Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: mlogit with alternative specific variables


From   John Hendrickx <john_hendrickx@yahoo.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: mlogit with alternative specific variables
Date   Tue, 9 Dec 2003 06:34:26 -0800 (PST)

I think what you should try to do is to respecify your -clogit- model
as an -stcox- model. The Cox proportional hazards event history model
is strongly related to the conditional logit model. If there are no
time varying covariates, choice of an alternative is indicated by a
time duration of 1, and non-choice is indicated as censured, the
pseudo-loglikelihood of the Cox model simplifies to the likelihood
function of the conditional logit model. Data structure is the same,
estimates and standard errors ditto. Once you've specified the
-clogit- model as -stcox-, the step to an event history model should
be straightforward

Examples using SPSS and SAS can be found at
http://www.xs4all.nl/~jhckx/mcl/ 
More information can be found in:

Allison, Paul D. and Nicholas Christakis
1994 “Logit models for sets of ranked items.” Pp. 199-228 in Peter V.
Marsden (ed.), Sociological Methodology
1994. Oxford: Basil Blackwell.

Hope this helps,
John Hendrickx

--- Peter Haan <phaan@diw.de> wrote:
> Dear Statalist,
> my question is about the multinomial logit model.
> My model includes both, individual and alternative specific
> variabels. 
> So far I have estimated my model with clogit, which works fine.
> Now I want to translate the clogit model into a mlogit including
> the 
> alternative specific variables. The reason for the model chnage is
> that 
> eventually I want to estimate a discrete time hazard model, which
> can be 
> written as a mlogit.
> As far as I understand Maddala (1983) mlgoit and clogit should be 
> equivalent with both individual and alternative specific variabels.
> However, so far I did not get the equivalent probabilities when I
> am 
> using mlogit, the problem I suppose is the wrong data structure.
> My structure: persnr outcomevariable (1 or2 or3)     indspecific
> vars    
> X(kat1)      X(kat2)      X(kat3)
> Hence, each ind has one row
> 
> Do you have an idea in which form I need to construct my data?
> 
> thank you very much for you help
> peter
> 
> -- 
> 
> Peter Haan
> Wissenschaftlicher Mitarbeiter
> Deutsches Institut für Wirtschaftsforschung (DIW Berlin)
> German Institute for Economic Research
> Abt. Staat / Dept. Public Economics
> Königin-Luise-Straße 5, 14195 Berlin
> Tel. +49/30/89789-165, Fax. +49/30/89789-114 
> 
> 
> *
> *   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/


__________________________________
Do you Yahoo!?
New Yahoo! Photos - easier uploading and sharing.
http://photos.yahoo.com/
*
*   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/



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