Date:         Sat, 5 Feb 2005 20:40:59 +0100
Reply-To:     Sakari Lemola <sakari.lemola@psy.unibe.ch>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Sakari Lemola <sakari.lemola@psy.unibe.ch>
Subject:      df-question in a repeated measures GLM (multivariate)
Comments: cc: Carola Smolenski <carola.smolenski@psy.unibe.ch>
Content-type: text/plain; charset="ISO-8859-1"

Dear list

We have got a problem with multivariate multifactorial repeated measures
analysis of Variances (GLM):
First we got outputs on "multivariate Tests" (between  and within-subjects)
Secondly we got the Mauchly's Test of Sphericity
Thirdly we got "tests of within-subjets Effects" and there again is an
overview on the multivariate within-subjects effects. And here comes the
point: the Error-df-values (and herewith the F-values and p-values) are very
different from the values we got under "multivariate Tests" (The
Error-df-values are more than 4 times higher!).

Does somebody know what happened? Which values should we take if there are
such differences?

Thank you very much for any help!

Sakari Lemola & Carola Smolenski
--
Sakari Lemola, lic.phil.
Institut für Psychologie, Universität Bern
Muesmattstrasse 45, CH-3000 B E R N / Switzerland
(+41) 31 631 4033