topic Re: Deviance in output of Poisson regression model with repeated subject statement in Statistical Procedures

Deviance in output of Poisson regression model with repeated subject statement

mbcox — Wed, 23 Mar 2016 21:44:42 GMT

I am running a poisson regression model for the number of admissions for a group of patients (identified by "ID"). The explanatory variables are post (pre vs post period) and case (case vs control) with an interaction term post*case. As each patient can have more than one admission, I have added a repeated subject statement to the model. However, after adding the repeated subject statement to the code the output for the model no longer gives the deviance. How can I specify that the output for the model include the deviance?

proc genmod data=alldata;

class ID;

model admit = post case post*case / dist=poisson link=log offset=logtime;

repeated subject=ID;

run;

Any suggestions would be appreciated!

Re: Deviance in output of Poisson regression model with repeated subject statement

lvm — Thu, 24 Mar 2016 20:41:04 GMT

When you add a repated statement, you are perform GEE, which involves a quasi-likeliood (not a true likelihood or true distribution). Thus, the deviance (which is based on a true likelihood) is not printed.

Re: Deviance in output of Poisson regression model with repeated subject statement

mbcox — Tue, 29 Mar 2016 16:23:12 GMT

Thank you for your reply. Do you know how I can run this type of model with the repeated subject statement and force SAS to output the deviance? Or do you have a suggestion of another model that would be better for this situation? Thank you.