R help - May 2004 - Scheire-Ray-Hare extension to Kruskal-Wallis

Hi,

I was wondering if it was possible to do the "Scheirer-Ray-Hare"
extension of the Kruskal-Wallis test using R, or if it was possible to use some
other non-parametric test for the following situation:

I have continuous data that is grouped by 2 factor variables.  I would like a
non-parametric test because the group variances are not equal.  I would like to
look for the effects of either factor variable and their interaction.

Thanks for any help,
Sarah 
	[[alternative HTML version deleted]]

"Sarah Skikne" <skikne at stanford.edu> writes:
> Hi,
> 
> I was wondering if it was possible to do the "Scheirer-Ray-Hare"
> extension of the Kruskal-Wallis test using R, or if it was possible
> to use some other non-parametric test for the following situation:
> 
> I have continuous data that is grouped by 2 factor variables. I
> would like a non-parametric test because the group variances are not
> equal. I would like to look for the effects of either factor
> variable and their interaction.
It is a common misconception that nonparametric tests do not assume
equal variances. In fact, they generally assume that entire
distributions are the same (under the null hypothesis). Testing for
interactions also pretty much assumes effects are on a particular
scale, which is not really very different from making parametric
assumptions. I'd consider a weighted linear model in such cases,
possibly supplemented by simulation or bootstrapping to get an
improved estimate of the p-values etc.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907