Proc Glimmix vs Proc Mixed

[Babloo]

Appreciate if someone give one simple example to understand the difference between Proc Glimmix and Proc Mixed.

[Rick_SAS · Accepted solution]

The difference between generalized mixed models (GLIMMIX) and linear mixed models (MIXED) are described in the doc: Introduction to Mixed Modeling Procedures.  An in-depth comparison is in the GLIMMIX doc:"Comparing the GLIMMIX and MIXED Procedures."

If you are familiar with generalized linear models (PROC GENMOD) and linear models (PROC GLM), the ideas are similar: the generalized models enable you to model a wider range of response variables, including binary, count data, lognormal data, and more.

[lvm]

Just to add to Rick's answer, GLIMMIX is for any (conditional) distribution in the so-called expoential family. Poisson, binomial, negative binomial, etc. Because the normal distribution is a member of this family, GLIMMIX can also be used for the normal distributions. Thus, many now use GLIMMIX instead of MIXED for normal response data. SAS continues to add new features to GLIMMIX, but new developments with MIXED are pretty much halted.  (But there are some features in MIXED for normal data that have not been implemented in GLIMMIX).