i am trying to use mixed-effect logistic regression (using proc glimmix) to estimate variance of non-subject-level random error term, together with some subject-level random effects. trying to use something like: proc glimmix data=testdata; class somegroup; model y=x1 x2 / dist=binomial link=logit solution; random intercept / residual type=vc; ** or: ; ** random _residual_; ** getting some number; random intercept / subject=somegroup; run; proc glimmix returns estimate for the random residual, often time some number close to 1. I am not sure if this estimated number really represent or can be interpreted as the variance of the random error term. Can someone please help? Thanks!
... View more