What You'll Learn
The Concept
Police organizations necessarily operate under uncertainty. Officers and administrators must frequently make decisions before complete information is available. Experience therefore plays an indispensable role in policing.
Experience, however, has limitations. A police department may implement a strategy, observe that crime subsequently declined, and conclude that its strategy caused the reduction. Yet crime could have decreased because of seasonal variation, demographic changes, changes in offending patterns, economic conditions, unrelated arrests, regression toward the mean, or numerous other factors.
The central problem is causal inference. Researchers are not merely interested in whether two events occurred together — they seek to determine whether an intervention actually produced an observed change. This distinction separates correlation from causation.
In a randomized controlled experiment, eligible units are randomly assigned to treatment and comparison conditions. Randomization attempts to reduce systematic differences between groups so researchers can more confidently attribute subsequent differences to the intervention. Police experiments involving patrol, hot spots, domestic violence responses, and body-worn cameras have contributed significantly to modern policing research.
Random assignment is frequently impossible or inappropriate in criminal justice. Researchers may instead use matched comparison groups, interrupted time series, difference-in-differences analyses, regression discontinuity designs, or other quasi-experimental methods. These approaches attempt to estimate what would probably have occurred without the intervention — what researchers refer to as the counterfactual.
One study should rarely settle an important policy question. A systematic review identifies and evaluates multiple studies addressing a specific research question. A meta-analysis may then statistically combine results to estimate an overall effect. This is particularly important because a program that succeeds in one city may fail in another. Implementation matters. Leadership matters. Officer behavior matters. Community characteristics matter.
Students must also distinguish statistical significance from operational importance. A large study may detect a very small difference that is statistically unlikely to result from chance. That does not necessarily mean the difference is meaningful enough to justify additional personnel, overtime, technology expenditures, community disruption, privacy costs, training burdens, or opportunity costs. Police executives must ask not only 'Did it work?' but 'How much did it work, under what circumstances, at what cost, and compared with what alternative?'
Research should inform decisions, not mechanically dictate them. A study conducted in a densely populated metropolitan area may not translate directly to a rural jurisdiction. A statistically effective strategy may require modification because of constitutional limitations, community expectations, staffing realities, or local crime patterns. Evidence becomes most useful when combined with research, local data, professional expertise, community context, and continuous evaluation.
Real-World Application
A police chief considering whether to add patrol officers to a high-crime district should ask: where exactly will those officers be deployed, doing what, at what times? Simply adding cars to random patrol is unlikely to produce measurable results based on available research. Directed patrol — officers assigned to specific locations during specific time windows based on crime analysis — has a substantially stronger evidence base. The question is not just 'how many officers' but 'where, when, and doing what.'
Why It Matters
For Police Officers
Officers who understand research design can evaluate claims about new programs and strategies critically. When an administrator proposes a new initiative, understanding the difference between correlation and causation helps officers ask the right questions about whether the evidence actually supports the approach.
For Students
Evidence-based policing is foundational to advanced criminal justice study. Students who understand research design, causal inference, and the limits of generalization can critically evaluate policy claims rather than accepting them at face value.
Think About It
Should police agencies require empirical evidence before implementing major new enforcement strategies?
How much evidence should be necessary before research findings influence policy?
Can a strategy be statistically effective but operationally undesirable?
How should agencies evaluate programs when randomized experiments are impossible?
Sources & Further Reading
U.S. Department of Justice, CrimeSolutions. Hot Spots Policing program rating: Effective for reducing overall crime. crimesolutions.ojp.gov.
Sherman, L. W. (1998). Evidence-Based Policing. Police Foundation.
Lum, C., & Koper, C. S. (2017). Evidence-Based Policing: Translating Research into Practice. Oxford University Press.
National Institute of Justice. Evidence-Based Policing resources. nij.gov.