Date:         Thu, 15 Mar 2001 10:59:40 -0500
Reply-To:     "Van Kirk, Jeff" <VanKirk@PSYCHIATRY.UCHC.EDU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Van Kirk, Jeff" <VanKirk@PSYCHIATRY.UCHC.EDU>
Subject:      Significant Box's Min GLMRepeated Measures
Content-Type: text/plain; charset="iso-8859-1"

In the past whenever I've encountered a significant value for the Box's M,
I've regarded the multivariate approach as "inappropriate" and proceeded to
report the univariate results (adjusted, as necessary for violations of the
sphericity assumption).  I've recently come across a paper in which the
authors performed a square root transformation on the dependent measures and
proceeded to report both the multivariate and univariate statistics.  Is
this appropriate?  It  would seem to me that  once the transformation was
accomplished, additional univariate analysis on those transformed values
would be misleading.   Jeff

                Jeffrey R. Van Kirk
                Department of Psychiatry, MC-2103
                University of Connecticut Health Center
                263 Farmington Ave.
                Farmington, CT 06030
                ph:  860-679-3613  fax: 860-679-1316
                e-mail:  VanKirk@Psychiatry.UCHC.EDU
<mailto:VanKirk@Psychiatry.UCHC.EDU>