(StataCorp, 2011). The negative binomial analyses comprised two models: (a) a main-effects model involving DRD4 status, intervention status (SAAF–T or control), and gender, and (b) tests of two- and three-way interactions involving the predictors in the main effects model. In these models, SAAF–T, DRD4, and their interaction predicted the logarithm of the mean count [log(μ)] in substance use. Thus, the estimated coefficients are in the log(μ) metric, meaning one unit change in a predictor corresponds to one unit change in log(μ). Exponentiation of the coefficients (eβ) provides the incident rate ratios (IRRs), which represent the increase or decrease in the frequency of substance use with each one-unit change in a predictor in the model. For interpretation purposes, percentage of change in IRR with each one-unit change in a predictor is calculated by first subtracting the IRR from 1 and then multiplying it by 100 [100*(1 - eβ)]. In the current study, the percentage of change was used as the index of effect size.
