From first principles to multi-scale decomposition: mutual information as a segregation index
Open Access
- 20 November 2026
- journal article
- Published by Springer Nature in EPJ Data Science
Abstract
Segregation operates across demographic and spatial scales and requires careful and principled measurement. In this work, we derive mutual information from well-known general approaches to constructing segregation indices and interpret the mathematical properties that are unique to it. The properties relate mainly to its behaviour when demographic identities and spatial units are coarse-grained. Mutual information decomposes into the contributions of regions and supergroups, which reveals insights that would be inaccessible otherwise. We demonstrate this with an exploratory case study of residential ethnic segregation in England and Wales. Overall, we help researchers identify when mutual information is the only suitable segregation index and provide open source software to support multi-scale analysis.Funding Information
- Mathematical Institute, University of Oxford
- Engineering and Physical Sciences Research Council (EP/V013068/1)