Computations for the familial analysis of binary traits (Q2488411)

Computer programs for estimation of the regression and correlation parameters in the multivariate logit regression model are discussed. Implementations of the maximum likelihood (MLE) and the generalized estimating equations (GEE2) estimates are compared. These programs are applied to data which simulate a genetic disease distribution in a family with three generations. The authors specially stress on the use of automatic differentiation software (www.autodiff.org).