[R] robust polr

[Prof Brian Ripley]

On 20/07/2012 22:06, Anthony Dick wrote:
> Hello,
>
> I am wondering if there is a robust estimation version of polr(),
> similar to the lm() -> rlm(). If not, can anyone suggest a way to do
> robust estimation with polr()?
>
> Thanks,
>
> Anthony

No: What would it mean to have a long-tailed distribution of a ordered 
categorical response?

There are ways to allow for misclassification  of training data in 
logistic regression, including polr, but they are very different in 
flavour and implementation from robust regression.

-- 
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