Date:         Fri, 14 May 2004 12:43:36 +0800
Reply-To:     Peggo Lam <peggo.lam@polyu.edu.hk>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Peggo Lam <peggo.lam@polyu.edu.hk>
Subject:      Re: Problems with intra-class correlation
Content-Type: text/plain; charset="iso-8859-1"

Hi Michael,

A low (or even negative) ICC value may also due to lack of variability among the subjects. Personally I will compute the 95% limits of agreement which represents how far a difference between any two repeated measurements from a single subject can be.

If you need further reference, the paper by White and van den Broek (2004) contrasted a number of methods for assessing reliability.

White, S.A., and van den Broek, N.R. (2004). Methods for assessing reliability and validity for a measurement tool: a case study and critique using the WHO haemoglobin colour scale. Stat Med, 23, 1603-19.

Cheers,

Peggo.