topic Prediction with PROC GLIMMIX in Statistical Procedures

Prediction with PROC GLIMMIX

longitudinal — Thu, 26 Apr 2018 19:36:06 GMT

To Community members,

If the predictions the same for the following 2 scenarios with PROC GLIMMIX:

1. fit the model with random intercept and predict with option pred(noblup ilink), suppose to be marginal mean as well?

2. fit the model without random intercept and predict the marginal mean

If not the same, why?

Thank you all!

Re: Prediction with PROC GLIMMIX

PaigeMiller — Thu, 26 Apr 2018 19:54:30 GMT

If you remove a term from the model, I would not expect to get the same prediction.

Re: Prediction with PROC GLIMMIX

longitudinal — Thu, 26 Apr 2018 20:11:02 GMT

Thanks. My confusion is how to interpret these two predictions. Seems to me that they are both marginal predicted probabilities without including the random effect. SAS has the option for prediction without random effect (noblup), even one specified random effect.