[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Suest and Random Effects.
-suest- relies on the scores at individual observations that are not
defined for the panel models. So you would have to rely on more
traditional means of testing. If yours are non-overlapping samples,
you can simply create dummy indicators of each subgroup, and then use
those dummies and their interactions within a single command. That
way, you would compare the models between samples. Comparing nested
models should also be explicit, at least for identical (sub)samples
On Wed, 20 Oct 2004 10:48:45 -0400 (EDT), Martha Martinez-Martinez
> I want to compare three different random effects models with different
> sample sizes (partitions of a bigger sample). Is there an equivalent of
> suest that I could use with random effects? Any other way of testing
> hypothesis between models?
> Martha Martinez.
* For searches and help try: