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

st: RE: Re: Marginal effect after xtlogit, fe


From   "Steve Stillman" <stillman@motu.org.nz>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Re: Marginal effect after xtlogit, fe
Date   Fri, 14 Oct 2005 16:22:01 +1300

This is related to the same reason that the individual fixed effects cannot be estimated from a conditional logit model.  It isn't possible to estimate the probability of a positive outcome at the individual level from the conditional fixed effects model (and thus to estimate a true marginal effect).  You can only make a relative comparison between groups of people w/ different covariates.

The approach I have taken in the past when using fixed effect logit models is to calculate the predicted probability of a positive outcome conditional conditional on a single positive outcome for the individual (predict option pc1) and then use these predictions to calculate pseudo marginal effects at the overall sample mean of this predicted probability 

for example, if this was a simple logit model, the true marginal effects would be calculated as:

marginal effect = b-hat * fxb, where fxb = p*(1-p) and p = exp(x-bar*b-hat)/ (1 + exp(x-bar*b-hat))

so an approach for a fixed effect logit model would be to calculate

marginal effect = b-hat * fxb, where fxb = p-hat*(1-p-hat) where p-hat is the average sample predicted probability (using pc1 option) 

Hope this helps,
Steve

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu]On Behalf Of Christine Yeo
Sent: Friday, October 14, 2005 2:42 PM
To: statalist@hsphsun2.harvard.edu
Subject: st: Re: Marginal effect after xtlogit, fe


Adding on to my previous message, I am trying to estimate the retirement
decision, thus it takes the value of 1 if individual retires and 0
otherwise.

Thanks alot.
Chris



> Hi all,
>
> I am trying to predict the marginal effect after fixed effect logit model,
>
> when I use the  command mfx compute, predict (pu0) nose;
>
> all the dy/dx are 0.
>
> I am not sure why.
>
> Greatly appreciate your help.
>
> Regards,
> Chris
>


*
*   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/



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