[R-sig-eco] Multinomial logist regression

[Karl Ove Hufthammer]

On Mon, 8 Feb 2010 18:29:06 -0500 Steve Friedman <friedman.steve-
Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org> wrote:

> Which package and function is more commonly used for ecological
> investigations of this kind?

I don't know, but for multinomial models in general, I have been happy 
with the 'multinom' function in the 'nnet' package.

> How can I develop a plot illustrating the probabilities of each of the six
> response vegetation types for a given concentration of nutrients (the
> independent predictors) ?

The 'effects' package is perfect for these kinds of plots, and has 
recently been extended with support for multinomial models:
http://www.jstatsoft.org/v32/i01

-- 
Karl Ove Hufthammer