Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: Zero-inflated Negative Binomial models for Panel data


From   Rodolphe Desbordes <rodolphe.desbordes@strath.ac.uk>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   RE: st: Zero-inflated Negative Binomial models for Panel data
Date   Thu, 11 Mar 2010 16:46:57 +0000

Stas,

I think that it has been previously mentioned on the Statalist that fixed/random effects zero inflation models can be estimated with LIMDEP. It is my understanding that the individual effects are only present in the count part of the model.

Rodolphe
________________________________________
From: owner-statalist@hsphsun2.harvard.edu [owner-statalist@hsphsun2.harvard.edu] On Behalf Of Stas Kolenikov [skolenik@gmail.com]
Sent: 11 March 2010 15:25
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Zero-inflated Negative Binomial models for Panel data

On Wed, Mar 10, 2010 at 10:48 PM, Jabr, Wael M <
wael.jabr@student.utdallas.edu> wrote:

> Does Stata support Zero-inflated Negative Binomial models for Panel data?
> I have researched some of the documentations but couldn't find a reference
> to that.
>

Does such a model exist at all in the literature?

Zero-inflation is a two part (mixture) model. I don't think it has a
sufficient statistic, so you cannot do any sort of fixed effects. Where
exactly do you want to place the random effects? You can say that you have a
random effect in the mixture equation only (inflation for zeroes); you can
say that you have a random effect in the main equation (binomial
probabilities) but not in the mixing equation; you can have random effects
in both (allowing them to correlate, I guess); or you can have a single
random effect that affects both equations with certain loadings (although
such model may be difficult to identify). Those would be four different
kinds of models that will have different fit to data.

-gllamm- does not support the negative binomial family. However since the
negative binomial model is the Poission model with a gamma random effect
(integrated over), you may be able to get some leverage out of Poisson
family.

--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
*   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/

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index