02-02-2016 10:20 AM
Does anyone know if there is a way to get odds ratios (and 95% confidence intervals) for a multiple logistic regression model where data was imputed with proc mi (fcs logistic regression method) and combined with proc mianalyze?
This is the code I used:
PROC MIANALYZE PARMS(CLASSVAR=CLASSVAL)=temp.parmsout covb(effectvar=stacking)=temp.covbout;
CLASS homeowner carowner Ethnicity;
MODELEFFECTS age_reg homeowner carowner Ethnicity;
This gives me the pooled parameter estimates from my 10 imputations and the variance information, but wondering if I can get pooled odds ratios? Or can I transform the estimates into odds ratios?
Any help would be much appreciated.
02-05-2016 12:33 PM
Alright, I just posted my own question - so I will pay it forward by tryig to tackle this one.
proc surveylogistic data=mydata varmethod=jk; by _imputation_; class X1 (ref='0') /param=ref;; model Y (event='1') = X1; ods output parameterestimates=ORs; run; proc mianalyze parms(classvar=classval)=ORs; class X1; modeleffects intercept X1; run;
I hope this helps, I just grabbed this old piece of code from when I ran a comparable question.