Proc Genmod

[darb]

Hi all,

I am trying to understand why in the example for GEE (page 3) the Type 1 option is being used rather than the Type 3 option. 

https://support.sas.com/documentation/cdl/en/statug/63962/HTML/default/viewer.htm#statug_genmod_sect...

Also it seems to contradict with what is described here: https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_genmod_sect...

Which says "This type of analysis is not available for GEE models, since the deviance is not computed for this type of model."

Thanks,

Doug

[StatDave]

The label on your first link seems to be to GENMOD documentation, but actually goes to a paper, so it's not clear which you really are referring to. However, there would be no reason to use the TYPE1 option when the REPEATED statement is specified because those tests only apply to a model fit using maximum likelihood estimation (which allows computation of the deviance) and the GEE model uses a different algorithm. Note that a maximum likelihood fit is done before using the GEE algorithm purely for the purpose of providing starting values of the model parameters for the GEE algorithm. That is why you ideally see two "algorithm converged" messages in the log when fitting a GEE model.