Message-ID: <2A8DE2E40F52E049B8FBB4634D32AFE602CA1C6F@groamrexm01.amer.pfizer.com>
Date: 2005-09-12T14:55:24Z
From: Rogers, James A [PGRD Groton]
Subject: Help with a more flexible funtion for multiple comparisio n of means
Jose -
Before implementing SNK and Duncan's, you may want to be aware of some
criticisms of these methods:
>From Hsu (1996),
"Newman-Keuls multiple range test is not a confident inequalities method and
cannot be recommended."
"Duncan's multiple range test is not a confident inequalities method and
cannot be recommended either. In the words of Tukey (1991), Duncan's
multiple range test was a 'distraction' in the history of multiple
comparisons, amounting to 'talking 5% while using more than 5%
simultaneous'"
@Book{Hsu1996,
author = {Jason C. Hsu},
title = {Multiple Comparisons: Theory and Methods},
publisher = {Chapman & Hall},
year = {1996}
}
-- Jim
LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}