I would like to compute the interrater agreement of two raters for variables with two or more categories. However, I assume that there will be a prevalence problem (i.e. certain categories are likely to appear more often in the data than others). Thus, I am interested in also getting the prevalence-adjusted bias-adjusted kappa (PABAK). As far as I have seen, the package {IRR} does not offer this option, but the epi.kappa function of {epiR} does so. However, I would prefer to use {IRR} as I would like to get Cohen's Kappa (epiR seems to be based on Fleiss' Kappa). Is there any other way in which I could obtain Cohen's Kappa and PABAK in R?
Thank you very much!
Mark Stevenson here, developer of the epiR package. I've updated the epi.kappa function in epiR to include Cohen's kappa as an option. I'll submit to CRAN shortly. Contact me by email if you'd like a pre-release version.