In baseline models, we first included control variables (i.e., child gender and parent education), study membership (i.e., a dummy variable with MLS=0 and AFDP=1) and interactions between control variables, study membership and the age-indicated time trends. This was a conservative strategy and we thus trimmed non-significant interactions for subsequent analyses. The interaction between one of the linear time trends (change between age 7 and 17) and study membership was significant for father-reported child externalizing symptoms and was thus retained. This interaction indicated that although externalizing scores decrease during childhood from age 2 to 7, this decrease slowed after age 7 to a greater extent in MLS than in AFDP (likely because AFDP contributed 11–17 year olds only to this analysis). As reported in Tables 2 and 3 in Model 1, we then added the three types of effects (i.e., distal, proximal, and time-varying) for each parent’s alcohol-related symptoms as well as interactions between these effects of parent alcoholism and study membership. Interactions including study membership were again trimmed when non-significant. Model 2 tested whether there were gender differences in the
