Self-reported demographic variables were obtained to adjust for potential confounds. Parental education was a dichotomous indicator of fathers and mothers who attended at least some college. Because education data were partially missing, a binary variable was constructed to identify missing data for both father and mother; this allowed a zero to be entered (rather than a missing indicator) so as to include the case in most computations, and permitted assessment of the relationship between missing education data and alcohol outcome. Mother’s report of family income was obtained and transformed into two binary income variables that represented the higher 20% and lower 20% income brackets. Offspring age identified those 21 years of age or older.
