> I am examining binary dependent variable panel data models using xtlogit
> or xtprobit.
>
> If I were using xtreg it would be easy to do a hausman test for fixed
> effects or random effects.
>
> Does anyone have any suggestions ariound assessing this for binary
> dependent variables, or any other post estimation tests for binary dep
> var csts models
No, I don't think this can be done explicitly and quickly. -xtprobit- is
good for random effects as it implies integration of the bivariate normals
which is something people have learnt how to do, over time. -xtlogit- is
good for fixed effects as there is some nice algebra that makes it boil
down to eliminate the fixed effects by conditioning, and essentially down
to -clogit-. (That's still a conditional rather than full likelihood
procedure.) Theoretically, you can do the random effects in panel logit
with the corresponding logistic distribution integrals, but I doubt
anybody ever wanted to do that, and likelihoods would not be comparable,
anyway, as long as the fixed effect estimator is a conditional one.
No such problems are relevant for the linear panel data model (-xtreg-)
as, analytically, it is linear, so you can again eliminate the fixed
effects by simply subtracting some things from other things, and, on the
other hand, the random effect integrals can be done analytically and lead
to the GLS estimation.
--- Stas Kolenikov
-- Ph.D. student in Statistics at UNC-Chapel Hill
- http://www.komkon.org/~tacik/ -- [email protected]
* This e-mail and all attachments to it are not intended to provide any
* reasonable point of view and was transmitted to you in error. It
* should be immediately deleted by all recepients unless they really
* enjoy communicating with the author :). Other restrictions apply.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
