scores) approximate the upper one-sixth of the distribution of intrapair RAPI differences across all ∼2,500 twin pairs in the epidemiological sample from which they were drawn. Similar, but substantially smaller, effects were found for individual RAPI and Mm-MAST scores and their intrapair differences at age 25. EDAC selection was clearly effective, for it yielded a selected sample of twins with markedly elevated rates of diagnosed alcohol dependency or abuse (51.6% of the 358 individual twins selected for ERP study), a sample unusually informative for within-family study of effects of adolescent alcohol exposure, albeit a selected sample not fully representative of the base population from which it was drawn. For all analyses, we used all twins from twin pairs in which both co-twins had usable ERP data, necessarily deleting pairs in which 1 or both co-twins had unusable ERP results. After pairwise deletion, the analyzed samples were the same for both correlational and HMR analyses with realized samples of 152 twin pairs for P2, 147 pairs for novel P3, and for target P3, for which SNR is smaller, a sample of 108 twin pairs; 90% of the pairs with usable ERP data were pairs for whom we had all of the
