Approximately 85% of the younger-cohort intake sample completed their age-17 follow-up assessment. Mothers of younger-cohort subjects who did not participate at the age-17 assessment were more likely to have alcohol dependence than mothers of participating subjects (22.1% vs. 9.6%; χ21 = 12.1, p < .001) and were more likely to have consumed more drinks at one time, F(1, 736) = 6.44, p < .05. Thus, high-risk adolescents were somewhat underrepresented in the second follow-up assessment, indicating some attrition bias in the younger cohort. However, the assessment of older-cohort twins was their intake assessment; by definition there is no attrition bias. Because our analytic strategy was predicated on obtaining replicated effects across both cohorts, the overall effect of any bias in the younger cohort is likely to be minimal.
