In Model III, controlling for demographic characteristics, coparent substance use/dependence, and psychiatric comorbidities of both parents and offspring, risk to Groups 2a, 2b and 2c could be equated. While offspring of unaffected parents with an alcoholic or cannabis dependent identical cotwin were at 1.76 times higher risk of early drinking, and without age interaction, risk to Groups 1, 2 and 3 did not significantly differ in post-hoc tests. Parental separation remained a strong predictor of early drinking in Model III, associated with 13.04 times higher likelihood of drinking through age 10, and 2.22 times higher likelihood from ages 11–13; from age 14 onwards, parental separation was associated with 41% higher likelihood of drinking.
