st: Mlogit and cross validation

[Sheela Athreya <athreya@tamu.edu>]

Hello,

I have a question about using multinomial logit--it's more of a  
theoretical than technical question, but I am hoping someone who uses  
this method can help.

I am using multinomial logit in the same way that biologists use  
discriminant function analysis--in this case, to determine the  
posterior probability of group membership of a set of 200 skulls,  
each of which is given an a priori assignment to one of five groups.

When using DFA, I know that it is necessary to cross-validate, i.e.,  
do a "leave one out" classification so that the analysis is rerun (in  
this case) 200 times, each time leaving one specimen unclassified so  
that it doesn't influence the matrix structure and, by extension, its  
own posterior probability values.

Is it necessary to do the same with multinomial logit?  And if so, is  
there a way to do it in Stata that does not require me to do it by hand?

Thank you in advance for your help.

Sheela

*******************************
Sheela Athreya, Ph.D.
Assistant Professor
Department of Anthropology
Texas A&M University
4352 TAMU
College Station, TX  77843
phone: 1-979-845-4785
fax: 1-979-845-4070



*
*   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/