Paired and Unpaired t-Tests for Agreement Coefficients

About this App

AgreeTest is an App that implements methods for testing the difference of 2 agreement coefficients for statistical significance. Two agreement coefficients are either correlated or uncorrelated. Each of the 2 agreement coefficients is calculated using ratings produced by a separate inter-rater reliability experiment that involves a sample of subjects and a roster of raters. If there an overlap between both subject samples or both rater rosters, then the resulting agreement coefficients are considered correlated, otherwise they are uncorrelated.

Testing corrrelated coefficients requires the use of a paired t-test, while testing uncorrelated coefficients requires the use of an unpaired (or independent) t-test. The paired t-test implemented in this App is based on my paper Testing the Difference of Correlated Agreement Coefficients for Statistical Significance.


Kilem L. Gwet, PhD