A measure of agreement for interval or nominal multivariate observations

This article addresses the problem of accounting overall chance-corrected i nterobserver agreement among the multivariate ratings of several judges. Mo difying an approach by Berry and Mielke, a generalization of Cohen's kappa coefficient is proposed. The generalized statistic accounts agreement for m ultivariate interval or nominal observations among several judges. The stat istic's metric is conventional, and in the univariate case it is equivalent to existing extensions of the kappa coefficient to several observers, as w ell as to a certain intraclass correlation coefficient. Application is exem plified using hypothetical and real-life data sets.