Todd Norwood

Crime researchers have noted spatial patterns in crimes for approximately 150 years. While there is recognition of the spatial variations in crimes, agreement on the causes of the variation has yet to be achieved. Nonetheless, research shows common linkages of crime types to socio-economics and demographics. Given the research, these linkages are considered first-order or deterministic effects. Stochastic or second-order effects occur from spatial dependence in crime rates. Property and drug crime data were acquired for the year 2003 in the Peel Regional Police jurisdiction for analysis at the census tract level of spatial detail. Global measures of spatial association confirmed significant spatial auto correlation within the crime types studied. To determine the nature of this spatial pattern in each crime type, local measures of spatial association are examined to explore the variations, followed by join-count ~tatistical analyses using a new S-Plus join-count algorithm that can handle first-order heterogeneity. Comparison to Cliff and Ord's (1981) original join-count statistic, which assumes first-order homogeneity, is made to separate first-order effects in crime rates from second-order effects in crime rates. The findings show that once the relationship of crime rates to socio-economic and demographic characteristics of areas are accounted for, there is no significant spatial auto correlation in the observed crime rates.