[R-sig-Geo] Test of independency in a contingency table for spatial variables

[Agustin Lobo]

Dear list,

I've classified a set of points according
to two different sets of descriptors (assume
a set of descriptors is biological, the other set environmental) into
N and M classes respectively, thus I have
2 vectors V1 and V2 recording each classification.
I've made a contingency table
conti <- table(V1,V2) of dimensions NxM

summary(conti) provides the chi statistic
to test for independence of both classifications,
but, I suspect, if I'm classifying points
the test should take spatial autocorrelation
into account. Which test would be appropriate?
The idea that comes to my mind is to randomize
the values of the descriptors among the points
and use a Monte Carlo approach. Is there any other
way?

Thanks!

-- 
Dr. Agustin Lobo
Institut de Ciencies de la Terra "Jaume Almera" (CSIC)
LLuis Sole Sabaris s/n
08028 Barcelona
Spain
Tel. 34 934095410
Fax. 34 934110012
email: Agustin.Lobo at ija.csic.es
http://www.ija.csic.es/gt/obster