Multidimensional Predictors of Substance Dependence

All statistical analyses were conducted using R version 4.2.1; there were no missing data in the analyses. To compare participants regardless of their preferred nicotine consumption method (cigarettes or e-cigarettes), a composite nicotine dependence variable was created by taking the highest value of the FTCD and e-FTCD scores; this gave each participant a single nicotine dependence score. Age of onset of regular use for each of the substances was calculated by subtracting reported years of regular use from current age; a composite nicotine age of onset was created by taking the lowest age between cigarette and e-cigarette age of onset variables. Three different hierarchical multiple regressions were conducted, in which dependent variables were levels of dependence on alcohol, cannabis, and nicotine, respectively. In the regression analyses, predictor variables were added in four stages: the first models contained demographic variables (age, sex, and education), the second models contained personality variables (hopelessness, anxiety sensitivity, impulsivity, and sensation seeking), the third models contained ages of onset of regular use across substances, and the fourth models contained dependence on the other two substances. Regarding nicotine dependence, the fourth model included an additional variable: dual use of cigarettes and e-cigarettes. The variables were added in this order as it was the most chronologically plausible order, as demographic variables are from birth, personality is mostly stable from childhood, and age of first use comes before substance dependence. Finally, post-hoc analyses were conducted on significant variables in the final models for each of the regressions. Participants were categorized by dependence on each substance (dependent vs. non-dependent) using the cut-offs described in Section 2.2. For continuous variables, the means of each group were compared via Welch's t-tests; for categorical variables (e.g., sex and dual use of cigarettes/e-cigarettes), the ratios were compared via Pearson's χ² tests.