Date:         Thu, 21 Jun 2007 04:20:57 -0700
Reply-To:     John Uebersax <jsuebersax@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         John Uebersax <jsuebersax@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Re: GLM linear trend test?
Comments: To: sas-l@uga.edu
In-Reply-To:  <1182421808.466275.76790@g4g2000hsf.googlegroups.com>
Content-Type: text/plain; charset="us-ascii"

Oops!  A typo.  The correct code is:

proc glm data = dsname;
   class x;
   model y = x;
   estimate  "linear"  x -3 -1  1  3;
run;

I neglected the minus sign in front of the first 3.

FYI, the following helpful webpage lists the weights for linear
contrasts with up 9 levels of the ordered categorical variable:

http://www.psych.yorku.ca/lab/sas/sasanova.htm

--
John Uebersax