We first examined bivariate age-varying associations between depressive symptoms and, respectively, daily smoking, marijuana use, and regular HED (Fig. 3). To this end, we implemented separate models in which each substance use behavior predicted depressive symptoms; models were estimated separately for males and females. For each substance use behavior, we observed a significant positive association with depressive symptoms during adolescence; this association persisted into early adulthood for daily smoking and marijuana use. Specifically, daily smoking was significantly associated with elevated depression scores from ages 12 to 31 for females and from ages 12.5 to 31 for males, with the strongest associations during early adolescence. Among 12-year old females, daily smokers had mean depression scores 4.3 points higher than non-daily smokers (95% CI = [2.6, 5.9]). This relationship peaked at age 13.5 for males: daily smokers reported mean depression scores 1.7 points higher than non-daily smokers (95% CI = [0.7, 2.8]). Some gender effects were observed, as the association among females was significantly stronger than the association for males from approximately age 14.5 to 16 and from 21 to 23. Marijuana
