How can I incorporate the MNAR statement into FCS MI for binary data

mcasas — Tue, 19 Feb 2019 23:22:42 GMT

Hello, I am trying to perform multiple imputation for binary data that is MNAR in SAS 9.4. The data comes from an adaptive test with large missing rates where basically 90% of the variables have missing data due to the adaptive algorithm. I need to impute the whole data set so that I can perform a confirmatory factor analysis. First, I used the FCS discrim within the proc mi statement including all my variables as class variables. My problem is that I cannot incorporate the MNAR statement and do not know how to specify the model for MNAR since I need to impute all the variables in the data set. I do not even get the MNAR statement activated when I typed it into my code.

I have read that the MNAR statement can be used along with FCS but I cannot find any example. Any suggestions would be greatly appreciated.

Below is the code that I have for the FCS part:

%macro myimp(mydata=repcat1, myoutimp=imp1);
proc mi data=&mydata nimpute=5 out=&myoutimp;
class q1-q50;
fcs plots=trace (mean std);
var q1-q50;
fcs discrim (q1-q50/classeffects=include);
run;
%mend;

Thank you!

Re: How can I incorporate the MNAR statement into FCS MI for binary data

SAS_Rob — Wed, 27 Feb 2019 17:05:42 GMT

What release of SAS are you running?

What do you mean when you say the the MNAR statement is not "activated"? Do you get an ERROR message in your LOG that suggests something went wrong with it?

Provided that you are running SAS/STAT 13.1 or SAS 9.4TS1M1 then you should be able to use the MNAR statement and the FCS statement together in a way similar to this example.

http://support.sas.com/documentation/cdl/en/statug/66859/HTML/default/viewer.htm#statug_mi_examples16.htm

topic Re: How can I incorporate the MNAR statement into FCS MI for binary data in Statistical Procedures

How can I incorporate the MNAR statement into FCS MI for binary data

Re: How can I incorporate the MNAR statement into FCS MI for binary data