Hmisc::transcan() vs FIML

data analysis
1

Hi

I find that some simulation work shows that Full Information Maximum Likelikhood (FIML) may offer better estimation than multiple imputation. It works even if there are missing values in dependent/response variable. I find not find reading about comparing Hmisc::transcan() with FIML. May one give me some suggestion to know more about them… Thank you very much for your help.

Best wishes

Sandro

2

This needs to be under datamethods topic datamethods.org/rms3 - please put it there as a reply to the last reply there, and I’ll remove it here.

3

Many thanks Professor Harrell. I am a bit foolish. I am not sure how do I move it there.

4

Copy and paste (to the end of that topic as a last reply), then I’ll delete here.