Re: st: test marginal effects in multinomial logit model

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Fri, 26/3/10, atsu june wrote:
> First, I run the following multinomial logit model.
> (Note: insure= Indemnity, Prepaid, and Uninsure)
>  
> mlogit insure age inc year89 year90 if
> men==1............(1)
> mfx2, at(mean) stub(mean)
> mlogit insure age inc year89 year90 if
> men==0............(2)
> mfx2, at(mean) stub(mean)
>  
> then I would like to test
> a)the difference between the marginal effect of "year89"
> and that of "year90" on "Prepaid" in eq.(1),
> b)the difference between the marginal effect of "year89"
> and that of "year90" on "Prepaid" in eq.(2), and
> c)the difference between a) and b)
>  
> Is it possible???

This boils down to interaction effects in the marginal
effect metric. This is not easy, work has been done to
implement such a thing for the binary logit case. See:

Edward C. Norton, Hua Wang, and Chunrong Ai (2004) 
"Computing interaction effects and standard errors in 
logit and probit models". The Stata Journal, 4(2): 
154--167
http://www.stata-journal.com/article.html?article=st0063

This should, in principle, be generalizable to the 
multinomial logit, but I don't think it has been done
yet. So, if you want to go along this road you'll need
to implement that yourself (or find someone els to do 
it for you). 

Alternatively, you can interpret your results in terms
of relative risk ratios, and interpret the interaction 
terms using the trick discussed here:
http://www.maartenbuis.nl/wp/interactions.html

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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/