st: Marginal Effects with Gllamm for Multinom Logit

["John Bunge" <jota.be@web.de>]

Hi

I have estimated a multinomial logit with random intercept using the gllamm routine, which seems to have worked fine. Now I would like to proceed to the computation of marginal effects, which however seems to be rather difficult. The gllamm helpfile does not provide any info on that. When I google "gllamm" and "marginal effects" I can only find some advice for a binary dc model, where, however, marg. effects computation was not based on gllamm at all, but rather on xtprobit, even though the enquirer was asking for advice on gllamm (http://statalist.blogspot.co.uk/2006/03/re-st-gllamm-and-marginal-effects.html).

Can anyone give a hint what I should do to compute the marg effects for this model? If I had to do it by myself, which would be the steps I would need to take, and what literature would be helpful to learn? I do not have any experience with programming in Stata, but I want to learn.

I am using this model in the context of a (unbalanced) panel. T=7 and N varies between 1000 and 2000 observations per year.

Thanks,

John.


*
*   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/