Date:         Fri, 10 Aug 2007 05:41:45 -0700
Reply-To:     shiling99@YAHOO.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Shiling Zhang <shiling99@YAHOO.COM>
Organization: http://groups.google.com
Subject:      Re: Negative R-square
Comments: To: sas-l@uga.edu
In-Reply-To:  <200708101044.l7A3KI2Q021835@malibu.cc.uga.edu>
Content-Type: text/plain; charset="us-ascii"

On Aug 10, 6:44 am, list_...@YAHOO.CO.UK (SAS List) wrote:
> I take the residuals from a mixed model and use those as the dependent
> variable in an OLS regression. For one of the groups (dummy=3) i get a
> negative r-square. Is this normal?
>
> proc reg data=all;
> model y= k o p ;
> by dummy;
> run;

I am not sure how you got a negative r-square. It is imposible if a
linear regression model contains a constant term. But WITHOUT a
constant term r-square could be any real number and it means nothing
in linear regression context.

HTH