PROC GENMOD

[ims]

Hello! I am emailing you with a brief coding question. I need to fit a marginal logistic regression model and used the following code:

PROC GENMOD DESCENDING;

   CLASS id trt (REF=first) week (REF=first);

   MODEL y=trt week trt*week / DIST=BINOMIAL LINK=LOGIT;

   REPEATED SUBJECT=id / WITHINSUBJECT=week TYPE=EXCH;

RUN;

unfortunately, I only get estimates from my model but cannot get 95%CIs and SE. Do you have any idea what could be wrong with the code?

I get the following error message:

error in computing the variance function

error in parameter estimate covariance computation

error in estimation routine

[StatDave]

With binary response models, this typically is the result of the data being sparse. Sparseness usually occurs when all responses in one subpopulation of your data (one trt-week combination) are the same. The way to fix this is usually to merge sparse subpopulations with others to remove any sparse subpopulations. For ordinary binary response models, this condition is called separation and is discussed further in this note.

[ims]

Perfect, thanks so much for getting back to me so quickly!!! I think I figured it out...