Study 1
Bland and colleagues (2021) conducted a crossover randomized controlled trial to examine the impact of Operation Rowan foot patrols on the prevalence and volume of serious violent crime in hot spots in Bedfordshire during a 90-day period from Nov. 9, 2020, to Feb. 6, 2021.
To define and select the hot spots for targeted patrol, all crimes designated by Bedfordshire Police analysts as serious youth violence were extracted and mapped by Lower-layer Super Output Area (LSOA). The monthly risk of such crimes in each hot spot was calculated on a rolling basis, given each area’s incidence of the crime in the previous 90 days. The analysis was further refined with the addition of crime harm weightings using the Cambridge Crime Harm Index and a four-tier classification of hot spots based on how frequently they would be targeted by serious youth violence over 1 year. The 21 highest-frequency hot spots were selected. The hot spots were all within urban areas but split between large-population towns and smaller towns. Each hot spot contained about 1,000 to 1,500 people and was an average of 2,000 square meters. Each hot spot was randomly assigned to two conditions (patrol or no patrol) for 90 days. The randomization was undertaken at the beginning of the experiment and given to the onsite project manager on a week-by-week basis. Supervising officers were issued their instructions for the week ahead by Friday of each week.
The unit of analysis was the location-day, with 21 locations over 90 days for a total of 1,890 units. For each day, seven hot spots were assigned to receive the foot patrol by a uniformed officer during the 10-hour late shift (2:00 p.m. to 12:00 a.m.). The remaining 14 hot spots each day were assigned to the control condition of no patrol. Data on 630 treatment units and 1,260 control units were collected during the 90-day period.
The source of daily crime data to measure outcomes in the hot spots was the Bedfordshire Police’s records management system. Only crimes that occurred in a public space were analyzed. Domestic abuse crimes, malicious communications crimes, blackmail, taking of indecent images of children, the facilitation of travel for exploitative purposes, and all crimes relating to the local prison were excluded. Crimes were categorized as follows: Violent crimes and robbery consisted of assault with and without injury, assaults on police and other emergency workers, harassment, threats to kill, both personal and commercial robberies, and other crimes against personal victims including sexual offenses, thefts, burglaries, criminal damage, and arson. Antisocial behavior (any action that lacked consideration for others but did not meet the definition of a criminal offense, such as noise disturbances or rowdy behavior) data were taken from the force command and control system. The CrimeSolutions review of this study focused on the violent crime and robbery and other crimes against personal victims outcomes.
An intention-to-treat framework was used for the analysis. Differences in the dependent variables were analyzed using two-tailed independent-sample t-tests and analysis of variance tests for analysis of consecutive days. To analyze crime displacement, 100-meter buffers were added to all 21 hot spots, and all dependent variable measurements that fell within them were extracted. No subgroup analysis was conducted.