> thanks a lot for the hint with gllamm. Is there also a way to perform
> fixed effects for ologit in xt.
Economists tend to love fixed effects because in the context of linear
regression, FE allows to get away with endogeneity assumptions. You can
also do conditioning in the fixed effect logit estimation and recieve
something easily estimable. However, in all other models there are no
sufficient statistics for those ancillary parameters, so there is no
closed form solution that conditions on the individual effects.
Introducing the dummy variables for panels is actually only working for
the linear model, too -- exactly because of the linearity of the response,
where all you need to do is to model the mean tendency.
Thus, one must be happy with the random effects because that is the only
way to estimate hierarchical models. Biased? May be. But there is nothing
else there. Besides, the matter of whether you believe the estimating
procedure to be biased because of the regressor endogeneity (i.e.,
correlation between the error term and regressors) hinges on whether you
view the ordered logit procedure as the one where there really is a latent
propensity of a response that was categorized into 1, 2, 3, etc., or you
just view the ordered logit model as a useful device to directly model the
probabilities of the responses given the covariates.
Sophia and Anders, the authors of -gllamm-, were briefly mentioning those
issues in their tutorial at NASUG yesterday.
--- 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 recipients 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/
