Date: Thu, 6 Jul 2006 22:21:44 -0700
Reply-To: David L Cassell <davidlcassell@MSN.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: David L Cassell <davidlcassell@MSN.COM>
Subject: Re: ARH(1) or SP(POW)
In-Reply-To: <200606261630.k5QAlPH0031279@malibu.cc.uga.edu>
Content-Type: text/plain; format=flowed
luciendiby@YAHOO.FR wrote:
>
>Dear all,
>
>I have data recorded unevenly (47, 107, 160 and 190 days) over time.
>When looking for the best covariance structure, the ARH (1) structure
>gives best result than the SP(POW) structure which is recommended for
>unequally spaced data! Is this finding realistic?
>
>Thank in advance!
>
>Lucien
Yes, your finding is realistic. The SP(xxx) spatial structures are
recommended
for unequally spaced temporal data because *sometimes* the effect at time t
decreases with time, in its effect on time t+1. It is entirely possible
that this
is not the case for your data.
It is also possible that one of the other SP(xxx) spatial covariance
structures
could fit the data better than the SP(POW), which does make a specific
assumption about the behavior of these 'decreases with time' I mentioned
above. It is also possible that the key factor is the 'H" in ARH(1), in
that the
variability from subject to subject that the ARH(1) structure addresses may
be at least as significant as the temporal structure within subject, so the
ARH(1) may only be doing a better job of describing the between-subject
behavior than the SP(POW).
Either way, ARH(1) may be an effective description of the behavior of your
data. Or you may want to plot the data out and see for yourself how the
behavior looks, giving you new clues about the covariance structure.
HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330
_________________________________________________________________
Express yourself instantly with MSN Messenger! Download today - it's FREE!
http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
