Researchers have derived mutual information from well-known general approaches to constructing segregation indices, establishing a mathematical framework for measuring segregation across demographic and spatial scales. In peer-reviewed research published in EPJ Data Science, Rohit Sahasrabuddhe and Renaud Lambiotte emphasize that because segregation operates across multiple scales simultaneously, quantifying it requires careful and principled measurement grounded in clear theoretical principles.
The authors derived the metric from general index-construction approaches and interpreted the mathematical properties that are unique to mutual information. According to the study, these unique properties relate mainly to how the index behaves when demographic identities and spatial units are coarse-grained. This mathematical analysis clarifies the precise theoretical behavior of the measure as researchers aggregate individual spatial units or group demographic categories into broader classifications.
A fundamental property identified by the researchers is that mutual information decomposes into the distinct contributions of regions and supergroups. The team reported that this multi-scale decomposition reveals analytical insights that would otherwise remain inaccessible under alternative measurement techniques. By separating these multi-scale components, the framework allows researchers to examine how broader regional distributions and demographic supergroups each contribute to overall observed segregation.
To evaluate the methodology, the authors demonstrated the approach through an exploratory case study of residential ethnic segregation in England and Wales. The work aims to help researchers identify the specific conditions under which mutual information is the only suitable segregation index for a given study, while providing accompanying open-source software designed to support practical multi-scale segregation analysis.
Subject of Research: Multi-scale segregation measurement using mutual information
Article Title: From first principles to multi-scale decomposition: mutual information as a segregation index
Article References: Sahasrabuddhe, R., & Lambiotte, R. (2026). From first principles to multi-scale decomposition: mutual information as a segregation index. EPJ Data Science. https://doi.org/10.1140/epjds/s13688-026-00716-4
Image Credits: AI Generated
DOI: 10.1140/epjds/s13688-026-00716-4
Keywords: mutual information, segregation index, multi-scale decomposition, residential segregation, ethnic segregation, spatial scales, demographic identities, coarse-graining, supergroups, regional contributions, England and Wales, EPJ Data Science, Rohit Sahasrabuddhe, Renaud Lambiotte, open-source software, spatial analysis, first principles
News Source: Denise Maddox. (November 20, 2026). Mutual Information Provides Multi-Scale Index for Measuring Demographic and Spatial Segregation. Scienmag.



