Determinants of Crime Incidence in Korea: A Mixed GWR Approach *

Abstract: It is true that crime is not exclusively an urban phenomenon, however, most crime and violence occur in cities and urbanized area. The present study investigated the determinants of crime incidence in Seoul city, focusing particularly on spatial and environmental effects on crime victimization. Since crime incidence is in general closely related to the spatial characteristics of a city, we adopted a geographically weighted regression model to incorporate these area characteristics in our statistical model. We also applied mixed GWR model in order to identify local and global effects of independent variables on crime incidence. The results of the model proved to be superior to those from standard spatial econometrics models (SAR, SEM, SAC) with respect to model fits and stability of the parameters estimated. We found that spatial dependency and spatial heterogeneity is particularly important in examining the crime incidence in Seoul. That is, crime in Seoul is primarily spatial; an area is more likely to be victimized by somewhere of the same city than somewhere of another area. Districts with frequent mobility and entertainment spot are more prone to violence and less safe than other districts. Also included among the determinants that contribute to crime incidence are density, size, proportion of youth age group, etc. Zoning is a particularly important factor to determine crime incidenc