Smallest is Better? The Spatial Distribution of Arson and the Modifiable Areal Unit Problem

Abstract

The aim of this study is to explore how the zonation and scale problems of the modifiable areal unit problem (MAUP) impact on the proportion of variance associated with surrounding areas in relation to micro-place levels of arson. MAUP is related to how geographical areas are constructed, with zonation related to how boundaries are drawn, and scale related to the size of areas. Arson point data from 2007 to 2011 are analyzed by means of hierarchical linear modeling in order to compute intra-class correlations (ICCs), the share of variance associated with the higher order geographical units, for geographical units of three different sizes and with three degrees of randomness. Real, administrative, geographical units of two sizes, with mean size of 1.2 and 0.4 square kilometers respectively, are compared both to semi-random and fully-random artificial geographical units of the same size, and to smaller types of units of 0.17 square kilometer size. The analysis shows that there is little difference between large and medium-sized geographical units, but there is a significant increase in the ICC at the smallest geographical scale. To understand the geography of arson this suggests that the smallest types of units are of the greatest importance. As regards the problem of zonation, the results show that more randomness of boundary placement is associated with lower ICCs. A key implication of these findings is that community preventive efforts may best be targeted at very small communities such as street blocks rather than larger neighborhoods.