On Fri, 25 Oct 2002 13:13:40 -0700
Howard Stauffer <hbs2@humboldt.edu> wrote:
> Oct. 25
>
> Non-parametric statisticians,
>
> Please resolve a controversial issue for me over use of the
> Wilcoxin-Mann-Whitney rank sums test. Is it appropriately applied to two
> populations that are heteroscedastic? References seem
> to be in conflict over this question.
>
> Thanks.
>
> Professor Howard Stauffer
> Humboldt State University
Yes, it is appropriate but it may not test a simple location shift in that
case. In other words, it won't test purely for a difference in means or
medians. It always tests a more general hypothesis, though, namely H0: Prob[X
> Y] = 0.5. X > Y means "a randomly chosen observation from the first
treatment group is larger than a randomly chosen observation from the second
group".
--
Frank E Harrell Jr Prof. of Biostatistics & Statistics
Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences
U. Virginia School of Medicine http://hesweb1.med.virginia.edu/biostat
|