←Back
678 – New Methods in Spatial and Spatiotemporal Modeling and Assessment
Distance Correlation as a Measure for Dependence of Distance Matrices with Complex Spatial Patterns: An Alternative to Mantel Test
Sponsor: Section on Statistics and the Environment
Keywords: Mantel test, partial Mantel test, distance correlation, partial distance correlation
C. Deniz Yenigün
Istanbul Bilgi University
Maria Rizzo
Bowling Green State University
The Mantel test is routinely used in many areas of ecology to test the independence of the elements in two distance matrices. Despite its popularity, the Mantel test has some well known disadvantages, and the fact that the test is only sensitive to linear or monotonic dependencies is often overlooked by the practitioners. In this paper we focus on resemblance matrices with nonlinear structures and point out that partial distance correlation is a powerful alternative to the Mantel statistic for measuring association of such matrices.