Date:         Fri, 2 Dec 2005 08:09:13 -0500
Reply-To:     jweedon@EARTHLINK.NET
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         jweedon@EARTHLINK.NET
Organization: http://newsguy.com
Subject:      Re: proc mixed covariance structure with non-alligned measures
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=us-ascii

On Thu, 1 Dec 2005 16:35:19 -0600, Howells_W@BMC.WUSTL.EDU ("Howells,
William") wrote:

>I am fitting a longitudinal mixed model in proc mixed and exploring
>serial correlation among the measurements within subject (repeated
>statement, type=).  The measures are not alligned, ie. different time
>points across subjects.  My question: am I restricted to the spatial
>covariance structures, eg. sp(exp), sp(gau), that take the exact times
>of measurement as a parameter?  Or does it also make sense to try the
>other structures like AR(1) and CS, etc.  Thanks in advance.

You can fit whatever model you want. Look at the fit criteria (AIC
etc.) and decide from those which is the best model.

JW