The effect of skewed distributions on matrix permutation tests

This paper examines the effect of skewed data distributions on matrix permutation tests of association. An empirical example is presented in which incorrect results are obtained using the Mantel statistic. Alternative procedures are considered, and two different data transformations are suggested which provide efficient and effective solutions. Methods for analysing related correlations as a means of evaluating the relative goodness‐of‐fit of alternative representations of a set of data are employed.