Date:         Wed, 14 Oct 2009 10:48:22 -0700
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: proc genmod - repeated measures
In-Reply-To:  <39c54bed-f584-4bd4-874c-5bdc4b8bf391@h14g2000pri.googlegroups.com>
Content-Type: text/plain; charset=us-ascii

--- On Wed, 10/14/09, nuria <nchapinal@YAHOO.COM> wrote:

> From: nuria <nchapinal@YAHOO.COM>
> Subject: proc genmod - repeated measures
> To: SAS-L@LISTSERV.UGA.EDU
> Date: Wednesday, October 14, 2009, 5:31 AM
> Hi,
>
> If I run a model like this, where treatment has only two
> categories
>
> proc genmod ;
> class cow treatmen;
> model sickness = treatment/dist=binomial type3;
> repeated subject=cow /type=cs;
> run;
>
> Do I need to look at the p-values for the type3 test or at
> Analysis Of
> GEE Parameter Estimates? I found sometimes the p-values are
> similar,
> but some other times aren't, and they may run into
> conflict.... for
> instance, treatment could be significant only in one of the
> tests...
>
> Thanks!
>

You should interpret the statistics in the Analysis of GEE
Parameter Estimates table.  However, I would suggest that
you use the GLIMMIX or NLMIXED procedures as an alternative
to fitting a GEE model.  The GEE model examines effects
marginally (average effect in the population) whereas the
GLIMMIX and NLMIXED procedures would produce conditional
effect estimates.  The conditional effect estimates ask
"What is the effect expected for an individual cow?"  My
guess is that would be the question you would prefer to
have answered.

Dale

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