Hello,
I'm looking to undertake MI with clustered data. Where I need to use MI for a level 1 variable.
The structure is students (level 1) within schools (level 2)
Where I have thousands of schools, and highly varying numbers of students within schools.
There is an FAQ on the Stata site here.
But due to the numbers of clusters and students the various approaches on the Stata site don't seem satisfactory.
It would seem that the FAQ was written quite a while ago. There is an emerging literature on undertaking multi-level MI, with programs available for another package with a one letter name.
For example, Audigier et al.

I may be able to cobble together an approach in Stata based upon the literature. But I wonder if this is on the cards for official inclusion in Stata?
For my work, there will be great benefits in using both the fixed and random effects in the MI process.

Regards,

Andrew