Detecting spatial movement of intra-region crime patterns over time

Many of the traditional measures of the degree to which crime patterns change over space and time have limitations. In particular most are unable to determine any change in spatial crime pattern within an areal unit. Usually studies measure the change in crime levels in contiguous areas (expressed as discrete sub-divisions of a study area), but this can become problematic due to difficulties such as the Modi. able Areal Unit Problem (MAUP). This paper describes a technique developed to allow researchers to examine intra-study region changes in crime patterns between two time periods without the need to aggregate crime counts to within-city areal boundaries. The method presented uses a random point nearest neighbor test combined with a Monte Carlo simulation. The process resolves problems of patterning and the MAUP that are common with a number of spatial displacement and pattern movement studies. This technique is demonstrated with example data from a city-wide police burglary crackdown in the Australian capital.