Since individuals often use multiple substances simultaneously, the bivariate association between a given substance use behavior and depressive symptoms may be confounded by concurrent substance use behaviors. Therefore, we also estimated a multivariate TVEM in which daily smoking, marijuana use, and regular HED simultaneously predicted depressive symptoms in order to estimate age-varying associations while adjusting for concurrent substance use behaviors (Fig. 4). Generally, the results from the multivariate model were similar to the results from the bivariate models, with the adjusted associations somewhat attenuated. However, in the adjusted analysis, there were no significant gender differences at any age for any of the three substance use behaviors. When adjusting for marijuana use and regular HED, daily smoking was significantly associated with higher depression scores for females during ages 12.5 to 31, and for males during ages 14.5 to 20.5 and 24.5 to 30.5. As in the bivariate model, this association was strongest during adolescence for both males and females, peaking at age 12.5 for females (1.8 point higher mean depression score, 95% CI = [0.05, 3.5]) and at age 16.5 for
