How to statistically compare the imputed datasets and the analyzed dataset in sas?

[Jiaoo]

Hello, there. I am a sas beginner. My quesion is that I want to do the sensitive analysis, to compare the demographic difference between the analyzed dataset and the multiple imputation  dataset. How can I solve it?Thank you very much for your help.

[Rick_SAS]

You don't say how you want to compare the two data sets, but let's assume you want to run univariate descriptive statistics on the numeric variables. Then you would run the following steps and compare the output:

proc means data=Orig_data;
   var _numeric_;
run;

proc means data=Imputed_data;
   var _numeric_;
run;

You could also see how the pairwise correlations compare:

/* by default, PROC CORR computes pairwise correlations */
proc corr data=Orig_data plots=none noprob;
   var _numeric_ ;
run;

proc corr data=Imputed_data plots=none noprob;
   var _numeric_ ;
run;

If you want to compare the data in a different way, please include sample data and specify how you want to compare the variables.

[Jiaoo]

Dear friend,

Really appreciate your help. That is pretty helpful.

If possible,can I ask more?  I want to compare, e.g, if the age or sex variables are statistically different between the original dataset and the imputed dataset. 

Firstly the original data and the imputed data was combined into one dataset. 

data combined;

set original imputed;

run;

Next,  the  statement of below was used, 

Proc freq data= combined;
Table sex* group / CHISQ EXPECTED DEVIATION NOROW NOCOL NOPERCENT;
run; 

Is my method correct?

Thank you very much for your time in advance.