Skip main navigation
We use cookies to give you a better experience, if that’s ok you can close this message and carry on browsing. For more info read our cookies policy.
We use cookies to give you a better experience. Carry on browsing if you're happy with this, or read our cookies policy for more information.

A tale of two experiments

The rational choice model (as well as common wisdom) tells us that frequent preventive police patrol should deter potential criminals from offending by increasing their perceived likelihood of apprehension and punishment.

However, as we saw from earlier weeks, it is not easy to actually obtain the causal effect of preventive patrol on crime based on non-experimental data. The main reason is that neighborhoods that receive more preventive patrols may be systematically different from neighborhoods that receive little.

But what if the intensity of police patrol were to be determined randomly? In this case, it becomes a lot easier to recover the causal effect of police patrol on crime. We will randomly assign neighborhoods into two groups, one group receiving a lot more police patrol than the other, and compare crime rates between highly patrolled neighborhoods and lightly patrolled neighborhoods. The difference can be viewed as the causal effect of additional police patrol on crime.

So far we dismissed such experiments as unrealistic and infeasible. However, the value of a well-designed randomized experiment has been well-known for a long time, and several police departments courageously undertook such policing experiments. Their findings tell us a great deal about the causal relationship between police and crime. Below, we will examine two policing experiments which had a major influence on police policies over the last few decades.

Kansas City Experiment

The first experiment was done in 1972 by the Kansas City Police Department. The Department wanted to find out whether their routine preventive patrol was really effective in reducing crime, and ran the following experiment. Fifteen police beats in the city were divided into five groups of three beats. Each group was chosen based on observable demographic, economic, and criminal characteristics so that the three police beats within each group were highly comparable to each other. Then, within each group, the Department randomly assigned one police beat as “control”, another as “reactive” and the third as “proactive”.

In the police beats assigned to the control group, police performed their routine patrol as usual. For the reactive group, police stopped preventive patrol altogether and only responded to calls for service. For the beats in the proactive group, the department significantly increased police visibility by sending out a lot more preventive patrol than usual. The department closely monitored weekly crime data so that they would stop the experiment in case of a noticeable crime increase from the reactive group. Fortunately this problem did not happen during the 12 month span of the experiment.

Surprisingly, the comparison between crime rates from the three groups revealed that additional police patrol did little to reduce crime, as crime rates from the three groups were very much similar. The finding led many researchers and policymakers to question the effectiveness of preventive police patrol for many years. For example, Carl Klockars (1983) argued that “it makes about as much sense to have police patrol routinely in cars to fight crime as it does to have firemen patrol routinely in fire trucks to fight fire.”

But before we conclude that police patrol really has no effect, let’s think a bit more about the experiment and its findings. Why do you think that the increased police presence did not lead to a noticeable drop in crime? Is it because potential criminals are just too impulsive and reckless, and do not mind committing theft and robbery in front of a police car? A more realistic explanation is that the “dose” of the treatment may not have been large enough. Typical police beats cover a large area (usually a few kilometers squared) and thousands of residents, and it would take unusually large increases in the police patrol effort before would-be offenders in the police beat realize any noticeable change in the level of police presence.

In either case, findings from the Kansas City experiment suggest that increased effort and investment on preventive police patrol may not be an effective crime-fighting strategy. But is this really the case?

Minneapolis Experiment

Recent advances in computing and mapping technology enabled researchers to easily visualize and analyze spatial patterns of crime. The spatial analysis consistently shows that crime is heavily concentrated in a small segment of “hotspots”. Just like how a small number of high-risk individuals are responsible for a disproportionately large share of crimes, few addresses and street corners account for a bulk of crimes in a city.

For example, criminologists Lawrence Sherman, Patrick Gartin, and Michael Buerger found that more than half of all police calls in Minneapolis originated from just 3 percent of street addresses, and that no police car was ever dispatched to about 40 percent of the street addresses and street corners during a calendar year. What would happen if police spends more patrol hours in these crime hotspots? Would this lead to a substantial reduction in crimes? Clearly, it should be a lot easier to significantly increase police presence in a few street corners and addresses than the entire police beat.

In order to test whether this hotspot-based intervention would yield a different result from the earlier beat-based experiment, criminologists Lawrence Sherman and David Weisburd designed and implemented the following randomized experiment in Minneapolis. With an endorsement and cooperation from the city government, Sherman and Weisburd identified 110 crime hotspots in the city and randomly assigned them into two groups. Hotspots in the control group received the usual level of police patrol, and those in the experimental group received about twice as many patrol hours as the control group.

After the experiment was concluded, Sherman and Weisburd compared crime rates between the two groups, and found that hotspots in the experimental groups had roughly 10 percent fewer calls for service. Their finding led to renewed academic and policy discussions on the use of preventive patrol as a potential crime-fighting tool. Similar experiments from Boston, Baltimore, Seattle, and Vancouver also support that hotspot-based preventive patrol can lead to a significant reduction in crime.


References:

  • Kelling, George L., Tony Pate, Duane Dieckman, and Charles. E. Brown. 1974. The Kansas City Preventive Patrol Experiment: A Summary Report. Washington, DC: Police Foundation
  • Klockars, Carl. Thinking about Police. New York: McGraw-Hill, 1991.
  • Sherman, Lawrence W., Patrick R. Gartin, and Michael E. Buerger. “Hot Spots of Predatory Crime: Routine Activities and the Criminology of Place.” Criminology 27.1 (1989): 27-56.
  • Sherman, Lawrence W., and David Weisburd. “General Deterrent Effects of Police Patrol in Crime “Hot Spots”: A Randomized, Controlled Trial.” Justice Quarterly 12.4 (1995): 625-648.

Share this article:

This article is from the free online course:

Economics of Crime

Hanyang University

Get a taste of this course

Find out what this course is like by previewing some of the course steps before you join:

Contact FutureLearn for Support