Re: st: Nemenyi test

[Ronan Conroy <rconroy@rcsi.ie>]

On 2013 Lún 29, at 07:22, Célia Dechavanne wrote:

> Thank you for your help.
> My first mail was not enough clear, sorry for that. The five groups are small (minimum n=11 and maximum n=26). The KW test gives a significant difference between the groups, now I would like to compare 2 by 2. But I can't perform pairwise comparison, I would like to compare the shift between the five groups. I hope is better clear now.
> Thank you so much for the time spent to answer.

Five groups compared two by two tests a lot of hypotheses, and it's hard to imagine that you actually have that many hypotheses. 

The questions I would put to you are

1. What hypotheses generated the five groups?
2. Is there a natural baseline group?
3. Are the groups produced by a single factor (such as five different diagnoses) or produced by combining categories of two factors (for example: one factor is low/high dose aspirin, another normal or enteric-coated tablets, and the fifth group is a treatment-free group to act as control)
In a case like this, you have two factors (dose and coating) and an interaction to test. 


I have a terror of hypotheses that are generated by multiple comparisons. Try to test only hypotheses that make sense. Sit down with your theory before you go any further.

Or maybe this is purely exploratory research and there are no a priori hypotheses. In which case you're on your own.

Ronán Conroy
rconroy@rcsi.ie
Associate Professor
Division of Population Health Sciences
Royal College of Surgeons in Ireland
Beaux Lane House
Dublin 2


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/