Exploratory spatial analysis of crimes against property in Turkey

Turkey, a rapidly developing country, is a junction point between Asia and Europe in terms of its social and economic structure. Turkey is both the most advanced economy in the Turkish-speaking world and the largest economy in the Muslim world. In the last few years, with the development of Turkey's economic and social structure, the level of criminality in the country has attracted attention. Consequently, we aimed to explore potential spatial associations of crimes against property rates across the 81 provinces of Turkey (NUTS3) from 1997 to 2009. Geographical information systems and explorative methods of spatial data analysis were employed in the analyses of crime rates. Since crime is a phenomenon that arises from the interaction between social, economic, psychological circumstances and, especially, geographical factors, this study attempts to rectify the possible deficiencies of traditional statistical analyses of geography. Because of the very different population sizes in each province, comparing the volume of crimes makes a major impact on the stability of the crime rates; therefore in this study an empirical Bayes smoothing method was used to interpret the crime rates correctly. Global spatial autocorrelation indices were used to test the spatial dependence of the distribution of the crime rates. Besides the excess risk rates, local spatial autocorrelation methods were used to detect and interpret the clustering of crime rates. In order to model the crime rates, a set of socioeconomic parameters (migration rates by province, gross national product according to purchasing power parity by province, registered number of touristic facilities and number of rooms by province, electricity consumption statistics, provincial unemployment rates and urbanisation index values were handled with geographically weighted regression. According to the results, criminal activities were non-random in terms of time and space. Property crime, except for smuggling, is densely clustered in the west and south-west of Turkey. The present study demonstrates the utility of spatial analysis and geographically weighted regression to detect some important geographical dimensions and crucial geographical aspects of property crimes in Turkey.