into account the shared variances between fathers’ and mothers’ parenting (so that unique effects of fathers’ and mothers’ parenting can be examined), errors terms were correlated between fathers’ and mothers’ parenting subscales, as were the residual errors for mothers’ and fathers’ positive parenting latent factors. The residual errors for adolescent risky drinking and conduct problems were also allowed to be correlated. Standardized loadings for each indicator and correlated error terms for the latent constructs are presented in Table 3. Risky drinking and conduct problems were specified as categorical variables in the SEM model, and thus the robust weight least square estimator (WLSMV) was used by default in Mplus.
