Spatial Proximity as a Measure of Activity-Space Segregation: Inferential Statistics and Sample Size Requirements

John Palmer, Universitat Autònoma de Barcelona

This article analyzes the performance of White's index of spatial proximity as a measure of activity-space segregation using sampled data. It relies on data collected from volunteers with a mobile phone application and data generated from computer simulations to construct empirical sampling distributions of the index's estimator at a range of sample sizes and to test a bootstrap approach for evaluating the uncertainty of any individual estimate. The empirical distributions suggest that this index may be estimated with essentially no bias using coarse trajectory data with sample sizes as low as several hundred people. In addition, the uncertainty of individual estimates can be approximated well using bootstrap methods. The article concludes that these statistical properties, combined with the index's flexibility for measuring segregation at a range of scales by modifying its distance function, make it a valuable tool for research on activity-space segregation.