large vapor clouds, as both groups gave negative utility scores to these product claims. These scores (shown in Supplement Fig. 2, right panel) suggest that particular product messaging may appeal more to women over men, or vice versa.
We next examined differences in nicotine level utility scores by gender (Fig. 3). Both men and women had positive utility scores for 6 mg/ml nicotine (men + 68.05, women + 33.28). Interestingly, women also had a positive utility score for 3 mg/ml nicotine (+ 37.65), while this had a negative scores for the men (− 9.04); conversely, women had negative scores for 12 mg/ml and 18 mg/ml nicotine (− 28.73 and − 35.26, respectively), while men had positive scores (+ 32.28 for 12 mg/ml and + 5.56 for 18 mg/ml). For both, 0 mg/ml nicotine had negative scores, with men having a substantially lower score than women. These findings suggest women generally prefer lower nicotine levels (optimal of 3 to 6 mg/ml nicotine) when compared to men (optimal of 6 to 18 mg/ml nicotine).
Utility scores for the five nicotine levels provided within the choice task. When split by gender, the preferred nicotine level for women was lower than the preferred level chosen by men. Elements with positive scores are shown above the dotted line, while negative scores fall below the dotted line.
For young adults studied here, men and women only had positive utility scores for the cig-a-like design (“Compact, looks like a cigarette (round, long design with glowing tip)”) while the other three device styles were scored negatively (boxy, flat compact design and both customizable eCig rigs). Men were interested in purchasing online or in convenience stores, but not in smoke shops; women were interested in purchasing in convenience stores or in smoke shops, but not online (not shown). Finally, the health warning ‘Not for sale to minors’ was rated negatively by both groups.
Taken together, segmenting our data by gender reveals differences in overall importance of eCig attributes in men and women, with varied preferences for specific product elements within those attributes.
