Protection Detail
Improving police behavior through artificial intelligence: Pre-registered experimental results in two large US agencies
Ian Adams, Kyle McLean & Geoffrey Alpert
Criminology, forthcoming
Abstract:
Police body-worn cameras (BWCs) generate extensive video data on officer behavior, yet resource constraints mean that only a fraction of this footage is ever reviewed. Recent advancements in artificial intelligence (AI) present an opportunity to automate these reviews, potentially influencing police conduct. We evaluate the impact of AI-led auditing in two pre-registered, randomized controlled trials conducted within two large US police agencies. The findings are mixed but positive: In a department operating under a consent decree, officers exhibited significant reductions in substandard professionalism, whereas in an agency without external oversight, officers increased the frequency of highly professional interactions. These results show that AI auditing of BWC footage can influence officer behavior, with its effects shaped by organizational and institutional contexts.
Walling Off Crime: An Analysis of the Local Deterrent Effects of Increased Border Control
Lawrence LaPlue & Randy McFerrin
American Law and Economics Review, forthcoming
Abstract:
This paper analyzes the deterrent effects of both border patrol agents and increased United States–Mexico fencing on reported crime rates in U.S. cities located along the border, between 1992 and 2016. We find that, for cities along the United States–Mexico border, increases in the number of border patrol agents, within patrol sectors, are associated with significant declines in property crime rates and that increased border fencing is associated with significant declines in both property and violent crime rates. Further investigation into the underlying mechanisms provides evidence that border control measures are associated with significant reductions in the apprehension of migrants in the border region with prior convictions for other, non-migration-related crimes. The avoided crime indicated by the baseline estimates represents an annual economic benefit, ranging, on average, from $2,300 for an additional border patrol agent to $650,700 for an additional mile of in-city fencing, relative to an estimated cost of $174,000 to hire an agent and $2.8 million to install a mile of border fencing.
Framing fairness: Understanding perceptions of police procedural justice in body-worn camera footage
Allison Cross et al.
Law and Human Behavior, forthcoming
Method: Adult White and Black participants from Prolific were randomly assigned to watch either a procedurally just or unjust BWC traffic stop video. They self-reported their perceptions of procedural justice.
Results: As expected, there was more variability in procedural justice perceptions of the unjust BWC video than the just video. Liberal-leaning participants believed the BWC video was less procedurally just than conservative-leaning participants. Contrary to our hypothesis, Black participants had more positive procedural justice perceptions of the BWC videos than did White participants. A serial mediation analysis revealed that these demographic differences typically operated through baseline procedural justice perceptions and identification with the officer and driver.
Behind the yellow sticker: Paradoxical effects of a visual warning of body-worn cameras on the use of police force
Noy Assaraf et al.
Journal of Experimental Criminology, December 2025, Pages 1215-1237
Methods: A six-month cluster-randomised controlled trial involving spatiotemporal police units was conducted in Miami Beach, USA. The units were randomly assigned to the experimental group, which included officers who wore BWCs featuring multiple yellow stickers and “VIDEO & AUDIO” logo (used as a visual warning). Officers in the control group were equipped with BWCs without yellow stickers. Neither group was required to announce the presence of the BWCs, thus isolating the visual warning from the audial warning. Poisson regression models estimate the treatment effect, with confirmatory subgroup analyses based on the proactive versus reactive interactions.
Results: Statistically significant differences in the use of force by officers, but in the contrary direction: higher rate of use of force due to equipping officers with BWCs with yellow stickers relative to BWCs without the yellow stickers. Relative change analysis indicates that yellow stickers cause an increase in the rate of use of force in incidents involving proactive policing, with a less pronounced increase in reactive policing.
The Law and Economics of Guilt and Shame
Ian Ayres, Joseph Bankman & Daniel Hemel
University of Chicago Law Review, January 2026, Pages 247-299
Abstract:
The negative moral emotions of guilt and shame impose real social costs but also create opportunities for policymakers to engender compliance with legal rules in a cost-effective manner. We present a unified model of guilt and shame that demonstrates how legal policymakers can harness negative moral emotions to increase social welfare. The prospect of guilt and shame can deter individuals from violating moral norms and legal rules, thereby substituting for the expense of state enforcement. But when legal rules and law enforcement fail to induce total compliance, guilt and shame experienced by noncompliers can increase the law’s social costs. We identify specific circumstances in which rescinding a legal rule will improve social welfare because eliminating the rule reduces the moral costs of noncompliance with the law’s command. We also identify other instances in which moral costs strengthen the case for enacting legal rules and investing additional resources in enforcement because deterrence reduces the negative emotions experienced by noncompliers. We end by exploring the implications of our framework for legal policy across “guilt cultures” and “shame cultures,” for the debate over shaming sanctions, and for other moral emotions such as resentment and virtue.
