Re: question about multinomial logit regression with frequencies

Thanks. This helps!

"Anon." <bob.ohara@xxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:4507DFF1.9050802@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
hyena wrote:
Sorry another question jump out.

in calculating P(i) = exp(bXj)/sum(exp(bXj)), Xj sum from 1 to number of
alternatives.

in the frequency data, how or where can we obtain the info of
alternatives that was not chosen since only the chosen ones are recorded
as counts?

With difficulty! If you how many classes there are, then the rest
obviously just have a count of zero. Maximising the likelihood can be
awkward at times (since the ML estimate of P(i) is 0, so the MLE for the
log of the count is -infinity).

If you don't know how mant classes you have, then you're OK only if you
know the form of the distribution of the expected counts (or proportions).
If you don't, then estimating the number of classes is impossible. In
fact, just using the data you can't even put a bound on it! Of course,
there is usually some information because of the particular context, so in
practice things may not be so bad. And depending on what exactly you're
doing, you may be able to ignore the missing classes.

HTH

Bob

--
Bob O'Hara

Dept. of Mathematics and Statistics
P.O. Box 68 (Gustaf H?llstr?min katu 2b)
FIN-00014 University of Helsinki
Finland

Telephone: +358-9-191 51479
Mobile: +358 50 599 0540
Fax: +358-9-191 51400
WWW: http://www.RNI.Helsinki.FI/~boh/
Journal of Negative Results - EEB: http://www.jnr-eeb.org



.

Relevant Pages