Re: st: Standard error of the marginal effect or predicted probability in a conditional logit

[Glenn Hoetker <ghoetker@igb.illinois.edu>]

Thank you, Maarten.  Unfortunately, I don't think pu0 is what I need, since I have a situation with exactly one choice per group.  So, I think I need pu1.  I didn't mention that in my earlier email, trying to keep things simple and brief.

I appreciate the quick help, regardless.

G.

On Aug 19, 2010, at 12:58 PM, Maarten buis wrote:

> --- On Thu, 19/8/10, Glenn Hoetker wrote:
>> Can anyone point me towards a way to calculate the standard
>> errors of either the marginal effects or the predicted
>> probabilities of a  conditional logit model 
> 
> -margins- in combination with -predict(pu0)- option. This will
> force the group specific deviations to be zero, i.e. you are
> getting the probabilities for the average group. That is 
> necesary because you need those deviations to compute the 
> probabilities and marginal effect if you don't set them at 0, 
> and Stata does not estimate these group specific deviations, 
> so they won't appear in the variance covariance matrix, so you 
> can't use the delta method for anything other than pu0.
> 
> Hope this helps,
> Maarten


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/