Date: Mon, 18 Jun 2007 17:16:18 -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: GLIMMIX vs. %GLIMMIX
Content-Type: text/plain; charset=iso-8859-1
--- Jeff <ap05money@ALPHAPOINT05.NET> wrote:
> Hi all,
> I am a SAS 8 user, which means I can't do spatial covariance mixed
> modeling with GLIMMIX. However, there is a %GLIMMIX macro online. The
> SAS PROC GLIMMIX page says that it has many of the features of
> %GLIMMIX, which tells me that %GLIMMIX will do anything that PROC
> GLIMMIX can do plus more. Is that true? Can I trust the older
> %GLIMMIX? Does anyone have experience with this?
It is a mistaken assumption to infer that the GLIMMIX macro "will
do anything that PROC GLIMMIX can do plus more." There is actually
more implemented in the procedure than is implemented in the macro.
But not everything available for the macro is found in the PROC.
In general, the two are very much alike. I would guess that 95%
of analyses have an exactly equivalent solution across the two
GLIMMIX flavors. The GLIMMIX procedure is much faster than the
There are some covariance specifications that are available only
to one or the other of the two flavors. I didn't recall that spatial
covariance structures were among those which were not available in
the GLIMMIX procedure, but would not doubt that for the earliest
GLIMMIX procedure release the spatial covariance patterns were not
available. The GLIMMIX modelling framework does not depend on the
type of covariance pattern one specifies, so you should be able to
use the GLIMMIX macro and specify a spatial covariance structure.
Certainly, the GLIMMIX procedure documentation which I have looked
at allows for spatial covariances structures.
Fred Hutchinson Cancer Research Center
Ph: (206) 667-2926
Fax: (206) 667-5977
Got a little couch potato?
Check out fun summer activities for kids.