Using PROC MI to replace missing values of variables belonging to a scale

[dtcleary]

Hello,

Very new to the SAS world and have been teaching myself over the past month. I am interested in using PROC MI to replace missing values for several variables from a single measure of self-efficacy (7-itme scale). See the code below, but my primary question pertain to PROC REG - what is the nature of one's model if I do not really have IV and DV's. I am simply trying to replace missing values for items on a scale - that is, no prediction of a DV from IV. What am I missing here?

PROC MI Data=EWCA.final seed=40080 out=miout;
mcmc;
var se1 se2 se3 se4 se5 se6 se7;
run;

PROC REG Data=collem Outest=a NOPRINT;
MODEL ??????; ***I am not sure what the model would be as I am not running a regression
BY REPLICATE;

PROC mianalyze data=a;
Modeleffects intercept se1 se2 se3 se4 se5 se6 se7;
Run;

[SAS_Rob]

Typically you would only use multiple imputation when you have a specific analysis in mind.  Once you impute the data, what do you plan to do with it?

[dtcleary]

I have 5 measures that I will be ultimately use as part of a cluster analysis. There are some missing data for each of the measures and I would like to replace missing values for each measure. Thus, suppose measure #1 had 7 items and items 2 and 5 have some missing values. I would like to use MI using all of the items on the measure. I would repeat this for each of the measures and then use the composite scores for each of the measures in a cluster analysis. Make sense?

[SAS_Rob]

Yes it makes sense.  To impute ordinal data like you have you would need to use either the DISCRIM or LOGISTIC method, with the fomer being preferred if you have 7 levels.  You can use either the FCS or MONOTONE statement, depending on the missing data pattern.  

That will take care of the imputation step, but you cannot use multiple imputation and cluster analysis because the theory does not lend itself to calculating combined estimates after running the cluster step due to the absence of point estimates and standard errors.

So I would not use multiple imputation in this case.  You could try some single imputation method or use FASTCLUS rather than CLUSTER which is more forgiving with missing values. 

 

[dtcleary]

Thank you very much for your replay and makes.



The key for me is being able to get the overall composite score for each of the measures and then to use those composite scores in the CA. So you are basically saying that it is best to use FCS in the imputation step without proceeding further down the typical 3 step MI process?

[SAS_Rob]

If you only wanted to impute then, yes, using the FCS DISCRIM or LOGISTIC method should work.