[R] Response to query re: calculating intraclass correlations

[Paul Mathews]

Karl,

If you use one of the specialized packages to calculate your ICC, make sure that you know what you're getting. (I haven't checked the packages out myself, so I don't know either.)

You might want to read David Futrell's article in the May 1995 issue of Quality Progress where he describes six different ways to calculate ICCs from the same data set, all with different interpretations. All of the inputs for these calculations come from the repeated measures analysis which is described in P&B.

Good luck!

P.

Message: 1
Date: Tue, 16 May 2006 10:09:54 +0000 (GMT)
From: Karl Knoblick <karlknoblich at yahoo.de>
Subject: [R] Interrater and intrarater variability (intraclass
	correlation	coefficients)
To: r-help at stat.math.ethz.ch
Message-ID: <20060516100954.6267.qmail at web26504.mail.ukl.yahoo.com>
Content-Type: text/plain; charset=us-ascii

Hello!
 
I want to calculate the intra- and interrater reliability of my study. The design is very simple, 5 raters rated a diagnostic score 3 times for 19 patients.
 
Are there methods/funtions in R? I only found packages to calculate interrater variability and intraclass correlation coefficients for matrices of n*m (n subjects, m raters) - I have n subjects, m raters and r repetitions.
 
Can anybody help?
 

Thanks!
Karl

-- 
Paul Mathews
Mathews Malnar and Bailey, Inc.
217 Third Street, Fairport Harbor, OH 44077
Phone: 440-350-0911
Fax: 440-350-7210
E-mail:
        Paul: pmathews at apk.net, paul at mmbstatistical.com
        Rebecca: rmalnar at apk.net, bek at mmbstatistical.com
Web: www.mmbstatistical.com