Our analyses adjusted for sociodemographic variables, but not for other correlates of early onset. Dawson and colleagues (2007) reported stress-related differences in ethanol intake for each onset group after adjusting for sociodemographic characteristics, and also after adjusting for family history of alcoholism, personal history of alcohol dependence and other psychopathology, as well as the interaction of stressors with covariates. As mentioned earlier, the risk of early drinking onset on subsequent consumption is often thought to be transmitted via a constellation of individual, familial, and genetic risk factors (e.g., Jessor and Jessor, 1977; McGue et al., 2001a,b; Pitkänen et al., 2008). Consistent with this hypothesis, adjusting for these variables in the NESARC study decreased the effect size associated with drinking onset age, but the stress-onset interaction effect remained. The interaction observed in the NESARC sample appeared to be largely based on high levels of past-year drinking among individuals in the youngest onset group who reported 6 or more past-year stressors. Individuals in this subgroup reported drinking on average 3.5 oz of alcohol per day (i.e., ~6 drinks), much more than early-onset
