MI ANALYZE by group(regions)

[Gilles-Protais]

Greetings community,

am trying to do multiple imputation and I which to have pooled estimates through the function MIANALYZE by region, but I don't how to include that in my code

proc mi nimpute=2 data=Work.query seed=45 out=want_imp1;
class age7 sgp region province ghq12 sex;
fcs regression (lnbmi) logistic (sgp) discrim (province) ;
var id hh wfin  age7 lnbmi    ghq12  province region sgp sex;
run ;
 

proc mixed data= want_imp1;
title 'PROC MIXED weighted + clustered GP after MI';
by _imputation_ regionch;
weight wfin;
class hh age7 sex sgp ;
model lnbmi = GHQ12 sgp sex*age7 / solution covb;
random intercept / subject=hh;
ods output SolutionF = mixed_GP_parms covb = mixed_GP_cov;
run;


proc mianalyze parms=mixed_gp_parms ;
title 'MIANALYZE of MIXED with GP as covariate';
class sgp sex age7 ;
modeleffects intercept sgp  age7*sex;
by region
run;

Thanks very much.

[SAS_Rob]

That code should work, assuming you have a typo in the MIXED code on the BY statement with the variable regionch instead of region.  Can you post the LOG so we can see exactly what is going wrong?

[Gilles-Protais]

[SAS_Rob · Accepted solution]

It looks like the data is not properly sorted.  Try sorting it first and see if that takes care of the issue.

proc sort data=mixed_gp_parms;
by region _imputation_;
run;

proc mianalyze parms=mixed_gp_parms ; 
title 'MIANALYZE of MIXED with GP as covariate'; 
class sgp sex age7 ; 
modeleffects intercept sgp age7*sex;
by region;
run;

[Gilles-Protais]

Thanks very much. It worked for me.