Date: Thu, 5 Nov 1998 09:27:30 +0100
Reply-To: ivanouw <ivanouw@post8.tele.dk>
Sender: "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From: ivanouw <ivanouw@POST8.TELE.DK>
Subject: Reliability - data structure
Content-Type: text/plain; charset="iso-8859-1"
I would be very glad if someone could help me with this problem:
I am analyzing data from a psychological test (Rorschach) for a Danish normative study and am having a problem trying to estimate the intertester reliability.
The test gives a lot of variables and calculated summary scores and indexes. Every tested person is represented as two cases/lines in the SPSS because of two raters scoring the same test. I would like to calculate
1. Percent angreement between the two reaters for every score
2. Percent agreement between the raters for groupings of scores of the same type (as defined by the psychological test)
3. Cohens Kappa - and weighted Kappa - for the same individual scorings and groupings of scores
4. Intra class correllations for some summary scores and indexes for each testperson.
The structure of the data is like this: The tested persons: A, B, C ... The two scorers 1 and 2
Variables Summary calculations and indexes
R W D Dd S .. . . . . Afr Wsum6
A1 22 10 12 0 2 0.56 36
A2 22 9 12 1 3 0.40 26
B1 24 13 9 2 1 0.77 15
B2 24 15 8 1 3 0.60 17
.
.
.
I am both unsure about choosing procedures in SPSS to perform these calculations and also how to handle this data matrix with the two different entries for each.
Can anybody help?
Sincerely
Jan Ivanouw
Research director Personality assessment project
University Hospital of Copenhagen
e-mail: ivanouw@post8.tele.dk
