An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers.