How to test overall within subject effect in Proc Genmod

[timothy19]

Hello Community,

I have this code to test within the subject effect for my data set.
I found out that the QIC of the type =ind is the best fit for the model but here is my next question; how do i test for the overall within-subject effect? Thanks

proc genmod data=d003   order=data;
class y_variable  sub_id  Position;
model y_variable =Time_in_Seconds Resistance__ohms Force_N   Position  / d=bin link=logit;
 repeated subject= sub_id / type=ind; *better model with better QIC;
run;
quit;

[StatDave · Accepted solution]

You would have to put your SUBJECT= variable in the MODEL statement, but this is likely to cause problems in the GEE fitting algorithm since that will add the need to estimate a parameter for each subject. Keep in mind that one of the main purposes and advantages of the GEE method is to avoid the need to estimate the possibly large number of subject parameters so that you can obtain good estimates and tests of the other effects in the model. The subject parameters are typically regarded as nuisance parameters with primary interest in the other parameters.