The results of the model fitting procedure are summarized in Table 2. The baseline ACE model (Model 1) included the unique twin environment (T) and age regression parameters. Dropping the T parameters (Model 2) resulted in no deterioration in model fit compared to Model 1, ΔAIC = −2.5, χ2(2) = 1.53, p = 0.47. Constraining the male and female A, C, and E paths to be equal (Model 3) resulted in significant deterioration in fit compared to Model 2, ΔAIC = 24.9, χ2(2) = 28.82, p = 6 · 10−7. Dropping the age regression parameters (Models 4 and 5) resulted in significant deterioration in fit, ΔAIC = 64.1, χ2(6) = 76.10, p = 2 · 10−14 and ΔAIC = 64.5, χ2(6) = 76.42, p = 2 · 10−14, respectively. Thus, the best fitting model was an ACE model with age regression where the parameters were allowed to vary across sex (Model 2). For males, variance components [95% CIs] for this model were: A = 21.3% [7.6, 28.5]; C: 2.7% [0.00, 9.7]; E = 76.1% [71.5, 82.8]. For females, variance components [95% CIs] were: A = 31.0% [18.8, 37.1]; C = 2.1% [0.0, 8.5]; E = 66.9% [62.9, 72.9].
