Study
Ariel and Partridge (2017) conducted a randomized controlled trial to measure the effects of hot spots policing at bus stops in London on crime, measured through victim-generated crime reports and bus driver incident reports. The experiment took place in London, the UK’s most populous urban metropolitan city, with 8.63 million residents in nearly 607 square miles. Overall, per 1,000 residents, London had 8.3 violence with injury offenses, 1.7 sexual assaults, 2.6 robberies, and 8.8 residential burglaries in 2014–2015. In terms of the bus network, London hosts one of the largest systems in the world, with over 9,000 buses, 675 bus routes, and 19,000 bus stops. The bus network attracts over two billion commuter trips per year. The MPS recorded 7.2 crimes per million passenger journeys during 2014–2015.
The study authors identified the hottest bus stops (i.e., bus stops that had a disproportionately higher count of driver incident reports relative to the other 19,000 bus stops in London) in the Greater London area, and randomly assigned them to the treatment or control conditions. None of the pretreatment between-group differences were statistically significant, therefore suggesting that the two groups were approximately equivalent prior to the launch of the intervention. The study authors drew a concentric series of buffers to test for the treatment effects in the area surrounding the bus stop at less than 50 meters, 50 to 100 meters, and 100 to 150 meters. Buffer zones were drawn to detect any possible displacement effects or diffusion of benefits and to ensure that no two hot spots and their surrounding areas overlapped, which could cause a treatment effect spillover.
The experiment involved 102 bus stops. Metropolitan Police Service officers were not instructed to patrol beyond the bus stop vicinity and were tracked using GPS. Outcomes were measured in terms of victim-generated crime reports to the police and bus driver incident reports (DIRs), within targeted and catchment areas. Victim-generated crime reports capture any instance in which an individual called on the police to report a crime, whereas DIRs are instances of criminal damage, fare evasion, and passenger disturbance on London’s bus network. At any given moment during the duration of the experiment (6 months), there were about 32 officers conducting patrols during hot hours. Each patrol unit had ownership of about two to four hot spots, depending on the travel distance between the bus stops.
Changes in the two outcome types, DIRs and victim-generated crime reports, were compared between the 6-month period before the trial and after the beginning of the trial, and then were compared with the differences between the treatment and control conditions. The data were also organized based on the time of the intervention (Monday–Friday, between 12:00 and 6:00 p.m.), and outside these hours. No subgroup analyses were conducted.
