Date:         Mon, 19 Aug 1996 09:21:35 +0200
Reply-To:     Hans-Peter Piepho <piepho@WIZ.UNI-KASSEL.DE>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Hans-Peter Piepho <piepho@WIZ.UNI-KASSEL.DE>
Subject:      Re: GLM with Tukey's option
Comments: To: david izraylovich <aldair@WORLD.STD.COM>

>Hi,everyone -
>
>Could anyone explain to me one phenomenon. I am using GLM for one-way
>ANOVA. Between-group factor has 4 levels. It gives me the overall
>significance with p-value 0.01. I am applying then Tukey option
>in means statement for pairwise comparisons and I do not get any
>significant differences. After changes alpha in means statement
>to 0.10 I finally got significance
>
>What could the matter be?
>
>Thanks.
>David Izraylovich
>aldair@world.std.com
>
>
The overall F-Test is based on the F-distribution, while Tukey's Test is
based on studentized ranges. Generally, if you get a significance by one
test, this does not mean, that another test of the same or a similar
hypothesis must also be significant. This has to do with error rates, power,
etc.....
 
Most importantly, if you do Tukey's test, forget the F-test. Tukey controls
experiment-wise error rate in the strong sense. So there is no need to
"protect" it by a preliminary F-test (as with Fisher's protected LSD).
 
Hans-Peter
_______________________________________________________________________
Hans-Peter Piepho
Institut f. Nutzpflanzenkunde  WWW:   http://www.wiz.uni-kassel.de/fts/
Universitaet Kassel            Mail:  piepho@wiz.uni-kassel.de
Steinstrasse 19                Fax:   +49 5542 98 1230
37213 Witzenhausen, Germany    Phone: +49 5542 98 1248