The sub-models tested and their corresponding model fit statistics are reported in Table 3. Models 1 and 2 were full Cholesky ACE and ADE models respectively. Parameter estimates for women and men were equated in these models. Because the log-likelihood of the ACE model was marginally better than that for the ADE model, the ACE model (Model 1) was selected for further examination. Model 3 tested an AE model with parameter estimates equated for women and men. This model resulted in a non-significant decrement in fit compared to Model 1, indicating that shared environmental influences were not significant. Model 3 therefore became the base model for additional analyses. In Model 4, we tested whether the parameters for women and men should be estimated separately. The improvement in fit when women and men were estimated separately was not significant, indicating that the parameter estimates can be equated across sex; Model 3 was therefore selected as the best-fitting model. (We were unable to estimate confidence intervals due to the computational intensity of the calculations for the models, so we present only the
