As COGA is a family-based study, cluster corrected standard errors were computed to account for the non-independence of these observations. All analyses were stratified by ancestry and age group, such that associations for EA adolescents (12–17 years old), EA young adults (18–31 years old), AA adolescents, and AA young adults were all analyzed separately. As epidemiologic (Eme, 2016) and neurophysiological studies (Iacono, Malone, & McGue, 2003; Porjesz & Begleiter, 1998) have found sex differences such that males endorse higher levels of externalizing behaviors and have smaller P3 amplitudes in comparison to females, sex was included as a covariate in all analyses. Follow up analyses including the interactive effects of sex were also performed. To test interactive effects relevant to Hypothesis 1 and 3, EXT PGS by sex, EXT PGS by age, and age by sex interaction terms were added to the base models (Keller, 2014). To test the interactive effects relevant to Hypothesis 2, the sex by P3 interaction term was added to the base model.
