st: mlogit, mfx2, predict and a bug?

["Stephan Brunow" <Stephan.Brunow@mailbox.tu-dresden.de>]

Dear Statalisters,

there might be a bug in STATA. 
Here is what we did in short:

.mlogit...
.outreg2...
.mfx2...
.predict (probabilities of all choices)

Then we looked for the maximum of the choice probabilities to get a
hypothetical choice and tabulate it with the true choice. Obviously, some of
our categogies did not get chosen by the model at all.
However, if we do 

.mlogit
.predict (probabilities of all choices)

And do the same thing to compare modeled maximum choice and true choice, weg
et a <normal> result where all choices are met to some extend. 

Is there a bug or what might be the reason?!

Stephan

*
*   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/