MIXED and GLIMMIX

[plf515]

Are there any analyses that can be done in PROC MIXED but not in PROC GLIMMIX?  If so, what are they?

[ballardw · Accepted solution]

From the documentation for Proc mixed in the OVERVIEW section:

The GLIMMIX procedure fits generalized linear mixed models (GLMMs). Linear mixed models—where the data are normally distributed, given the random effects—are in the class of GLMMs. The MIXED procedure can estimate covariance parameters with ANOVA methods that are not available in the GLIMMIX procedure (see METHOD= TYPE1, METHOD= TYPE2, and METHOD= TYPE3 in the PROC MIXED statement). Also, PROC MIXED can perform a sampling-based Bayesian analysis through the PRIOR statement, and the procedure supports certain Kronecker-type covariance structures. These features are not available in the GLIMMIX procedure. The GLIMMIX procedure, on the other hand, accommodates nonnormal data and offers a broader array of post-processing features than the MIXED procedure.

[plf515]

Thanks. This is helpful.  I am not going to mark it as a "solution" because I'd like to wait and see what other people say, but, if nothing happens, I will mark it.