Among EA families, prevalence of parental separation ranged from 10% (0-20%ile of propensity score distribution) to 79% (>80%ile of propensity score distribution). Within-quintile comparisons of EA family background characteristics are presented in Table 4, with multivariate results provided in Table S23. In summary, parental alcoholism (maternal or paternal) is nearly absent in the lowest quintile of predicted probability of parental separation, but a dominant factor in the highest quintile. A similar pattern is observed for maternal smoking and both maternal age at twins’ birth and paternal education, where very young mothers and fathers with less than high school education were over-represented among higher risk quintiles. Given excellent matching on family background, within-quintile comparisons of twin substance involvement by parental separation were conducted. As shown in Table 5, parental separation in EA families continued to predict riskier twin outcomes across the risk spectrum. In most cases, differences within strata reached statistical significance with the single odd exception of alcohol use in 61-80th percentile families, where rates of early alcohol use were unusually low in separated families.
