Dear list,
I have data of 6 different coaches each rating two video fragments on a
small number of items, call it scale A. Each coach has to judge the same two
fragments.
(Actually I have this structure 4 times: the same judges judging two other
videos on scale B, another two videos on scale C and again two other video's
on scale D. As the scales are different, I think I have essentially four
times the same problem).
I was asked to assess the intraclass correlation coefficients.
Now I have three questions:
1 How do I best setup the data file? The example in the book by Marija
Norusis (SPSS 13.0 Statistical Procedure Companion, chapter 18) suggests a
file with only two records for each videofragment being rated and 6
variables for each judge's score.
2 If so: Are these two records enough to assess the ICC?
3 If I get away with it so far: is the following syntax correct?
RELIABILITY
/VARIABLES=judge1 to judge6
/SCALE('ALL VARIABLES') ALL/MODEL=ALPHA
/STATISTICS=DESCRIPTIVE
/ICC=MODEL(MIXED) TYPE(CONSISTENCY) CIN=95 TESTVAL=0 .
Thank you in advance,
Antoon Smulders
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
