Although our study is one of few to examine the separate effects of parental separation versus parental alcoholism on offspring substance involvement, there are a number of limitations. First, parental alcoholism was modeled without regard to onset or remission. Our results likely underestimate risks for offspring exposed to chronic parental alcoholism to the extent that some parents no longer met criteria during childrearing years. We also did not examine potential mediators of risk from parental separation or parental alcoholism, such as compromised parenting behavior (e.g., lax or inconsistent monitoring) and presence of a stepparent, which will be an important focus of future work. Follow-up of twins as they age into periods of highest risk of substance dependence is also planned. For African Americans, in addition to reduced interpretability of survival-analytic findings, we have limited statistical power because of a much smaller sample of African relative to European ancestry twins, which may have (i) obscured potentially important age interactions in survival models and (ii) contributed to inadequate numbers of intact African American families across the range of risk propensity for parental
