Dear Steffen,
I believe a model like that can be fitted using the following syntax:
proc mixed;
class subject time;
model Y=;
random intercept / subject=subject;
repeated time / type=un(1) subject=subject group=subject;
run;
However, it seems to me that the estimated subject-specific variances should
al be zero as only one observation is available for estimating each of them.
Best regards,
Geert Verbeke
Biostatistical Centre
K.U.Leuven
Kapucijnenvoer 35
B-3000 Leuven
Belgium
Tel: +32 16 33 68 91
Tel (sec): +32 16 33 68 92
Fax: +32 16 33 69 00
Web: http://www.med.kuleuven.ac.be/biostat/
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list
[mailto:[log in to unmask]]On Behalf Of Steffen Witte
Sent: vrijdag 28 februari 2003 10:04
To: [log in to unmask]
Subject: query / bivariate model, R-Matrix
Dear all,
I am trying to fit a bivariate model with SAS (proc mixed). The errors
(R-Matrix) have the following structure
ai b
b ci
for all i=1,...,n subjects. So, on the main diagonal are different
variances but the covariances should be the same for all subject. How do I
use the subject/group/type options in the repeated statement?
Best Steffen
---------------------------------------------------------
Dipl.-Math. Steffen Witte
Universitaet Heidelberg, Abteilung Medizinische Biometrie
Im Neuenheimer Feld 305, D-69120 Heidelberg
T: +49-(0)6221-56-4371, F: +49-(0)6221-56-4195
http://www.biometrie.uni-heidelberg.de
mailto:[log in to unmask]
---------------------------------------------------------
|