12-29-2016 01:45 PM
I am working with SAS NLMIXED to run a bivariate-GLMM using two correlated random-effects to join the two responses together.
I’ve chose the Gaussian-Hermite (G-H) method for approximating the likelihood. I know the G-H formula for the univariate case but I couldn’t find a proper formula for the bivariate model, when I have two responses together.
I want to know which “G-H formula” SAS NLMIXED is using to approximate the likelihood function for bivariate regression models.
I appreciate your responses.