CM,
For NB2 the variance is quadratic in the mean. In Stata this is -nbreg,
dispersion(mean)-
For NB1 the variance is a multiple of the mean. In Stata this is -nbreg,
dispersion(constant)-
See the Stata FAQ "How do you specify the variance function in nbreg to coincide
with Cameron and Trivedi's (Regression analysis of count data, page 62) NB1 and
NB2 variance functions?" at http://www.stata.com/support/faqs/stat/nbreg1.html
Hope this helps,
Scott
----- Original Message -----
From: "Chih-Mao Hsieh" <[email protected]>
To: <[email protected]>
Sent: Tuesday, December 09, 2003 3:58 PM
Subject: st: Testing hypotheses on variance of dependent variable: NBREG
> Hi all,
>
> This question is part Stata, part statistics in general. I have a dep.
variable made up of count data. High overdispersion. I want to be able to test
hypotheses on both the expected value of the dep. variable, as well as the
variance of the dep. variable, regressing on a bunch of independent variables
that I have.
>
> After examining nbreg's options, it appears testing my expected value
hypotheses uses the basic command:
>
> . nbreg depvar [indepvars]
>
> Is there a way to formulate a variance or dispersion parameter for each
observation, which I can regress on my ind variables? I have heard something
about a Negbin II (??), but not sure how to proceed in Stata.
>
> CM
>
>
> *
> * 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/
