Modifications of the Syrjala Test for Testing Spatial Distribution Differences Between Two Populations

• Activity Number: 412 - Data Science and Machine Learning Topics
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Eric McKinney* and Juergen Symanzik
• Companies: Utah State University and Utah State University
• Keywords: Continuous Bivariate Data; Permutation Test; Environmental Data; Eye-Tracking Data

Abstract

Determining whether two spatial distributions are statistically equivalent is the goal of the Syrjala test. When using continuous bivariate data, we show that the original Syrjala test produces different results depending on the data aggregation steps. In this article, we propose modifications to the previous version of the Syrjala test and make comparisons via simulations and an application. Simulation results indicate greater power and a more appropriate type one error rate for our modified Syrjala test. Furthermore, our new approach can be used for environmental data (for which the Syrjala test was originally developed), but also for data that originates from an eye-tracking study conducted at Utah State University.