Ratcliffe and colleagues (2011) conducted a randomized controlled trial to assess the impact of the Philadelphia Foot Patrol strategy in reducing crime during the summer, when violent crime levels in Philadelphia, PA, tended to peak. The experiment was intended to isolate the specific impact of foot patrol as a crime-reduction strategy in order to examine its feasibility to create meaningful reductions in violent crime. The evaluation took place during peak summer months in 2009.
The study began with the identification of hot spots using data from the incident (INCT) database of the Philadelphia Police Department for 2006, 2007, and 2008. High-crime areas identified by the database were then assigned to spatial units, producing 120 total locations for inclusion in the experiment. A randomized block design was then used to separate the locations into two groups; 60 were assigned as control locations, and 60 were assigned to receive treatment. Treatment locations were assigned two pairs of officers to engage in intensive foot patrol policing, while control locations received no foot patrol policing. The officer pairs were assigned either a morning (10 a.m. to 6 p.m.) or an evening shift (6 p.m. to 2 a.m.) that they policed Tuesday through Saturday nights. The pairs alternated morning and evening shifts every other week. This meant that the areas were not assigned foot patrols from 2 a.m. to 10 a.m. each day, and from 2 a.m. Sunday right through to 10 a.m. Tuesday each week.
A comparison of crime frequency in both the 60 target areas and the 60 control areas before and during the operational period was used to determine the impact of the strategy. Data from the immediate 3 months before the start of the experiment was compiled for use as a measure for a pretreatment period. The operational period of the experiment was executed using two phases. Phase 1 commenced on March 31, 2009, with officers in 24 foot patrol areas. Phase 2 commenced on July 7, 2009, and totaled 36 patrolled areas. Over the first 12 weeks of each phase combined, this theoretically provided for 57,600 hours of foot patrol activity. A linear regression model was used to determine the statistical significance of changes in violent crime for both groups between pretreatment and operational levels. Analyses of displacement and diffusion effects and incident type frequency were used to supplement the analysis of crime trends for the experiment.
Reported violent crime. Data was provided by the Philadelphia Police Department from the INCT database, which assigns a Uniform Crime Report classification to all entries. Violent crime was defined as criminal homicide, all robberies (except cargo theft), and a majority of aggravated assaults. Violent crimes that were likely to take place indoors and were not expected to be prevented by foot patrol, such as rape, were excluded from the analysis.
Displacement and diffusion effects. The total net effect of the operation was calculated to assess the possibility that the strategy displaced violent crime to nearby areas, or that violent crime reductions were diffused to nearby areas. This was examined using an analysis of effects in nearby buffer areas, which encompassed the area of about 1,000 feet around target sites. Effects were calculated by examining the ratio of the crime reduction in the target areas after factoring in the general change in the control areas and then taking into consideration any displacement or diffusion to the buffer area.
Incident type frequency. The frequency of various minor criminal incidents was measured to examine the impact of increased proactive police activity after implementation of the foot patrol strategy. Such incidents were chosen as measures because of the likelihood that they would be handled by a patrol officer, as opposed to a call for service. A pedestrian stop was recorded when an officer conducted a field interview, a stop-and-frisk, or a search of suspect in the street. A vehicle stop was recorded when an officer pulled over an occupant in a vehicle. Disturbances were recorded for incidents that involved police disruption of rowdy behavior. Incidents involving narcotics were also recorded. Disorder incidents included public order offenses such as prostitution, public drunkenness, loitering, or violation of city ordinances. Arrests were also recorded, to provide possible indications of proactive police work.
Sorg and colleagues (2013) conducted a randomized experiment to examine the longitudinal deterrent effects of foot patrols in violent crime hot spots. In 2009, 240 rookie police officers were assigned to 60 of Philadelphia’s violent hot spots soon after their graduation from the police academy. The officers patrolled in pairs on a day (10 a.m. to 6 p.m.) and night (6 p.m. to 2 a.m.) night shift, 5 days a week (Tuesday through Saturday). Two phases were deployed to coincide with academy graduation dates. Phase 1 deployed March 31, 2009 and ended August 31, 2009. Phase 2 deployed July 7, 2009, and ended September 28, 2009.
Researchers identified experimental areas by reviewing violent crime event data that included homicide, aggravated assault, and robbery. This information was taken from the Philadelphia Police Department’s incident database from 2006–2009. Police leadership agreed to 129 foot beats for the study. From this list, the researchers adjusted beats that were too large or overlapped with other beats, and dropped the four lowest crime beats, which resulted in a total of 120 experimental beats (60 target foot beats and 60 control foot beats).
To measure displacement, buffer zones were drawn around treatment locations, which measured displacement. Buffer zones averaged 2.8 miles of street, and could not overlap with the experimental areas and could not cross physical barriers. Overall, 55 buffer zones were created across10 foot patrol areas.
To assess the effects of the Philadelphia Foot Patrol experiment over time and at its conclusion, multilevel growth curve models were used to analyze the outcome data. Four separate modes were run: 1) model estimating treatment effects; 2) model estimating initial deterrence decay; 3) model estimating posttreatment effects; and 4) model estimating residual deterrence decay. This review focused on the treatment effects and posttreatment effects models.
The outcome of interest was violent crime counts in each of the treatment and control hot spots aggregated to 2-week time periods. The same violent crime incident categories were used to identify violent crime hot spots and the outcome variables: homicide, robbery (excluding cargo thefts), and pertinent classifications of aggravated assaults (excluding categories that foot patrols are unlikely to affect, such as assaults against police or assaults in schools). Data from the Philadelphia Police Department’s incident database were taken from approximately 1 year before Phase 1 began (April 1, 2008).