In the next model all variance components were constrained to be the same in men and women. We tested whether variance components due to A and C were significantly different from zero. Finally, γ, the genetic correlation in DOS twins, was constrained at 0.5 (or φ , the shared environmental correlation, was fixed to 1). Significance of the parameters was tested by comparing the fit of the nested models to the fit of less restricted models. Goodness-of-fit of the sub models was assessed by likelihood-ratio test. The difference in log-likelihoods between the nested models follow a χ2 distribution, with degrees of freedom (df) equal to the diference in the number of paramaters in the two models. According to the principle of parsimony, models with fewer parameters are preferred if they do not give a significant deterioration of the fit (p>0.01).
