Optimizing the geographic deployment of hot spot patrols with license plate readers

Abstract

To determine optimal methods of patrol with license plate readers (LPRs) for maximizing detection and apprehension. Data from two randomized experiments were jointly analyzed. In the first, a four-officer squad conducted short daily operations to detect stolen and other vehicles of interest at randomly selected road segments (averaging 0.6 miles) at varying times of day for 2 weeks per location. Based on random assignment, the unit operated with LPRs on some routes and conducted extensive manual checks of license plates on others. In the second experiment, the squad conducted similar operations focused on larger hot spot areas (averaging 1 mile2) using the same research design. The first experiment emphasized longer surveillance on primary hot routes, while the second emphasized shorter visits to the main hot routes and more roving surveillance around these routes. Officer log data were used to examine the likelihood that the patrols resulted in detection of wanted vehicles, arrests, or recoveries of stolen vehicles. Chi-square tests and multivariate logistic regressions were used to examine LPR performance across experiments. LPRs increased the likelihood of all outcomes (which were generally rare) in both experiments, though only the effects on detections were statistically significant. Differences across the experiments were not statistically significant, but LPR performance was generally better during the second experiment, particularly with regard to how much LPR use improved performance over manual plate checks. Results suggest LPR patrols emphasizing roaming operations within larger hot spot areas may be most optimal, but the effects of LPR patrols may not be substantial without large-scale deployment.