Would it be valid to impute the variable for dataset B, and then to only conduct analysis on dataset B?
For example:
Code:
webuse lbw, clear set seed 12345 generate wave = round(uniform(),1) replace bwt = . if wave==1 //artificially creating missing data in half of dataset mi set mlong mi register imputed bwt mi impute chained (pmm, knn(5)) bwt = age lwt i.race, add(5) rseed(12345) mi estimate : glm age lwt i.race bwt if wave==1 //command runs, but is only conducted on part of sample where all data was missing originally
0 Response to Is it okay to use multiple imputation to create imputed values in a second dataset?
Post a Comment