The Kansas City Preventive Patrol Experiment was designed to test, empirically, the validity of one of the major strategies of modern policing: routine, visible, motorized, “random” patrol. In modern democratic societies, 60% to 65% of police personnel are assigned to such patrol activities (Bayley 1994). The most important, albeit sometimes implicit, objectives of this strategy are to deter crime, arrest offenders, and reduce the fear of crime. However entrenched this strategy might have been, by the early 1970s no systematic measurement or credible evidence had been presented to demonstrate the effectiveness of such an approach. In an effort to address this evidence gap, a task force, including researchers from the Police Foundation, and patrol officers and supervisors of the Kansas City Police Department’s South Patrol Division, worked to develop and implement an experimental research design that would test the effectiveness of routine preventive patrol.
Experimental Design
Within the South Patrol Division’s twenty-four-beat area, nine beats were eliminated from consideration as unrepresentative of the city’s socioeconomic composition. The remaining fifteen-beat, thirty-two square mile experimental area encompassed a commercial-residential mixture, with a 1970 resident population of 148,395 persons, and a density of 4,542 persons per square mile. Median family income ranged from $7,320 to $15,964 (in 1970 dollars). The percentage of African American residents ranged from 1% to 78%. The fifteen beats were computer matched on the basis of recorded crime, number of calls for service, ethnic composition, median income, and transiency of population into five clusters of three comparable beats each. Within each cluster, one beat was assigned to one of three experimental conditions:
1. Reactive, in which routine preventive patrol was eliminated and officers were instructed to respond only to calls for service.
2. Proactive, in which patrol visibility was increased by two to three times its usual level, both by the assignment of additional marked patrol vehicles and the presence of units from adjacent “reactive” beats.
3. Control, in which patrol was maintained at one marked car per shift.
Implementation
Implementation of the experiment began on July 19, 1972. By mid-August, both the Police Foundation and the Kansas City Police Department recognized that several problems had arisen that fundamentally threatened the integrity of the experiment. The first problem was that the South Patrol Division had fallen to a dangerously low level of manpower for experimental purposes. To correct this problem, the department assigned additional police officers. A second problem involved violations of the experimental guidelines. Specifically, it was determined that officers assigned to “reactive” beats were not adhering to the guideline that stipulated that they not enter their beats except in response to a call for service. Additional training was provided, and administrative measures were taken, to ensure adherence to the guidelines. A third problem was boredom among officers assigned to “reactive” beats. To counter this, the guidelines were modified to allow an increased level of activity by “reactive”-assigned officers in “proactive” beats. The revised guidelines stressed adherence to the spirit of the project, rather than to unalterable rules.
On October 1, 1972, the experiment resumed. It continued successfully for twelve months, ending on September 30, 1973. Findings were produced in terms of the effect of experimental conditions on a wide variety of outcome measures, as described later.
Data Sources
The task force decided to test the possible effects of routine preventive patrol by collecting a wide variety of data from as many diverse sources as possible. Sources used included departmental data, surveys of community residents, surveys of commercial managers, surveys of persons encountered by police, a response time survey, surveys of police officers, participant observer surveys, officer activity analyses, and others. A summary of the various data sources is provided next:
• Community survey. A survey of approximately twelve hundred, randomly selected residents was conducted before the experiment began; the sample was spread throughout all fifteen experimental beats. Respondents were asked about their fear of crime, attitudes about their neighborhood, satisfaction with police service, victimization experiences, and other matters. One year later, twelve hundred respondents were again interviewed, with six hundred chosen from the original groups (producing a repeated sample), and six hundred chosen randomly (for a nonrepeated sample.)
• Commercial survey. A survey of representatives of commercial enterprises in all of the experimental beats was conducted both before and one year after the experiment began. As with the community survey, respondents were asked about their fear of crime, attitudes about the neighborhood, satisfaction with police service, and victimization experiences.
• Encounter survey. Because household surveys tend to interview relatively few persons who have experienced actual contact with the police, an additional survey was conducted of those persons in the experimental area who experienced direct encounters with police officers. The survey was conducted during a four-month period (July-October 1973) and collected information from 331 citizens who were involved in either an officer-initiated incident (vehicle check, pedestrian check, or a traffic violation) or citizen-initiated incident (one in which the citizen called for police service). Respondents were asked questions about the nature of their encounter with the police and their evaluation of the service provided.
• Participant observer transcriptions. Participant observers were hired to make observations of police activity in all fifteen experimental beats. These observations served two purposes: (1) to monitor adherence to experimental guidelines and (2) to observe and describe interactions between officers and citizens.
• Recorded crime data. Monthly totals for recorded crime, by type, for each experimental beat over the October 1968 through September 1972 (pre-experimental) period and over the October 1972 through September 1973 (experimental) period were extracted from departmental records.
• Traffic data. Monthly totals concerning two categories of traffic accidents (noninjury and injury/fatality) were collected for each experimental beat for the October 1970 through September 1972 (pre-experimental) period and for the October 1972 through September 1973 (experimental) period.
• Arrest data. Monthly arrest data were collected, by crime type, for each experimental area for the three-year period prior to the experiment and during the experimental year.
• Response time data. Computerized records concerning the time required for police to respond to calls for service were not available, making comparisons between pre-experimental and experimental periods impossible. To compensate, police response time in the experimental area was measured by a response time survey completed by the participant observers and citizens who had called the police for service. In measuring the time taken by the police in responding to calls, emphasis was placed on field response time (that is, the amount of time occurring between the time a police unit on the street received a call from the dispatcher and the time when that unit contacted the citizen involved). In measuring citizen satisfaction with response time, the entire range of time required for the police to respond to a call was considered, including time spent talking with the police operator, time taken by a police dispatcher, and field response time.
Findings
Overall, the experiment found that there were no significant differences among the three experimental conditions concerning recorded crime, reported victimization, citizen fear of crime, citizen satisfaction with police service, police response time, arrests, traffic accidents, or other major indicators. In particular:
• As revealed in the victimization surveys, there were no significant differences among or between the experimental conditions concerning recorded residence and nonresidence burglaries, auto thefts, larcenies, robberies, or vandalism, crimes traditionally considered to be deterrable through preventive patrol.
• In terms of rates of reporting crimes to the police, few differences and no consistent patterns of differences occurred across experimental conditions.
• In terms of recorded crime, only one set of differences across experimental conditions was found, and this was judged likely to have been a random occurrence.
• Few significant differences and no consistent pattern of differences occurred across experimental conditions in terms of citizen attitudes concerning police services.
• Citizen fear of crime, overall, was not affected by experimental conditions.
• There were few differences and no consistent pattern of differences across experimental conditions in the number and types of anticrime protective measures used by citizens.
• In general, the attitudes of businessmen toward crime and police services were not affected by experimental conditions.
• Experimental conditions did not appear to affect citizen satisfaction with the police as a result of their encounters with police officers.
• Experimental conditions had no significant effect on either police response time or citizen satisfaction with police response time.
• No significant differences in traffic accidents or injuries were found across experimental conditions.
Conclusion
Overall, the results of the Kansas City Preventive Patrol Experiment challenged the conventional wisdom that “random” patrol had a significant effect on crime and fear of crime. The study was not without its detractors. Some critics argued that Kansas City was not representative of the vast range urban environments in the nation. Others contended that the study did not contain enough beats, that the samples of citizens surveyed were too small, and that the statistical power of the tests was too limited. Nevertheless, the study, by calling into question basic assumptions of police strategy, led to a wave of empirical research that has caused a wide-ranging reappraisal of police strategy and tactics.
