06-22-2012 08:07 AM
Could you give a reasonably full description of your experimental design? With that info, there are several in the community who will be able to help. However, with what you have given so far, it is hard to help--in fact, I can't even say that PROC GLM is the best tool for what you wish to accomplish.
Hope this will get you to post a more complete question.
06-24-2012 10:10 PM
Since you talk about binomial variables , Or maybe you can consider using proc reg.
"in analysis with binomials variables."
About your question, you need to check residual item of your model , to see whether your model has https://communities.sas.com/dict://key.0895DFE8DB67F9409DB285590D870EDD/heteroscedasticityheteroscedasticity. i.e. your residual item like a curve or increasing or decreasing .