[R] [Fwd: Trend test and test for homogeneity of odd-ratios]

Jacques VESLOT jacques.veslot at good.ibl.fr
Fri Aug 18 18:13:20 CEST 2006


I partly answered my question since independence_test() function in coin package apparently do 
Cochran-Armitage trend test just like Eric Lecoutre's function tabletrend() - slightly modified here:

 > independence_test(pheno ~ geno, data = dat2, teststat = "quad", scores = list(geno = c(0, 1, 2)))

         Asymptotic General Independence Test

data:  pheno by groups 1 < 2 < 3
chi-squared = 0.2268, df = 1, p-value = 0.6339

 > tabletrend(with(dat2, table(pheno, geno)))
[1] 0.6338308

-------- Message original --------
Sujet: Trend test and test for homogeneity of odd-ratios
Date: Wed, 16 Aug 2006 17:39:33 +0200
De: Jacques VESLOT <jacques.veslot at good.ibl.fr>
Pour: R-Help <r-help at stat.math.ethz.ch>

Dear r-users,

I am looking for some R functions to do Cochran-Armitage trend test for 2*3 tables (binary phenotype
vs. genotypes) and for testing the homogeneity of odds ratios within 2*3*k tables (binary phenotype
vs. genotypes vs. strata).

In R-Help archives, I've found a 2003 script by Eric Lecoutre for Cochran-Armitage trend test and a
script for Breslow-Day test for 2*2*k tables.

Could someone please tell we if there were some functions available on CRAN to do such tests ?

Thanks,

jacques



More information about the R-help mailing list