[R] Analysis of ordinal categorical data

[Prof Brian Ripley]

On Wed, 5 May 2004, John Fox wrote:

> One approach to an ordinal response variable is the proportional-odds model,
> implemented in the MASS package as polr(). The proportional-odds assumption
> may not hold, however.

And you can find out by comparing the fit with a multinom fit -- see the 
MASS scripts for a worked example.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595