I am fitting a model with fixed and random effects, and making a prediction that involves a nonlinear function of the fixed effects estimates and predictions of the random effects. Is there a way to output the approximate prediction variance matrix (P) described on Pg 2786 of SAS/STAT® 9.22 User’s Guide - The GLIMMIX Procedure?
If so I could apply the delta method in IML, using P from GLIMMIX.
I can do this in NLMIXED, but my data set is large with many random effects and NLMIXED is slow whereas GLIMMIX is really fast (even with method=laplace).
BTW, adding the laplace approximation to GLIMMIX was a great benefit.