On Jan 09, 2012 at 11:48am Christoph Liedtke wrote: > I should be detecting some non-significance between groups I and III at > least, but the test comes back with > extremely low p-values. Where am I going wrong?
Nowhere, I think. This does seem to be an error in coin. You should send your example to the maintainers of the package. Apart from the visual you have provided, other reasons for thinking that this is an error are the following. First, if you redo the analysis excluding habitat II, then the contrast is not significant, as expected. Secondly, if you repeat the full analysis using package nparcomp then you get the results you are expecting, based on the graphical representation of the data. See the examples below. ## drop habitat == II NDWD <- oneway_test(breeding ~ habitat, data = droplevels(subset(mydata, habitat != "II")), ytrafo = function(data) trafo(data, numeric_trafo = rank), xtrafo = function(data) trafo(data, factor_trafo = function(x) model.matrix(~x - 1) %*% t(contrMat(table(x), "Tukey"))), teststat = "max", distribution = approximate(B = 900000)) print(NDWD) print(pvalue(NDWD, method = "single-step")) ## use nparcomp library(nparcomp) npar <- nparcomp(breeding ~ habitat, data = mydata, type = "Tukey") npar Regards, Mark. ----- Mark Difford (Ph.D.) Research Associate Botany Department Nelson Mandela Metropolitan University Port Elizabeth, South Africa -- View this message in context: http://r.789695.n4.nabble.com/Unexpected-results-using-the-oneway-test-in-the-coin-package-tp4278371p4281329.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.