Fifth, we considered whether the associations between AUD and marital outcomes were robust to or modified by polygenic loading for alcohol problems. Polygenic scores provide a useful global marker of genetic risk, though we recognize that at present polygenic scores generally account for a small amount of variance in alcohol phenotypes (Kranzler et al., 2019; Sanchez-Roige et al., 2018; Walters et al., 2018). Sixth, in the mixed effect Cox regression analyses of predicting marriage, several violations of the proportional hazards assumptions were detected. In the presence of a violation of proportional hazards assumptions, the estimated coefficient is interpretable as an average effect over time, although the coefficient can be biased to the extent that the slope of plotted residuals across time varies in magnitude and direction (Xu & Gamst, 2007). Seventh, we explored and did not find any evidence of sex differences in the pattern of results. However, likelihood of marriage and divorce were higher among females compared to males, which mirrors national data (Aughinbaugh et al., 2013).
