I am tasked with running a sample size calculation and also statistical analysis of a study looking at the agreement between two 'tools'. This is a binary outcome so looking mainly at the different forms of kappa. However, here is the complexity to the data:
- There are multiple cases that we intend to recruit. i.e. people that we are measuring
- There are multiple raters too. i.e. multiple people using the two tools to make the measurements.
The focus is on the agreement between the two tools. Therefore, I want to run a sample size calculation to recruit a good combination of raters and cases.
It seems that this extends beyond the current capabilities of kappa, which can't account for this many dimensions to the data. Any ideas on how I could proceed?
Thanks in advance, Dean
