How do I perform correlation analysis with multiple imputed dataset?

[Gregorytus07]

Hi everyone,

I used the PROC MI statement to perform multiple imputation for a dataset with missing data. The second step is to use PROC CORR to conduct the correlation analysis and generate an output dataset to be used with PROC MIANALYZE. I was able to generate the correlation analysis by the imputation number but couldn't get a pooled result using the PROC MIANALYZE. This is the code I used below:

PROC CORR DATA=PREP_MI_TTT COV OUTP=PREP_H1;
VAR STIGMA_P NORMS_T;
BY _IMPUTATION_;
RUN;

PROC MIANALYZE DATA=PREP_H1 EDF=30;
MODELEFFECTS STIGMA_P NORMS_T;
RUN;

[sbxkoenk · Accepted solution]

Usage Note 38739: Combining multiply imputed correlation matrices using PROC MIANALYZE
https://support.sas.com/kb/38/739.html

Koen

[Gregorytus07]

Thanks, Koen! This is really helpful.

Do you know how I can assess the earlier example referenced in the first statement  "The example titled "Combining Correlation Coefficients" in the PROC MIANALYZE documentation illustrates how to combine sample coefficients for the correlation between two variables computed from a set of imputed data sets by using Fisher's z transformation." 

I think the referenced exampled would be an additional guide considering that I am trying to compute the correlation coefficient between two variables.

[Gregorytus07]

Thanks, Koen! I found it.
https://documentation.sas.com/doc/en/statcdc/14.2/statug/statug_mianalyze_examples11.htm