2.4. Statistical Analysis

All analyses were conducted using Stata version 15.1 (StataCorp LLC, College Station, TX, USA) to obtain weighted point estimates and 95% confidence intervals for baseline sample characteristics, past 30-day e-cigarette use at follow-up, and percentage exposed to various channels of e-cigarette advertising and content. Associations among exposure to baseline advertising/product content, cigarette smoking, ever use of e-cigarettes, participant characteristics and past 30-day e-cigarette use at follow-up were measured by weighted multivariable logistic regression using svy: logit command in Stata.

The combined data set of AmeriSpeak and KnowledgePanel contained study-specific base sampling weights derived using a combination of the final panel weight and the probability of selection associated with the sampled panel member. These weights were adjusted to account for survey non-respondents to decrease potential nonresponse bias associated with sampled panel members who did not complete the survey interview for the study. The nonresponse adjusted survey weights for the study were then adjusted via a raking ratio method to external population totals (using Census data) associated with the following socio-demographic characteristics: age, sex, education, race/Hispanic ethnicity, and Census Division, to create the final study weights. Raking and re-raking were done during the weighting process such that the weighted demographic distribution of the survey completely resembled the demographic distribution in the target population. Final study weights were raked to external benchmarks on “ever use of e-cigarettes”, using the 2017 National Youth Tobacco Survey for the teen, and the 2017 National Health Interview Survey for the adult benchmarks of e-cigarette use, respectively, creating a study-specific-post-stratification weight variable. This weight variable was used in all analyses to adjust for sources of sampling and non-sampling error and allow the generalizability of the results to a national level.