Offspring alcohol use phenotypes were regressed on mediator variables, the transmitted component of parental polygenic scores, and a series of exogenous covariates (offspring sex, offspring age at last assessment, and offspring 10 within-ancestry principal components). Mediator variables were regressed on the transmitted and nontransmitted components of the parental polygenic scores and the covariates, except offspring sex. Offspring birth cohort was included as a covariate in the models examining parental marital discord as a mediator, but excluded from models examining parental divorce because of problems with model convergence. Parental alcohol use phenotypes were regressed on the corresponding parental polygenic scores (maternal phenotype on maternal polygenic score and paternal phenotype on paternal polygenic score) and the exogenous covariates, except offspring sex. Parental relationship variables (i.e., divorce and discord in the respective models) were regressed on all parental polygenic scores.
