Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: nonsense category in ordered logit


From   Marcello Pagano <pagano@hsph.harvard.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: nonsense category in ordered logit
Date   Tue, 22 Feb 2005 09:41:12 -0500

Sounds good to me, assuming you have enough data, except
that the interpretation rightfully is a little more involved. Something like, the same covariates affect the probability
of 2 and 3 similarly relative to a constant, although 2 is more(less) likely than 3; i.e. that constant *may* be different.

So the statement you make is relative to the covariates, and is thus
also quite dependent on the model you have used to introduce
these covariates.

m.p.



E. Michael Foster wrote:


Good morning,


I don't know that I understand your question. Suppose we get rid of all the
other outcomes and just concentrate on the outcomes 2 and 3. Now we
have a binomial. Are you asking whether the covariates are all equal
to zero, or do you want further that the probability of each is one-half?

Nope, I'm happy to let the constant / thresholds determine the proportion in each of those two categories.

So, if I just run an ordinary logit on that distinction (2 v. 3) and do a likelihood ratio test for the betas, I'd have demonstrated what I hope to show? I find it intuitive, but there's often something technically wrong with my intuition.

thanks,
Michael


*
* 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/
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index