Randomized Controlled Trial in Criminology
Also known as: Criminological Field Experiment, Experimental Criminology Trial, Place-Based Randomized Trial, Justice RCT
A randomized controlled trial (RCT) in criminology evaluates a justice intervention — such as hot-spots policing, a deterrence message, or a reentry program — by randomly assigning units (places, people, or cases) to receive the intervention or to serve as controls. Because assignment is by chance, treatment and control groups are statistically equivalent at baseline, so any later difference in crime or reoffending can be attributed to the intervention rather than to selection. Sherman and Weisburd's 1995 Minneapolis hot-spots patrol experiment helped establish the design as the gold standard of experimental criminology.
Key highlights
- Randomization removes selection bias, giving an unbiased estimate of the causal effect even of unmeasured confounders.
- The intention-to-treat analysis is simple, transparent, and robust, requiring few modeling assumptions to interpret.
- Provides the strongest internal validity available for evaluating crime-control and justice interventions.
- Supports clear, decision-relevant statements about whether a program works and by how much.
- Block and cluster randomization adapt the design to place-based policing and multi-site program evaluation.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use an RCT when an intervention can be ethically and practically assigned at random to a meaningful number of units, and you need credible evidence that the intervention — not selection or secular trends — caused a change in crime or justice outcomes. It is well suited to evaluating policing tactics at places, sentencing and supervision options, treatment and reentry programs, and deterrence communications. It is less appropriate when randomization is impossible or unethical (you cannot randomly assign neighborhoods to severe disadvantage), when the number of available units is too small for chance to balance the groups, or when spillover between treatment and control units contaminates the comparison. Observational designs such as propensity weighting or regression discontinuity are fallbacks when an experiment is infeasible.
Strengths & limitations
- Randomization removes selection bias, giving an unbiased estimate of the causal effect even of unmeasured confounders.
- The intention-to-treat analysis is simple, transparent, and robust, requiring few modeling assumptions to interpret.
- Provides the strongest internal validity available for evaluating crime-control and justice interventions.
- Supports clear, decision-relevant statements about whether a program works and by how much.
- Block and cluster randomization adapt the design to place-based policing and multi-site program evaluation.
- Random assignment is often infeasible or unethical for many justice questions, limiting where the design can be used.
- External validity can be weak: an effect found in one jurisdiction or trial population may not generalize elsewhere.
- Spillover and displacement between treatment and control units (especially nearby places) can bias estimates if ignored.
- Non-compliance, attrition, and treatment dilution can erode the difference between arms and complicate interpretation.
- Many criminological trials are small or under-powered, so chance imbalance and wide confidence intervals remain a risk.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
Why randomize places instead of people in policing experiments?
Crime is highly concentrated at a small number of micro-places, and many policing tactics — patrol, problem-solving, environmental changes — act on locations rather than individuals. Randomizing street segments, addresses, or hot-spot clusters lets the trial estimate the effect of the place-based tactic directly, and block randomization within crime-level strata improves balance and precision when the number of places is modest.
What is the difference between intention-to-treat and treatment-on-the-treated?
Intention-to-treat (ITT) compares groups as randomly assigned, regardless of whether each unit actually received the intervention; it preserves the unbiasedness of randomization and answers 'what is the effect of offering the program?'. Treatment-on-the-treated (the local average treatment effect) rescales the ITT by the difference in uptake to estimate the effect among those whose behavior was changed by assignment, but it relies on additional instrumental-variable assumptions.
How is spillover handled in place-based trials?
Spillover — where the intervention at a treated place affects nearby control places through displacement or diffusion of benefits — can bias the simple difference in means. Designs address it by using buffer zones, randomizing larger clusters, measuring outcomes in catchment areas around treated and control units, and analyzing displacement and diffusion explicitly rather than assuming control places are unaffected.
Sources
- 1.Sherman, L. W., & Weisburd, D. (1995). General deterrent effects of police patrol in crime hot spots: A randomized, controlled trial. Justice Quarterly, 12(4), 625–648.
- 2.Weisburd, D. (2003). Ethical practice and evaluation of interventions in crime and justice: The moral imperative for randomized trials. Evaluation Review, 27(3), 336–354.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 22). Randomized Controlled Trial in Criminology. ScholarGate. https://scholargate.app/criminology/randomized-controlled-trial-criminology