Date:         Thu, 8 Feb 2007 09:51:25 -0800
Reply-To:     Dale McLerran <stringplayer_2@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Dale McLerran <stringplayer_2@YAHOO.COM>
Subject:      Re: Catmod and Bowker's test
In-Reply-To:  <6250203B042D8349A920AA610383093602735804@UTHEVS3.mail.uthouston.edu>
Content-Type: text/plain; charset=iso-8859-1

--- "Swank, Paul R" <Paul.R.Swank@UTH.TMC.EDU> wrote:

> I am trying to recreate some analyses done in an article (Standard
> and Nonstandard Log-Linear Symmetry Models for Measuring Change in
> Categorical Variables, von Eye, C. & Spiel, ,C., The American
> Statistician, vol. 50, Nov., 1996). The idea is to recast the Bowker
> test for symmetry in a k by k matrix as a log-linear model. This can
> be done by specifying the design matrix directly. I have managed to
> recreate the results, at least as far as the parameter estimates are
> concerned, but the reported chi-squares are different for the model
> test (6.92 vs. 6.15 as it should be, each with 3 df). The authors
> also report the chi-square for the main effects model which also
> differs from my SAS output when I run that model (50.74 vs 66.643 as
> they report (both df=4). I cannot figure out why I can get the right
> parameter values but not the correct model chi-squares. Any ideas?
> My code is below.
>
> data example1;
> input pv1 pv2 count;
> datalines;
> 1 1 11
> 1 2 2
> 1 3 8
> 2 1 4
> 2 2 5
> 2 3 12
> 3 1 1
> 3 2 13
> 3 3 120
> ;
> proc catmod;
> weight count;
> model pv1*pv2 = _response_ / design;
> loglin pv1 pv2;
>
> proc catmod data=example1;;
> weight count;
> model pv1*pv2 = (1 0 0 0 0,
>                  0 0 1 0 0,
>                  0 0 0 1 0,
>                  0 0 1 0 0,
>                  0 1 0 0 0,
>                  0 0 0 0 1,
>                  0 0 0 1 0,
>                  0 0 0 0 1) (1='main effect 1',
>                              2='main effct 2',
>                              3='symmetry 1',
>                              4='symmetry 2',
>                              5='symmetry 3');
> run;
>
> Paul R. Swank, Ph.D. Professor
> Director of Reseach
> Children's Learning Institute
> University of Texas Health Science Center-Houston
>

Paul,

With regard to the chi-square test for symmetry, are you certain
that the value reported by the authors is correct?  Bowker's test
statistic from PROC FREQ has a value of 6.15.  Perhaps the authors
reported the wrong statistic.  That does happen occassionally.

Dale

---------------------------------------
Dale McLerran
Fred Hutchinson Cancer Research Center
mailto: dmclerra@NO_SPAMfhcrc.org
Ph:  (206) 667-2926
Fax: (206) 667-5977
---------------------------------------



____________________________________________________________________________________
Be a PS3 game guru.
Get your game face on with the latest PS3 news and previews at Yahoo! Games.
http://videogames.yahoo.com/platform?platform=120121