Date: Tue, 22 Feb 2005 15:39:11 -0500
Reply-To: Roy Money <firstname.lastname@example.org>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Roy Money <email@example.com>
Subject: intraclass correlation coefficient question
Content-type: text/plain; format=flowed; charset=us-ascii
I have 79 ratings for 17 subjects - 5 ratings for 13 ss, 4 ratings for 2 ss
and 3 ratings for 2 ss - .
The ratings were done by 9 raters, only two of which rated all 17 subjects.
It seems that I have several choices
a one way random effects model with 3 raters on all 17 subjects
a one way random effects model with 4 raters on 15 subjects
a one way random effects model with 5 raters on 13 subjects
a two way random effects model for the 3 raters who did the same 14 subjects
However in each case I am unable to use all the ratings data.
Is there anyway to get a measure of overall agreement among all the raters
across all the subjects?