[R] Evaluating model fits for ordinal multinomial regressions with polr()

Ben Haller rhelp at sticksoftware.com
Wed Feb 16 17:41:44 CET 2011


  Hi all.  I'm just starting to explore ordinal multinomial regression.  I've been doing logistic regression in the past, and I've gotten used to various functions for getting pseudo-R2 metrics and other metrics to evaluate the quality of a model.  For example:

val.prob()
NagelkerkeR2()
ClassLog() 
LogRegR2()

  But these don't seem to work for the models generated by polr().  Does anyone have any recommendations for ways to evaluate the goodness-of-fit and such things for ordinal multinomial regression models?  I've searched Google and looked at the docs for the package (MASS) that defines polr(), but haven't discovered anything.

  Thanks!

Ben Haller
McGill University



More information about the R-help mailing list