R help - Mar 2013 - Assumptions of oneway_test in coin package

Hi everybody,

I want to test differences in means among independent groups (more that
two). The assumptions of normality and homocedasticity are violated, but I
don´t want to use rank tests like Kruskal-Wallis, because I lose too much
information under this rank transformation. Can I use
oneway_test(VAR~GROUPS, method=approximate(B=9999)) in coin package in this
situation? I think, as this test uses permutation, I don´t have to meet the
assumptions of normality and homocedasticity. Am I right? Thank you very
much

A hug


                                               Diego PJ

	[[alternative HTML version deleted]]