Re: st: xtnbreg pre and post estimation

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Fri, 28/5/10, Jabr, Wael M wrote:
> I am using xtnbreg fixed effects but I can't seem to find a
> way to:
> 1) compare the performance of xtnbreg over xtpoisson
> (basically test for overdispersion)

This is a fairly complicated test, as it's null hypothesis is
on the edge of the parameter space (a variance is equal to 0).
See: -help j_chibar-. In other words, don't expect an easy 
answer to that question.

> 2) use Hausman to compare fixed effects versus random
> effects.

If you want to know the influence of contextual variables 
(variables on the higher/group level), than you have no choice
but to go for a random effects model. The ansumptions are not
that different from the assumptions you make when useing a 
regular regression. So if you aren't worried when using 
linear regression, you also don't need to be worried when
using random effects regression. However, fixed effects 
regression is a very convincing design if you are only 
interested in variables that change in lowest/individual level.
So if that is the case, go for fixed effects regression.
In other words, the Hausman test is a bit superfluous.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/