Dear allstat,
The non-parametric Wilcoxon Rank Sum Test is the test of choice when
comparing two independent
samples with identical non-normal shape of their distribution. I am
looking for an alternative test to use when
the second assumption is not met, i.e. the two shapes look completely
different.
I am looking for an appropriate test to compare several ordinal outcomes
(scores with bounded range)
for two groups of patients with very different distributions. n1 is about
50 and n2 is about 10.
Further, I am looking for a test which can provide me with a quantifiable
effect size (e.g. difference in median).
I heard that the Median Test is the two-samples equivalent of the
one-sample Sign Test but I could not
find any references.
I thank you for your contribution,
Monica
------------------------------------------
Monica Daigl Cattaneo
AO Clinical Investigation and Documentation
Stettbachstrasse 6
8600 Dübendorf
Switzerland
Phone: +41 44 200 2468
Direct: +41 44 200 2472
Fax: +41 44 200 2460
[log in to unmask]
www.aofoundation.org
|