Hello Michele,
A couple of thoughts - I am not sure your full model actually actually
solved. A condition number that large is usually an indication that
something went wrong. You might try more nip points or initially dropping
the level 2 random intercept as it seems to be quite small in your null
model.
Are you running -gllapred- immediately after running -gllamm-? I use
-estsave- to save results from -gllamm- as it saves the -gllamm- results
right in the data and can be retrieved at any time to be used with
-gllapred-.
You might also have a look at why so many of your samples are disappearing
between your null and full model.
Good luck!
Cam Gillies
