The extent of correlated change in personality and problematic alcohol involvement was examined in a series of bivariate LGMs following the specifications described earlier. On the basis of the results of the univariate models, all bivariate LGMs consisted of an intercept and a nonlinear slope (i.e., the slope loading for the last assessment was freely estimated) for each construct of interest. To enhance interpretability, the freely estimated slope loadings (at the final measurement occasion) for the respective alcohol and personality constructs in each bivariate growth model were constrained to be equal. Furthermore, FH and sex were included as exogenous variables (i.e., exerting a direct influence on all study constructs) in all models.
