Rank range test for equality of dispersion

Abstract

This paper exploits the computational simplicity of the range of a set of data to formulate a twosample scale test called the Rank Range test. The performance of the test statistic is compared with other tests of scale. The exact distribution of the Rank Range test statistic is generated empirically
through the unconditional permutation approach by actually obtaining all the distinct permutations of the ranks of the variates in an experiment; this exposes the trimodal nature of its probability distribution. The tables of exact critical values are produced.