Specifying the right model, and using the appropriate covariance matrix
parameterization of the covariance matrix, a 95% CI can be obtained from
proc mixed in SAS. The calculation is based on a Wald-type test statistic,
using asymptotic normality of the ML or REML estimates.
Put the two variables underneath each other in one variable Y, and create a
factor (dummy) MEAS which is 1 for the first variable and 2 for the second.
Further, ID is a subject identification variable.
Fit following model:
proc mixed data=...;
class MEAS ID;
model Y=MEAS;
repeated MEAS / type=toeph subject=ID;
run;
I hope this helps.
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 70 15
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 Eric Wong
Sent: woensdag 11 juni 2003 3:36
To: [log in to unmask]
Subject: 95% CI for correlation coefficient
Dear all,
Is there a way to calculate 95% CI for Spearman's correlation coefficient?
Many thanks
Eric
|