06-19-2017 08:00 AM
I was just wondering what the best imputation method for categorical variables is. Considering many categorical variables (with exception of variables such as postcodes etc) dont have too many levels, will it not be misleading to impute their missing values?
I have always thought creating a 'one-size-fits-all' variable such as CCC and imputing that for every missing level might be more reasonable. Can someone help out with any ideas or suggestions?
06-19-2017 10:11 AM
That's why you do multiple imputation to reduce chance of bias AND to understand the effect of the imputation.
This isn't a SAS question though, it's a statistics question better answered by statisticians or Google.