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

Re: st: Missing chi squared in negative binomial

From   Maarten buis <[email protected]>
To   [email protected]
Subject   Re: st: Missing chi squared in negative binomial
Date   Mon, 4 Jan 2010 10:44:56 +0000 (GMT)

--- On Mon, 4/1/10, joe j wrote:
> I ran a couple of negative binomial estimations - three
> specifications for each of my five dependent variables - 
> and in just some models I find chi2 missing. 

Your model specification looks too complicated (= too many
covariates) for your data. The negative binomial is a pretty 
complicated model as it involves estimating a variance of a 
variable you haven't observed (the alpha). So each 
observation in the data contains relatively little 
information that can be used in the estimation. This means 
you require a lot of observations to get a reliable 
estimate. 603 observations for estimating 45 parameters 
means about 13 observations per parameter, which to me
just looks too small for a model like this. So I am not
surprised that Stata is complaining.

Hope this helps,

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


*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index