Date:         Fri, 16 Oct 1998 13:48:38 -0800
Reply-To:     Dennis Gene Fisher <afdgf@UAA.ALASKA.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Dennis Gene Fisher <afdgf@UAA.ALASKA.EDU>
Subject:      Repeated measures and geometric mean
Content-type: text/plain; charset=us-ascii

We are working on an analysis of antibody titers that were taken at
three different points in time for each participant.  We want to analyze
the data using repeated measures ANOVA.  The problem is that any figures
of the mean titers are supposed to be in geometric means, not
arithmetric means.  The only form of repeated measures ANOVA that we
know about uses arithmetic means.  Is it possible to tell PROC GLM (or
some other proc) to use geometric means instead of arithmetic means, or
are we supposed to transform the data in some fashion before we do the
analysis, or would it be OK to do the analysis on untransformed
arithmetic means and still do the graph with geometric means???  What do
people normally do in this situation.  We would appreciate all
opinions.  Thanks in advance.
 
--
Dennis G. Fisher, Ph.D.
IVDU Project
University of Alaska Anchorage
3211 Providence Drive
Anchorage, AK 99508