A third limitation is that the measures of life events used in both samples tapped events that were in the normative range of the types of changes and transitions that adolescents and their families are likely to face. Whether the same indirect and gene-environment interaction effects would be observed in the context of more extreme stressors (e.g., physical, emotional, or sexual abuse) remains an important direction for future research. Fourth, we are unable to account for parental GABRA2 genotype in the present samples. Fifth, it would have been ideal to use measures of parental antisocial behavior and alcohol problems in the CDP sample that were obtained prior to mothers’ and teachers’ first reports of their adolescent’s externalizing behavior; however these data were not available. In view of evidence that antisocial behavior and alcohol problems are relatively stable in adulthood (Bennett, McCrady, Johnson, & Pandina, 1999; Neumann, Wampler, Taylor, Blonigen, & Iacono, 2011; Odgers et al., 2008; Petersen et al., in press; Pitkanen et al., 2008), we believe it is appropriate to conceptualize the measures as global indices of parental externalizing
