Skip to content
Prev 1962 / 20628 Next

Weighted-ML estimates

I wonder if in your situation you might be better off using a completely 
EM approach treating both cluster indicator variables and the random 
effects as the missing data?

Murray Jorgensen
H c wrote: