Dear all,
I am trying to find a power calculation formula for a reliability study, where the primary outcome is binary and therefore I am looking at the kappa statistics. I want to know how many raters I would need (n), if the raters are rating a set of 3 to 5 videos, where the rating is 'yes' or 'no'.
The only reference I have found is the one by Allan Donner (Statistics in Medicine, 1998). Here it computes the number of vidoes that we would need for a fixed number of raters. However, I need the number of raters given 3 to 5 vidoes. It is difficult to re-arrange this and use it for the number of raters as it requires the kappa estimates that we would expect for each pair of raters. We don't know how many raters, or what there kappas are even likely to be. However we do know that the overall kappa is likely to be 0.7 (with a minumum of about 0.4).
Any help would be most welcome.
Many thanks in advance.
R. Lall (Phd.)
Warwick Emergency Care and Rehabilitation
Division of Health in the Community
Warwick Medical School
University of Warwick Gibbet Hill Campus
Coventry
CV4 7AL
|