Autocorrélation spatiale et déplacement de la criminalité

Abstract

This research examines spatial autocorrelation and crime displacement as facts and concepts. The objective of the study concerns both the methodological theory and the practice of applied research in criminology. The two expressions used in the title can be considered a fragmentation of the concept of turbulence. The authors examine the influence of the presence of spatial autocorrelation in the data on tests of the significance of statistics. They attempt to detect it in a part of central Montreal by examining the distribution and evolution of the two types of crime: armed robbery and burglary. Three methods are used: the known coefficients and correlograms, spatial grouping with forces contiguity, and trend surface analysis. The last two methods lead to convincing results. On the other hand, the displacement could not be observed during the time studied. In conclusion, they consider the implications for the methodological practices and usual research strategies in criminology.