-------- Original Message -------- Subject: Stability in CVA Date: Thu, 04 Sep 2008 17:22:56 +0200 From: [EMAIL PROTECTED] To: morphmet <[EMAIL PROTECTED]> Additional to Jim Rohlf's comment and Norm Campbell's note it should kept in mind that Canonical variates (or as it is also called, Multiple Discriminant function Analysis) suffers from a tiresome malaise in some work in that the coefficients of the canonical vectors are often unstable under resampling. Norm Campbell advocates the use of the correlations rather than the variance-covariances, which goes far towards stabilizing the coefficients, particularly if an aim of the analysis is to reify the coefficients. A biologically oriented account of Campbell's fundamentally important work is given in. Richard A. Reyment MULTIDIMENSIONAL PALAEOBIOLOGY Pergamon Press (Now Elsevier) (1991) Richard A. Reyment Paleozoologiska avdelningen Naturhistoriska Riksmuseet Stockholm ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
