Results predicting adolescent externalizing behaviors are presented in Table 1. The level-1 predictors in the model accounted for approximately 22% of the variance in externalizing t-scores. The first-order effects suggested that high levels of peer delinquency were prospectively associated with increases in externalizing behavior 3 years later. There was no evidence for first-order effects of covariates or GABRA2 on externalizing t-scores. There was evidence for a statistically reliable GABRA2 × Peer delinquency interaction term. As depicted in Figure1, the simple slope of peer delinquency on externalizing behavior was statistically significant for both A-carriers (β = 1.85, P < 0.001) and adolescents with the GG genotype (β = 4.65, P < 0.001) although this association was stronger for adolescents with the GG genotype. At low levels of peer delinquency, adolescents with the risk genotype (GG) had lower rates of externalizing behavior compared to A-carriers. In contrast, at high levels of peer delinquency, adolescents with the risk genotype, GG, had greater rates of externalizing behavior compared to A-carriers. The lower and upper bounds of regions of significance were −9.60 and −0.25, respectively.
