Date: Thu, 17 Mar 2005 16:44:00 -0500
Reply-To: "Penfold, Robert" <PenfoldR@pediatrics.ohio-state.edu>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Penfold, Robert" <PenfoldR@pediatrics.ohio-state.edu>
Subject: Mixed model "Hessian matrix is not positive definite"
Content-Type: text/plain; charset="iso-8859-1"
Recently I posted a question regarding error/warning during mixed modeling.
Turns out that it's important to scale your variables such that they are in the same order of magnitude (e.g. express population in 100's of thousands) in addition to centering.
Rob Penfold, PhD
Office of Clinical Science
Columbus Children's Hospital
700 Children's Drive, room J1413
Columbus, OH, 43205-2696
(614) 722-3589 voice
(614) 722-3544 fax