Date: Tue, 6 Sep 2005 09:38:41 -0400
Reply-To: "Chan, Loretta" <Loretta_Chan@klinegroup.com>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Chan, Loretta" <Loretta_Chan@klinegroup.com>
Subject: Scheffe VS Trimming Sample Sizes
Content-Type: text/plain; charset="US-ASCII"
I was given 3 data series and told to conduct a test to compare 3 groups
of means. After looking at the uneven sample sizes of the 3 groups (60
VS 40 VS 25), I ran an ANOVA and the Scheffe test. I was subsequently
told to trim off the excessive data points. In other words, I am to run
ANOVA and a contrast with 3 groups of 25 cases each.
If my memory serves me right, I was told to avoid throwing away data if
I can find an equally robust technique to achieve the same goal. Are
there deficiencies in the Scheffe test that I am not aware of? Is it
better to perform a contrast or a post-hoc Scheffee?