PROC PLM: how to get pulled odds ratios on multiply imputed data

[ejay0503]

Hello, all!

Do anyone know how to get pulled odds ratios for the PROC PLM procedure on "multiply imputed data"? 

1) I just ran PROC MI to do multiple imputation; 

2) ran PROC SURVEYLOGISTIC to run the model; 

3) ran PROC MIANALYZE to get pulled results (including odds ratios). 

4) Since I have 2 interaction terms in the model, I ran PROC PLM to conduct simple slope analyses; 

5) ran PROC MIANALYZE to get pulled results for Step 4. 

After Step 5, I only got pulled estimates but NOT pulled odds ratios. How do I get pulled odds ratios for Step 5? I would really really appreciate if you could share your insight with me! 

[Reeza]

exp(Parameter_Estimates) => Odds Ratio

[ejay0503]

That's for the reply, Reeza!

I did include "exp" in my PROC PLM statement (shown below), but my question is, since my data is multiply imputed, how to get "pulled odds ratios" for the plm procedure. What should I include in the PROC MIANALYZE statement to get pulled odds ratios from the PROC PLM step?

ods graphics on;
proc plm restore = perm.plm;
estimate 'slope x, y = 60'   x 1   x*Y 60 /e   exp ;

ods output Estimates = est_ds;
run;
ods graphics off;

proc sort data = est_ds; by DV_r label _imputation_; run ;

proc mianalyze data = est_ds; by DV_r label ; modeleffects estimate; stderr stderr; run;

Thanks! 

[Reeza]

PLM doesn't fit a new model, it scores it or tests hypothesis under the basis of the model. The estimates are the same as from the original model. But this may be more complex than I understand and you should wait for someone else to answer your question. 

[ejay0503]

Thank you again for the quick reply! Yes, I will wait to hear from others' thoughts too.