R help - Jun 2005 - Analysing ordinal/nominal data

Hi!

I'm looking for a solution to analyse data, which consists of 
dichotomous  responses (yes/no) for 2 multinomial ordinal variables. I 
was trying glm() and got hierarhical models treating all variables as 
nominal, but I can't figure out how to tell glm() to use a model for 
ordinal data like this:

log(Mij) = intercept + X + Y + Z + beta*(x-x')*(y-y')

where beta is a regression factor for interaction between X and Y.

Do you know a trick to code it in R or point me to some documentation?

thanks a lot,

Piotr Majdak

On Thu, 16 Jun 2005, Piotr Majdak wrote:
> I'm looking for a solution to analyse data, which consists of
> dichotomous  responses (yes/no) for 2 multinomial ordinal variables.
Please explain how you get a binary response for a `multinomial ordinal 
variables'?  If you intend these variables to be explanatory variables, in 
what sense are they `multinomial'?
> I was trying glm() and got hierarhical models treating all variables as 
> nominal, but I can't figure out how to tell glm() to use a model for 
> ordinal data like this:
>
> log(Mij) = intercept + X + Y + Z + beta*(x-x')*(y-y')
>
> where beta is a regression factor for interaction between X and Y.
What are Mij, X, x, x', Y, y, y' and Z?  One normally fits a logistic 
regression to a binary response.
> Do you know a trick to code it in R or point me to some documentation?
Probably no `trick' is required, but we need to start from a complete and 
accurate description of the model you want to fit.

This could be a simple as

R    2-level factor
U, V ordered factors

glm(R ~ U + V + as.numeric(U)*as.numeric(V), family = binomial)

-- 
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