[R] polr question

Prof Brian Ripley ripley at stats.ox.ac.uk
Wed Nov 8 16:18:51 CET 2000

> From: "Graeme D Hutcheson" <mewssgdh at fs1.ed.man.ac.uk>
> To: r-help at hypatia.math.ethz.ch
> Date: Wed, 8 Nov 2000 14:44:40 GMT
> Subject: [R] polr question
> Hi,
> I am trying to reproduce in R an analysis reported in Agresti
> (1996; page 214) and Agresti and Finley (1997; page 602) which
> uses the proportional odds logistic regression model to compare
> political ideology.
> I am using the "polr" function from the MASS library which provides
> a number of statistics, but does not appear to provide -2 LOG L
> for the intercept only model. This statistic is useful as it
> would enable me to reproduce exactly the analysis which is
> currently reported using SAS and enble me teach a bit more using R.

Did you fit an intercept-only model to get it?  As in the example
on the help page

     options(contrasts=c("contr.treatment", "contr.poly"))
     house.plr <- polr(Sat ~ Infl + Type + Cont, weights = Freq, data = housing)
     house.plr0 <- polr(Sat ~ 1, weights = Freq, data = housing)

Note that -2 LOG L is not actually well-defined as it depends on the
grouping of the data assumed, and so is not a statistic.  I assume
that you want differences of that quantity, as in

house.plr0$deviance - house.plr$deviance

(maybe I need to write anova.polr ....)

Again, whether testing against the intercept-only model makes sense
is pretty debatable.   Are you seriously entertaining no effect
of the regressors?

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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch

More information about the R-help mailing list