To account for differences in patterns of drinking and broad sociocultural factors,21 focal analyses were stratified by genetically inferred continental groups (ie, EA and AA). These groupings broadly aligned with self-identified race and serve as a proxy stratifying variable.20 Separate models were fit for parental separation and parental relationship discord in view of prior evidence that parental separation and quality of the parental relationship exert separate and unique effects on children’s adjustment.40 For all a priori hypotheses, we used a p-value threshold of p < .05 for inference criteria.
