A network of neighborhoods: Predicting crime trips in a large Canadian city

• In Event: Crime and Place: Shifting Crime Risks By Night and Day

Abstract

Criminal target choice has been described as a multistaged process: an offender first selects a suitable area and then chooses a specific target. While many authors showed interest in specific target selection, often with crime prevention goals in mind, relatively few studies of the factors explaining area selection are available. In particular, previous studies have been unable to distinguish crime generators/visit detractors (factors appealing to anyone) from crime attractors/offender detractors (factors especially appealing for offenders). This presentation aims to analyze relations between 506 census tracts (CT), from offenders' homes to crime locations. In the analysis, a crime trip occurs when an offender commits a crime outside his home CT. Two types of predictors are considered: relational measures (relating to dyads of tracts) and characteristics of the destination tract. Multilevel negative binomial regression is used to analyze more than 15,000 trips for violent and property crime. Results indicate that various factors are significantly related to the number of crime trips. Following a recent categorization, these factors relate to differences in reward (residential and visiting population size, presence of schools or bars), differences in effort (distance between tracts, major roads linking both tracts) and differences in risk (social disorganization).

Author

• Rémi Boivin, Université de Montréal