Subject: | |
From: | |
Reply To: | |
Date: | Wed, 27 Sep 2006 18:13:22 -0500 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Erika,
I'm dealing with the same issue. Winship and Mare discuss to some
degree extensions of the Heckman approach:
Christopher Winship and Robert Mare. 1992. "Models for Sample Selection
Bias." Annual
Review of Sociology 18: 327-50.
Greene's "Econometric Analysis" contains a
couple of paragraphs addressing the issue. I believe Greene also has some
unpublished paper on this, but I have not been able to track it down.
As far as estimating a two-stage ordered probit, my understanding is
that NLOGIT
---the fancier and more expensive version of LIMDEP--- can do it.
Regards,
renato
>
> Maarten Buis wrote:
>
>>
>> --- Erika Franklin Fowler wrote:
>> > Has anyone worked with selection models (a la Heckman) where the
>> dependent
>> > variable of interest requires an ordered probit? I'm familiar with
>> the
>> > heckman and heckprob commands in Stata, but the only models I've
>> seen with
>> > an ordered model have bootstrapped the standard errors. In
>> particular, I'm
>> > hoping someone out there has or has seen a pre-programmed routine.
>>
>> Within Stata there is a user written program called -ssm- which
>> is a wrapper for -gllamm-. Among others, it does a selection
>> model for an ordered probit and maybe an ordered logit. It is
>> described in detail in the last issue of the Stata Journal
>> (volume 6 issue 3).
>>
>> > Also, there was talk on this list several years ago about models with
>> > selection based on multiple processes, where two conditions (say c1
>> and c2)
>> > are observed separately (not quite Poirier), but the dependent
>> variable is
>> > only observed if c1=c2=1. Has anyone published on this issue?
>>
>> I remember seeing those types of models in G. S. Maddala (1983)
>> "Limited-Dependent and Qualitative Variables in Econometrics"
>> Cambridge University Press.
>>
>> HTH,
>> Maarten
>>
>> -----------------------------------------
>> Maarten L. Buis
>> Department of Social Research Methodology
>> Vrije Universiteit Amsterdam
>> Boelelaan 1081
>> 1081 HV Amsterdam
>> The Netherlands
>>
>> visiting adress:
>> Buitenveldertselaan 3 (Metropolitan), room Z434
>>
>> +31 20 5986715
>>
>> http://home.fsw.vu.nl/m.buis/
>> -----------------------------------------
>>
>> ********************************************************** Political
>> Methodology E-Mail List Editor: Karen Long Jusko
>> ********************************************************** Send
>> messages to [log in to unmask] To join the list, cancel your
>> subscription, or modify your subscription settings visit:
>> http://polmeth.wustl.edu/polmeth.php
>> **********************************************************
>
>
>
> --
> _______________________________
> Indridi H. Indridason
> Associate Professor/Associate Chair
> Department of Political Science
> Oddi v/Sturlugotu
> University of Iceland
> 101 Reykjavik
> Iceland
>
> e-mail: [log in to unmask]
> tel: +354 525 5209 [Iceland]
> tel: (585) 233 4604 [U.S.]
> homepage: www.hi.is\~ihi
>
> **********************************************************
> Political Methodology E-Mail List
> Editor: Karen Long Jusko <[log in to unmask]>
> **********************************************************
> Send messages to [log in to unmask]
> To join the list, cancel your subscription, or modify
> your subscription settings visit:
>
> http://polmeth.wustl.edu/polmeth.php
>
> **********************************************************
--
*****************************************************************
Renato Corbetta
Assistant Professor
238A Ullman
Department of Government
The University of Alabama at Birmingham
1212 University Blvd.
Birmingham, AL 35294-3050
Phone: 205-934-2336
Fax: 205-975-5712
Web site: http://www.dpo.uab.edu/~corbetta/home.htm
*****************************************************************
**********************************************************
Political Methodology E-Mail List
Editor: Karen Long Jusko <[log in to unmask]>
**********************************************************
Send messages to [log in to unmask]
To join the list, cancel your subscription, or modify
your subscription settings visit:
http://polmeth.wustl.edu/polmeth.php
**********************************************************
|
|
|