We conducted a series of analyses guided by the framework of Baron and Kenny (1986) for assessing whether one variable mediates the association between two others. This allowed us to determine whether any effects of maternal maximum consumption might be explained, or fully mediated, by effects of maternal alcoholism. We first determined whether maternal maximum drinks and alcohol dependence separately were significantly associated with each offspring outcome measure. Analyses included offspring gender to adjust for sex differences in the outcome measures. We next included an interaction between maternal characteristic and offspring gender to assess whether gender moderates associations between maternal predictor and offspring outcome. If not significant, this term was dropped. The third step included both maternal characteristics as predictors. This allowed us to determine whether maternal alcohol dependence might explain any significant associations between maternal maximum consumption and offspring characteristics. If so, then significant associations between maternal maximum drinks and offspring characteristics should become nonsignificant net the effects of maternal alcohol dependence. We used the same approach to determine whether measures of drinking during pregnancy might account for any associations we found.
