Date: Thu, 11 Jan 2001 16:16:56 +0100
Reply-To: ghattas badih <ghattas@LUMIMATH.UNIV-MRS.FR>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: ghattas badih <ghattas@LUMIMATH.UNIV-MRS.FR>
Subject: testing b=0 against b>0 in a linear regression
Content-Type: text/plain; charset="iso-8859-1"
Hello,
I am looking for a way to test in a linear model y=a +b1x1 + b2x2 + b3x3
hypothesis such as :
b1=0 vs b1>0
and
b1 > b2 > b3
How can I do this.
I have seen such things in proc model (SAS ETS) but I don't need to work in
a context of time series ...
Thanks.
----------------------------------------------------------
Ghattas Badih, Ph.D.
ghattas@lumimath.univ-mrs.fr
Département de maths, Fac des Sciences de Luminy, case 901
163 avenue de Luminy,
13288 Marseille cedex 09 - France
Tel : (33)(0) 4.91.82.90.89 FAX : (33) (0)4.91.82.93.56
Portable : 06.13.61.21.31
------------------------------------------------------------
| 
