Rodolfo,
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> Rodolfo Coelho Prates
> Sent: 22 October 2007 13:29
> To: [email protected]
> Subject: st: Hausman test
>
> Dear Statalisters
>
> I estimeted a panel model, and it�s used, pooled regression,
> fixed efect and randon efect. After this, when the stata run
> the hausman test, the result is negative. The message is that
> the matrix is not positive definite. In this form, I don�t
> know if the fixed efect is more efficient than randon efect.
> How can I do?
Short answer: install -xtoverid- from SSC (from within Stata, -findit xtoverid- and follow the links. It will give you a Hausman test stat after estimation with random effects that is guaranteed to be non-negative and can (unlike the standard Hausman test) be made robust to clustering and the like.
Longer (sort of) answer: this question comes up on Statalist fairly regularly. Try googling on the keywords "statalist negative hausman" and browse through some of the Statalist postings, e.g., one of my contributions:
http://www.datasets.org/statalist/archive/2005-08/msg00762.html
Cheers,
Mark
> Best regards,
>
> Rodolfo
>
> *
> * 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/
>
*
* 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/
