model selection for longitudinal data using proc glimmix

[393310]

Hi all, 

Does anyone have any advice as to how I can automatically select the best model in proc glimmix? I am looking at longitudinal data with repeated measures and my outcomes are ordinal. I have 20 explanatory variables. To my understanding, glimmix does not have the option to do stepwise selection so is there another procedure that would work for this type of data and allows stepwise selection? 

Would I be able to use GLMselect to select the predictors for the model then run those in glimmix? Or would the fact that GLM select does not take into account the repeated measures give inaccurate results?

Thanks in advance!

[PaigeMiller]

There is this: https://support.sas.com/resources/papers/proceedings14/1822-2014.pdf

However, I have to disagree with your statement that implies stepwise produces the best model. It does no such thing, it has many many drawbacks that you need to be aware of.