|
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: RE: Comparing models - Akaike, Wald-Test?
Thank you all for your helpful answers.
I just tried "estat ic". It works and gives me, what I want. Btw. what kind of test is BIC? Does it work like the AIC?
I am using a generalized Poisson-model to capture underdispersion. It was downloadable from the site of the book of Dr. Hilbe.
There is an indicator for underdispersion which is around 0,7. But I donŽt know how to interprete it.
If anyone has a hint, I am grateful.
Thx again,
Markus.
> -----Ursprüngliche Nachricht-----
> Von: [email protected]
> Gesendet: 14.04.08 21:42:16
> An: <[email protected]>
> Betreff: st: RE: Comparing models - Akaike, Wald-Test?
>
> Markus Köller wrote:
>
> >>I want to use the Akaike criterion to compare both approaches and wondered whether there is any function in stata to compute that criterion or whether it is easy to derivate from the results.<<
>
> estat ic
>
>
> >>The other point is, that I am wondering, whether the Wald-Test is useful to compare both models? The standard poisson model has a Wald chi2(22) of 1410,56 and the generalized Poisson model has a Wald chi2(22) of 1333,46. Does that mean, the standard model fits better or is the Wald-Test not appropriate for that comparison?<<
>
> What generalized Poisson model are you running? The dfs seem fishy to me. If you are running a negative binomial model, Stata performs a LR test whether the additional parameter for overdispersion is needed. This test appears at the bottom of the nbreg output, along with an important footnote about the test.
>
> IMO, you should take a look at the book by Long and Freese (http://stata.com/bookstore/regmodcdvs.html), which discusses this in detail.
>
> Jay
>
> *
> * 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/
>
_______________________________________________________________
Schon gehört? Der neue WEB.DE MultiMessenger kann`s mit allen:
http://www.produkte.web.de/messenger/?did=3016
*
* 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/