Six time points of repeated-measures data were available for both life events and CBCL externalizing. Two time points of repeated-measures data were available for TRF externalizing. Accordingly, we adopted a linear mixed modeling (LMM) approach to test for an indirect effect of parental externalizing problems (antisocial behavior and alcohol problems) on subsequent adolescents’ externalizing that is transmitted through life events, and to test whether life events and adolescents’ GABRA2 genotype interacted to qualify this indirect effect. Linear mixed modeling allows for the incorporation of missing data and models the covariance structure with fewer parameters relative to a repeated-measures ANOVA. We used restricted maximum likelihood estimation, a continuous autoregressive correlation structure, and random intercepts and slopes. Parental antisocial behavior and alcohol problems were run in separate models, as were CBCL and TRF externalizing outcomes.
