Road-Network Versus Isotropic Distance in the Geographic Profiling of Serial Homicide: A Multinational Comparison

Purpose: Urban road networks impose structural constraints on movement that isotropic distance metrics fail to capture. This study evaluates whether road-network topology, via OpenStreetMap shortest-path algorithms, produces spatial priority surfaces differing from Euclidean and Manhattan distance in the geographic profiling of serial homicide. Purpose: Urban road networks impose structural constraints on movement that isotropic distance metrics fail to capture. This study evaluates whether road-network topology, via OpenStreetMap shortest-path algorithms, produces spatial priority surfaces differing from Euclidean and Manhattan distance in the geographic profiling of serial homicide. Methods: A verified multinational sample of 29 serial homicide cases (277 offences) from the UK, continental Europe, and North America (1962-2012) was analyzed using a repeated-measures design comparing Search Cost% (SC%) and Error Distance (ED). Results: No significant differences emerged between metrics in SC% (Friedman χ²(2) = 2.396, p = .302) or ED (χ²(2) = 3.493, p = .174). Network did not differ from Euclidean in ED (V = 77, p = .093) or SC% (V = 75, p = .056). Neither OSM coverage (ρ = .043, p = .848) nor geographic dispersion (β = -222.6 m/km, p = .084) predicted Network relative performance. Network required approximately 346 times an assumed isotropic benchmark, with the most demanding case exceeding four hours. Implications: Road-network and isotropic metrics showed comparable search efficiency, and the substantially costlier Network metric showed no statistically reliable precision advantage in this sample, cautioning against its routine use in operational geographic profiling of serial homicide