Edge Correction for Exact Tests on Nearest-Neighbor Contingency Tables for Testing Spatial Segregation

• Activity Number: 112
• Type: Contributed
• Date/Time: Monday, July 30, 2007 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Elvan Ceyhan*+
• Companies: Koc University
• Keywords: Association ; complete spatial randomness ; edge effects ; exact inference ; buffer zone correction ; toroidal correction

Abstract

Nearest neighbor contingency tables (NNCTs) are used in testing spatial patterns of segregation or association in various fields. For general contingency tables, exact tests are conservative in testing independence, but we have shown that the most conservative versions of the exact tests are appropriate for NNCTs for small samples (i.e., total number of points < 100). The null patterns are either complete spatial randomness (CSR) or random labeling (RL). The CSR model assumes that the study region is unbounded for the analyzed pattern, which is not the case in practice. In this article, we discuss the edge (or boundary) effects on various exact tests for NNCTs under the CSR case. We demonstrate that buffer zone edge correction can severely affect the results of these tests, while the toroidal correction has a mild effect on them. The right-sided exact test is robust to the edge effects.