topic Re: Proc hpgenselect with matched data in Statistical Procedures

Proc hpgenselect with matched data

Liamb — Mon, 15 Jul 2024 13:32:35 GMT

Hello,

How to apply stepwise selection with proc hpgenselect to matched data?

Thanks

proc hpgenselect data=tab;
class paired var1 var2 var3 var4 ;
model wbc= var1 var2 var3 var4/ dist = negbin link=log ; 
/*random intercept / subject=paired;*/
selection method=stepwise details=all;

run;

Re: Proc hpgenselect with matched data

sbxkoenk — Tue, 16 Jul 2024 11:59:56 GMT

Matched data? You mean data where treated units are matched to control units? Or you mean pre- and post- (treatment) measurements?

I am moving this topic to "Statistical Procedures" - board where it belongs.

Koen

Re: Proc hpgenselect with matched data

Liamb — Tue, 16 Jul 2024 12:01:26 GMT

I mean data where treated units are matched to control units

Re: Proc hpgenselect with matched data

SteveDenham — Tue, 23 Jul 2024 13:59:44 GMT

If you are locked into model selection techniques like stepwise, and you realize that it really isn't suited to mixed models, consider treating the difference between the treated and control subjects for each pair as your response variable. Without knowing your data structure, consider this:

proc hpgenselect data=tab;
class paired var1 var2 var3 var4 ;
model wbc_diff= var1 var2 var3 var4/ dist = negbin link=log ; 
selection method=stepwise details=all;
run;

The key here is in calculating the response variable wbc_diff. There are several ways, but all of them depend on the current structure of the dataset "tab".

You might be happier using method=lasso/choose=aicc, as it provides at lest some protection against type I error inflation due to multiple testing.

SteveDenham