Abstract
The problem of comparing K independent groups is a common task in statistics. While the tests for univariate data are already well established, the choice of an appropriate test for multivariate data is much more delicate.
Moreover the assumptions of the traditional multivariate analogues of the univariate tests are usually considered too strict for most ecological multivariate data sets. Analysis of Similarities (ANOSIM) is a distance-based test that is often considered to be a nonparametric analogue of a standard F -test that is suitable for multivariate date.
This paper compares the two most common versions of ANOSIM test that have appeared in literature.