Richard Williams wrote:
>>My experience is that it is rare to have a model where the
proportional odds assumption isn't violated!<<
True, my feeling was that it was often capitalization on chance. How
often do you find it replicated?
>>Often, though, the
violation only involves a small subset of the variables, in which
case gologit2 can be useful.
Absolutely.
>> You might also want to consider more
stringent alpha levels (e.g. .01, .001) to reduce the possibility of
capitalizing on chance. You can also try to assess the practical
significance of violations, e.g. do my conclusions and/or predicted
probabilities really change that much if I stick with the model whose
assumptions are violated as opposed to a (possibly much harder to
understand and interpret) model whose assumptions are not violated.<<<
Right, this is about what I was thinking---be a more stringent about the
test. I wonder if anyone's done good simulation studies to see the
properties of the Brant test?
>>Finally, while I happen to like gologit2, there are a lot of other
categorical and ordinal models out there that might be worth a look
depending on the problem.<<
I hope no one misunderstands me on this point: The generalized model is
very useful but I think giving up the proportional odds assumption in
the face of "small but significant" violations is a bad idea. It's kind
of like taking the goodness of fit statistics in SEM or confirmatory
factor analysis too seriously.... If the violations are real, that's a
totally different question.
Jay
*
* 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/
