NCJ Number
209645
Journal
Journal of Quantitative Criminology Volume: 21 Issue: 1 Dated: March 2005 Pages: 1-26
Date Published
2005
Length
26 pages
Annotation
This study tested the relationship between the complexity and accuracy of geographic profiling strategies.
Abstract
Decisionmaking research has often relied on the assumption that more complex decisionmaking strategies lead to more accurate predictions. One area where the complexity-accuracy assumption is being challenged is in the area of geographic profiling. This study tested the following two hypotheses: complex geographic profiling strategies would result in more accurate prediction compared to less complex strategies and complex geographic profiling strategies would be more accurate than less complex strategies on more complex tasks. Data for this study were crime site and home locations of 16 serial residential burglars who committed 10 or more burglaries in a semi-rural county of the United Kingdom between 1997 and 1999. The study tested 11 geographic profiling strategies: 6 were spatial distribution strategies and 5 were probability distance strategies. The computational complexity approach was used to classify these strategies. Applying these strategies to the data, the study found that all probability distribution strategies were substantially more complex than all spatial distribution strategies, but were not necessarily more accurate. The results also showed that more complex strategies were not significantly more accurate than less complex strategies when the geographic profiling task was more complex. Rather, accuracy tended to increase with increasing task complexity across all strategies. Implications for police policies and procedures are discussed. References, tables