DO CRIME HOT SPOTS MOVE? EXPLORING THE EFFECTS OF THE MODIFIABLE AREAL UNIT PROBLEM AND MODIFIABLE TEMPORAL UNIT PROBLEM ON CRIME HOT SPOT STABILITY

The aim of this research was to explore the apparent disconnect between academicians and practitioners with respect to the stability of crime hot spots. Calls-for-service data from Arlington, TX, and Charlotte-Mecklenburg, NC, for the years 2005-2010 were used to test the effects of grid cell size and period duration length on the effectiveness, efficiency, and measured hot spot stability of the random walk model. This research relied on the Prediction Accuracy Index (PAI) for measurement of the effectiveness of the random walk method; this research also introduced and relied on the Prediction Efficiency Index (PEI) for measurement of the efficiency of the random walk method. In both cases, the percentage of grid cells that were classified as a hot spot in two adjacent time intervals T1 and T2 was used to measure the hot spot stability of the random walk method. The results of this research indicated the general relationship that the PAI value decreases with increasing grid cell size. Charlotte-Mecklenburg showed a general positive relationship between grid cell size and PEI, whereas no such general relationship was found for Arlington. As the period duration was increased, no general finding was observed with regard to its effect on the PAI. Conversely, a negative relationship was found in both cities with respect to the effects of period duration on PEI. In terms of hot spot stability, four findings were ultimately identified: (1) as period duration increased, the percentage of dual hot spots increased; (2) different crime types were found to have different hot spot stabilities; (3) the hot spot break point technique affected the percentage of dual hot spots relative to the number of hot spots that the break point classifies; and (4) grid cell size did not affect the measured stability in a similar manner among the two cities. While not generalizable, this research provides a framework for police departments to conduct similar analyses to provide a transparent, effective, and efficient means of defining place and time with regard to crime hot spots.