Extension of Kendall's Concordance Test Where Ties Are Allowed

摘要

Kendall's measure of concordance and associated probability tables establishes critical regions for testing the null hypothesis of random rankings of M items by N judges. The M items are ranked according to an agreed upon criterion, such as beauty and cost effectiveness, or to the efficacy of a compression scheme, which was the basis for this Research Note. The concordance probability values are approximated by the F-distribution and the tests require that each judge produce a valid ranking of the M item, i.e. ties among the items are not allowed. The purpose of this work was to develop exact probability tables for a limited number of values of the parameters N and M where, in fact, ties were allowed. (Author)