I have been asked to calculate a 3-way kappa statistic (for agreement beyond
chance) in a situation with 3 observers.
In SAS, the following code compares observer 2 and observer 3 for each level
of observer 1 and calculates an overall kappa statistic.
proc freq; tables obs1*obs2*obs3 /agree; run;
The overall kappa is dependent on the order. e.g. obs2*obs1*obs3 will produce
a different overall kappa.
Is this what is meant by a 3-way kappa? If this is something different, does
anyone have a reference for calculating one?
I will post a summary of replies.
Thanks in advance.
Angela Crook
Public Health
KCW Health Authority
London W2 6LX
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|