topic Re: How to do the impute phase for multiple imputation in Statistical Procedures

How to do the impute phase for multiple imputation

Therain — Thu, 23 Feb 2017 19:59:52 GMT

Hello,

I use SAS 9.4 and I have a data set which has 7 variables: ID, a, b, c, d, e, f. There are some missing values for variables a-f. a, b and c are categorical variables and d, e, f are continuous variables. I tried the following way for the imputation phase but the missing values are still missing and not be imputed. Does anyone know the reason? Thank you!

proc mi data =have nimpute=100 out=want;
class a b c;
fcs discrim(a = b c d e f/classeffects=include);
fcs discrim(b = a c d e f/classeffects=include);
fcs discrim(c = a b d e f/classeffects=include);
var a b c d e f;
run;

Re: How to do the impute phase for multiple imputation

Haikuo — Thu, 23 Feb 2017 21:27:35 GMT

Posting your question in SAS/STAT sub-forum would make it more visible to the experts in this area. https://communities.sas.com/t5/SAS-Statistical-Procedures/bd-p/statistical_procedures

Re: How to do the impute phase for multiple imputation

Therain — Fri, 24 Feb 2017 20:49:43 GMT

Thanks! I will move my question!