ABSTRACT This paper develops a spatial formulation of the Jensen–Shannon divergence to measure multigroup compositional dissimilarity between each municipality and its surrounding context. Building on information theory, the index compares local group distributions with neighbourhood‐lagged compositions defined through two distance‐based systems: a Gaussian kernel and a Max–Min sequence of connectivity thresholds. The measure is symmetric, bounded and decomposable into group‐specific contributions, allowing identification of both the magnitude of local divergence and the groups responsible for it. Using data on the 10 largest foreign citizenships across 7887 Italian municipalities in 2024, the analysis reveals that significant local divergence is concentrated in a limited subset of municipalities and exhibits a clear multiscalar structure. South‐Asian and selected North‐African groups consistently account for the highest contributions, while many Eastern‐European groups show patterns of local alignment with surrounding areas. The Gaussian kernel produces smooth spatial gradients, whereas the Max–Min system highlights breakpoints in connectivity and reveals stepwise evolutions in local mismatch. A Monte Carlo spatial randomisation test confirms that the observed patterns deviate systematically from space‐composition independence. Overall, the spatial Jensen–Shannon framework provides a flexible diagnostic tool for identifying enclave‐like configurations and assessing multiscalar socio‐spatial contrasts with potential relevance for integration policies.
Massimo Mucciardi (Fri,) studied this question.