3

I am involved with designing a new diagnosis method for a routine clinical procedure. I want to compare the agreement between operators to the agreement between methods. Is there such a thing as a multidimensional Kappa test, and if so how do I perform it?

EDIT: I could carry out a number of individual Kappa tests, finding the agreement between methods for each operator, as well as the agreement between operators for each method, but there may be some interaction between operators and methods which would be useful to know.

Doragan
  • 69
  • 3
  • Not quite clear what you need. Kappa is a test for k X k frequency table with the same categories in both "dimensions" of the table (the two raters). The exist also multi-rater kappa, which is for k X k X k... multiway table. – ttnphns Jul 30 '14 at 19:00
  • There will be two tests. Lets call them A and B. There will also be 2 (or more) raters. I'll call them X and Y. Both raters will rate 10-20 different reports in both A and B. Each patient will be tested with both A and B. There may be some interaction between the raters and the tests, so simply using a string of Kappa tests, to test the agreement of the raters, and the agreement of the tests, may not be the best method. – Doragan Jul 31 '14 at 08:33

0 Answers0