In the path model, fathers’ and mothers’ ADS were specified as predictors of their own and the other parent’s parenting behaviors, as well as adolescents’ risky drinking and conduct problems. Fathers’ ADS and mothers’ ADS were specified as correlated. Fathers’ and mothers’ parenting behaviors were specified as predictors of adolescent risky drinking and conduct problems. Adolescents’ age, gender, and ethnicity, family structure, parental education, paternal/maternal antisocial behavior and internalizing problems were included as covariates to predict adolescent risky drinking and conduct problems. To evaluate indirect effects (e.g., father ADS → father parenting → adolescent risky drinking), we used the MODEL CONSTRAINT command in Mplus to specify new parameters that represent the multiplication of coefficients for the two specific paths (e.g., father → father parenting (a), and father parenting → adolescent risky drinking (b)) that comprise the indirect pathway. These new parameters (ab), which were tested along with other path coefficients in the model, provided a test of indirect effects. MODEL CONSTRAINT is the recommended method for estimating indirect effects in Mplus in the context of multiple imputation (Muthén & Muthén,
