This was to do with missing values: due to the pattern of missing
observations, omitting one of your variables from the model
resulted in an increase in the sample. There is code in place
that is intended to avoid this effect, but it was not correctly
set up for the case of panel models. This should be fixed now.
Hello Allin...
It works perfectly in the case of panel data models with fixed effects but
crashes in case of random effects models...
Thanks for your help...
Mariusz
Poland