Date:         Wed, 13 Sep 2000 16:24:40 +0200
Reply-To:     Dr Olaf Kruse <olaf.kruse@VST-GMBH.DE>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Dr Olaf Kruse <olaf.kruse@VST-GMBH.DE>
Subject:      Re: Poisson Vs Negative Binomial Regressions
Content-Type: text/plain;charset="iso-8859-1"

Ogbonnaya John Nwoha wrote:
> How can I determine the better of a poisson regression result and a
> negative binomial regression result.  Attached are the outputs of an
> actual regression.  Which one is better?


I didn't study the attachments in detail. But the foremost reason to move
from a Poisson- to a Neg-Binomial-model are problems with overdispersion,
that usually lead to an underestimation of the variance-terms and
confidence-intervals. (It makes your model look more "significant"). As you
see from your results, the std.-errors are always higher in the
Neg-Binomial-model.

I suppose that you'll receive similar results with respect to the
parameter-values and predicted values, if you eliminate all insignificant
terms from your Neg-Binomial-model and run the same model as a
Poisson.model.


Cheers,

Olaf

+-----------------------------------------------------+
 Dr. Olaf Kruse
 VST -Gesellschaft fuer Versicherungsstatistik
 Roscherstr. 10
 30161 Hannover/FRGermany

 mail: Olaf.kruse@vst-gmbh.de
 phone:++49-511-339 599 21
 fax:  ++49-511-388 57 13
+-----------------------------------------------------+