st: Re: calculating inverse mills ratios from a multiple logit.

["R.E. De Hoyos" <redeho@hotmail.com>]

Srini,

Since the residuals in the logistic regression are i.i.d. with extreme value 
distribution, to get the inverse of the mills ratio from the conditional 
probability of a multinomial logit you will have to make a normal 
transformation of it. You can read all about this procedure in Lee (1983) 
Econometrica.

There is a Stata command that handles this:

http://www.delta.ens.fr/gurgand/selmlog12.htm

And the survey version of it is available from the SSC archives:

ssc install svyselmlog

I hope this helps,
Rafa

----- Original Message ----- 
From: "Srini Sankaraguruswamy" <sankaras@georgetown.edu>
To: <statalist@hsphsun2.harvard.edu>
Sent: Tuesday, February 14, 2006 9:53 PM
Subject: st: calculating inverse mills ratios from a multiple logit.



> Dear Statlist members
> I am trying to calculate inverse mills ratios from a first stage multiple 
> logit
> y ~ f(explantory variables)
> where y takes on 4 values 0,1,2,3 and the y are not ordered.
>
> I was wondering if anyone has coded this in stata.
>
> I would like to then use these inverse mills ratios in the second stage 
> regression as
>
> Z ~ f(new explanatory variables, IME1-IME3)
>
> Thanks so much for your help
> Srini
>
>
>
> *
> *   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/
*
*   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/