[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: RE: Countvariables/Poisson Regression in heterogeneouschoice models

From   jverkuilen <>
To   <>
Subject   RE: st: RE: Countvariables/Poisson Regression in heterogeneouschoice models
Date   Tue, 24 Jun 2008 21:09:56 -0400

Unbiased? Forget about that, doesn't exist in ML for counts (in general).

It sounds to me like you have a meta-analysis problem. It is not simple because  you are going to have to deal with the composite link issue, because you have different response scales. 

Rabe-Hesketh and Skrondal have an article in Psychometrika on this and should have sample code and the full cite on 


-----Original Message-----
From: "Sebastian Sattler" <>
Sent: 6/24/2008 5:32 PM
Subject: Re: st: RE: Countvariables/Poisson Regression in heterogeneous choice models

Hello Jay,
thanks for your fast answer.

I still read something about poisson/negative binomial regression analysis and the "poisson"/"nbreg"-commands in the Long book.

My problem is an adequate group comparision of coefficients (in different models; or with a pooled dataset using interactions of the group variable and all other independent variables).

I have three different studies and I want to compare the results between them.

I do not know any procedure which gives me unbiased estimates.


-------- Original-Nachricht --------
> Datum: Tue, 24 Jun 2008 17:16:10 -0400
> Von: "Verkuilen, Jay" <>
> An:
> Betreff: st: RE: Countvariables/Poisson Regression in hetergenous choice models

> Sebastian Sattler wrote:
> >Hi everybody, I am searching a procedure which helps me to estimate the
> networksize of respondents (count variable).>
> You should probably read up on count regression and its various
> relatives, e.g. truncated count regression (for your limited range
> responses), as well as exposure/offset terms, which can be used to
> control for rates. The Stata Press book by J. S. Long and J. Freese
> (2005), Regression Models for Categorical and Limited Dependent
> Variables with Stata, 2nd Edition, is a good place to look. 
> Ordinal regression models aren't really appropriate for counts. 
> JV
> *
> *   For searches and help try:
> *
> *
> *


Sebastian Sattler
Voltmannstraße 156

33613 Bielefeld

Mobil:        0174/ 9784307


Ist Ihr Browser Vista-kompatibel? Jetzt die neuesten 
Browser-Versionen downloaden:
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index