Spatially varying profiles of bicycle robbery targets at the street-level

The crime literature widely recognizes the influence of people flow in the creation of crime opportunities. While efforts have been made to account for human mobility in crime prediction, previous studies are restricted to aggregated measures of people displacements, overlooking both incidents that occurred during trips and the socioeconomic profile of potential targets. This paper analyzes bike robberies targeting cyclists during their trips in São Paulo (Brazil) from a street-level perspective, using origin and destination survey data to provide socioeconomic information on pass-through cyclist flows, such as income, age and sex. Hotspots are identified using network kernel density estimation, and a geographically weighted logistic regression explores the spatially varying impact of 15 urban features on whether or not a street segment is likely to be within the influence range of a crime occurrence. Evidence showed that not only does the cyclist profile have an influence on bike robberies, but also that the effect of all risk factors varies spatially. Bike robbery likelihood varied from −2% to 5% given a 100-unit increase in the cyclist volume. Increasing the cyclist age in 1 year would cause a 4% increase in the robbery probability in some streets, whereas it would reduce by 9% in other. In addition to identifying bike robbery risk factors, the proposed approach indicated the streets where a higher variation in the risk factors leads to a higher increase in crime likelihood, providing valuable information to the cyclist community and police patrolling toward improving security levels during bike trips.