Date: Fri, 4 Jul 2003 14:59:24 +0300
Reply-To: Akin Pala <akin@DR.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Akin Pala <akin@DR.COM>
Subject: Re: REPEATED, UN, CS intraclass correlation
Is there any way to see messages related to me at the top instead of having
to scroll down ever time I open outlook express 6? How do you guys do it?
Anyhew, what I do is repeated and I have growing goat kids. Growth measures
(weight, height etc) were taken every week for 17 weeks. So I might think
there is an order within kid. To answer your question, I guess changing the
order would certainly change the correlation structure. So I am assuming is
the ante (1) method is just fine, here is the corr matrix for the first
animal, untill week 9, calculated with CS:
Row Col1 Col2 Col3 Col4 Col5 Col6 Col7
Col8 Col9
1 1.0000 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
2 0.3961 1.0000 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
3 0.3961 0.3961 1.0000 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
4 0.3961 0.3961 0.3961 1.0000 0.3961 0.3961 0.3961
0.3961 0.3961
5 0.3961 0.3961 0.3961 0.3961 1.0000 0.3961 0.3961
0.3961 0.3961
6 0.3961 0.3961 0.3961 0.3961 0.3961 1.0000 0.3961
0.3961 0.3961
7 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 1.0000
0.3961 0.3961
8 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
1.0000 0.3961
9 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 1.0000
10 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
11 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
12 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
13 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
14 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
15 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
16 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
17 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961 0.3961
0.3961 0.3961
here is the corr matrix for the first animal, untill week 9 calculated
with ante(1):
Estimated R Correlation Matrix for KNo 2202
Row Col1 Col2 Col3 Col4 Col5 Col6 Col7
Col8 Col9
1 1.0000 0.8166 0.7718 0.7551 0.7284 0.6962 0.6725
0.6503 0.6337
2 0.8166 1.0000 0.9451 0.9247 0.8919 0.8525 0.8235
0.7963 0.7760
3 0.7718 0.9451 1.0000 0.9784 0.9437 0.9021 0.8714
0.8426 0.8211
4 0.7551 0.9247 0.9784 1.0000 0.9646 0.9220 0.8906
0.8611 0.8392
5 0.7284 0.8919 0.9437 0.9646 1.0000 0.9559 0.9233
0.8928 0.8700
6 0.6962 0.8525 0.9021 0.9220 0.9559 1.0000 0.9660
0.9340 0.9102
7 0.6725 0.8235 0.8714 0.8906 0.9233 0.9660 1.0000
0.9669 0.9423
8 0.6503 0.7963 0.8426 0.8611 0.8928 0.9340 0.9669
1.0000 0.9745
9 0.6337 0.7760 0.8211 0.8392 0.8700 0.9102 0.9423
0.9745 1.0000
10 0.6014 0.7364 0.7792 0.7964 0.8257 0.8638 0.8942
0.9248 0.9490
11 0.5846 0.7159 0.7575 0.7742 0.8027 0.8397 0.8693
0.8991 0.9226
12 0.5697 0.6976 0.7382 0.7544 0.7822 0.8183 0.8471
0.8761 0.8990
13 0.5603 0.6860 0.7259 0.7419 0.7692 0.8047 0.8331
0.8616 0.8841
14 0.5525 0.6766 0.7159 0.7317 0.7585 0.7936 0.8215
0.8496 0.8719
15 0.5376 0.6583 0.6965 0.7119 0.7381 0.7722 0.7994
0.8267 0.8483
16 0.5113 0.6261 0.6625 0.6772 0.7020 0.7345 0.7603
0.7863 0.8069
17 0.4919 0.6023 0.6373 0.6514 0.6753 0.7065 0.7314
0.7564 0.7762
Ante correlations are much higher and are different in different weeks. I
think when I say ante (1) as opposed to saying somthing like ante (2), it
takes first degree corr. only, adjacent corr only. Not sure if I am right
though...
I am in animal science and I had been doing nothing but animal science
stuff, but it seems that next week I will have some repeated gender data and
later something from a clinical trial with no correlation within subject (or
that is what they say). So this is a verrry helpful discussion for me!
--
Akin Pala, Ph.D.
http://akin.owns.it
Tel: (286) 218 00 18 ext. 1349
"Akin Pala" <akin@dr.com> wrote in message news:3f0191b7@shknews01...
> I can calculate intraclass correlation if I type
> proc mixed etc.;
> repeated/type=cs subject=id;
> by correlation=common variance/common var+residual variance which is
> something like
> 0.74/(0.74 + 0.57) = 0.57.
> RCORR gives me a 1.
> Using UN with repeated/type=un subject=id;
> gives me some other number like 1.30
> Is there any way to calculate intraclass correlation using Unstructured
> (UN).
> Also, is there any way to calculate that number in proc genmod repeated
> statement? It does not accept rcorr; not that I saw any use for it :)
> So, is there any one out there who knows the answer to those? I am just
> confused about UN vs CS and I want to calculate intraclass corr using UN
> because UN gives me Akaikes IC=+89.1 while CS gives me -90.1 and Schwarz
> bayesian is -90.1 for UN and -92.1 for CS.
> Thanks and I appreciate any answer...
>
>
> --
> Akin Pala, Ph.D.
> http://akin.owns.it
>
>
