Regression analyses were used in Models 1a and 1b in Table 2 to examine prediction of early initiation of substance use (at wave 3). We first checked for potential multicollinearity among variables. VIF scores ranged between 1.00 for early initiation of substance use SES-related risk and 1.07 for admixture proportion, and all measures of tolerance were above .90, suggesting that multicollinearity was not present. To examine the shape of the interaction effect of SAAF with genotype on early initiation of substance use, we first regressed the early initiation of substance use score at age 13 for each participant on SAAF participation (0 vs. 1), 5-HTTLPR genotype (s allele carriers = 1; others = 0), wave 1 substance use, and the control variables of gender and SES risk exposure. We then examined the moderating effect of 5-HTTLPR by entering the interaction of SAAF and genotype in the second step of the regression. As can be seen in Table 2, model 1a, the main effects for SAAF and genotype were nonsignificant. However, in model 1b, the interaction of SAAF participation with the
