We are kinda stuck with a problem.. I guess it's both SAS and a statistics related problem. Since repeated measurements are conducted on our subjects we cannot assume independence, so we are looking into Generalized Estimating Equations.
In this case we want to answer the question: Is there a difference between measurements depending on which education level the subject has?
The data we are working with is as follows:
NR_OF_SUICID_ATTEMPTS (categorical, ordinal with 3 levels)
EDU (categorical, ordinal with 4 levels)
ID (unique number for each subject, and has the value 1,2,...,n)
COUNTER (measurement for ID(i), and has the value 1 or 2)
And the code in SAS:
proc genmod data=bipolar.Uppfoljning_2uppf_fardig desc;
class ID COUNTER EDU(REF = '0') / PARAM = ref;
model NR_OF_SUICID_ATTEMPTS = EDU counter / link=cumcll dist=multinomial;
repeated subject=ID / type=indep covb corrw;
So, our main question is for you: Is there a way of testing for difference between repeated measurements in SAS proc genmod? We have looked around on google, but we are not really sure what to look after, or what the name is of our problem.