Basic program for a significance test for two clusters in euclidean space as measured by their overlap

Abstract

Abstract A BASIC program, OVCLUST, is given for a significance test for the distinctness of a pair of clusters of points in Euclidean space. The input data may be given in one of three forms: (1) the coordinates of the points in the full space; (2) the distances of the points to the cluster centroids together with the intercentroid distance; or (3) the intercentroid distance and the standard deviations of projections of points onto the intercentroid axis. The method is based on an index of disjunction W that corresponds to a degree of overlap V. The test consists of ascertaining whether the observed overlap is significantly less than an expected overlap, by determining whether a t statistic derived from the observed overlap exceeds a noncentral t value for the expected degree of overlap.