The univariate models for the timing of first alcohol use and AD are shown in Table 2. For each phenotype, the full model, which included separate estimates for females and males, is shown above the final model. For the timing of first alcohol use, equating A, C, and E across females and males did not result in a significant difference in model fit (Δχ2 (1)=0.87; p=0.83). A deterioration in model fit was observed both when the A pathway was dropped (Δχ2 (1)= 23.61; p<0.001) and when the C pathway was dropped (Δχ2 (1)= 7.14; p<0.01). Thus the best-fitting model was an ACE model that equated A, C, and E values in females and males. Under this model, genetic influences accounted for 37% and shared environmental influences accounted for 14% of the variance in the initiation of alcohol use. Individual-specific factors (E) accounted for the remaining 49%. In the univariate alcohol dependence model, A, C, and E could be equated across sexes (Δχ2 (1)= 0.11; p=0.99). Dropping the C pathway did not result in a significant change in model fit (Δχ2
