I need to do Combined analyses of GLM mixed, where we calculate the effect of Location, Treatment, LocationxTreatment.   Thanks. I see you have switched to PROC GLIMMIX, you might want to update your original post to reflect this.   proc glimmix data=IMPORT; class TRT REP LOC; model LHC = TRT REP LOC LOC*TRT / solution ddfm=kr; random LOC; run;   Your words (quoted above) do not indicate that REP is in the model, and perhaps it should not be there anyway. But that depends on what REP really means and how you want it to be used in the model. If it indicates a replicate observation under identical conditions, I would not put REP in the model. But that's a decision you have to make, I can't make that decision, as you have not explained what REP is or how it should be used. ... View more