A Second Chance at Schooling? Unintended Consequences of Prison Education
Romaine Campbell & Logan Lee
Cornell Working Paper, January 2026
Abstract:
Prison education programs are rapidly expanding across the United States, but we have little causal evidence on the impact of prison education. We estimate the im-pact of prison education on reincarceration, future employment and education, and in-prison misconduct using detailed administrative data from Iowa. To address selection bias, we instrument for courses taken during incarceration with an opportunity metric based on course availability. We find that participating in education increases the likelihood of reincarceration within three years of release. The effect is driven by revocations rather than new crimes and is more pronounced for post-secondary education and among white prisoners. Exploring mechanisms, we find that participating in education increases the likelihood that prisoners are released with significant supervision requirements -- likely contributing to increased revocations. Our findings highlight a potential unintended consequence for policymakers and practitioners to consider as prison education programs continue to proliferate.
Does Distance from Home Matter in Prison? Effects on Visitation and Recidivism
Danielle Nemschoff
University of Chicago Working Paper, January 2026
Abstract:
This paper studies how the distance between prison and an incarcerated individual’s home affects their likelihood of recidivism. Leveraging a unique dataset covering more than 20,000 incarcerated individuals and over 200,000 prison visits, I exploit quasi-random variation in home-to-prison distance generated by facility assignment rules. I find that a 100-mile increase in placement distance raises prison readmission within 3 years by almost 4 percent. This effect is driven by a reduction in visitation, with individuals placed farther from home receiving significantly less visits. While social support is theorized to reduce recidivism, there is limited causal evidence on how maintaining these connections through visitation during incarceration affects recidivism. To address this, I use distance as an instrument for visitation, and find that an additional visit per month lowers the likelihood of re-incarceration by roughly 14 percent within one year post-release and 7 percent within three years post-release. I also show that an additional visit per month shortens the fraction of sentence served by one percent and reduces housing instability by 12 percent, the former consistent with a reduction in misconduct and the latter an important mechanism for successful post-release outcomes. Counterfactual estimates suggest assigning individuals closer to home could reduce recidivism by 2 to 4 percent.
The effect of civilian monitoring on fatal encounters
Matthew Harvey
Contemporary Economic Policy, forthcoming
Abstract:
Does civilian monitoring affect police use of force? Using evidence from the Safe Passage program, I exploit geographic and temporal variation in the program's release in a difference-in-difference framework. I find that fatal encounters decrease in areas where the program was implemented by 33%. While this effect cannot be directly tied to the presence of monitors, it is robust to multiple subsets of fatal encounters and levels of fixed effects. The program also leads to an increase in police presence as captured by investigatory stop and a very small increase in police usage among civilians. However, civilian sentiments as captured by self-reported trust and safety scores do not change in response to the Safe Passage program; I also find no statistically significant change in complaints against officers after the program.
Effect of juvenile justice financial sanctions on youths’ recidivism
Luyi Jian, Jennifer Skeem & Jaclyn Chambers
Law and Human Behavior, forthcoming
Method: We accessed, linked, and analyzed data from county and state agencies for a sample of 2,401 youth under supervision. We applied a rigorous causal inference approach (targeted maximum likelihood estimation) combined with machine learning to test the hypotheses.
Results: Financial sanctions overall modestly increased the likelihood of both probation violations (from an estimated 9% to 14%) and rearrests (from an estimated 54% to 58%) -- but fees alone did not significantly predict either outcome. The effects of financial sanctions on recidivism were not moderated by the youth’s race, socioeconomic status, or cumulative risk.
Evaluating the effectiveness of a fatigue training intervention for the Seattle Police Department: Results from a randomized control trial
Lois James, Stephen James & Loren Atherley
Journal of Experimental Criminology, December 2025, Pages 1121-1135
Methods: Using a randomized control trial (RCT) experimental design, employees from the Seattle Police Department were exposed to a fatigue training intervention. Measurement included wrist actigraphy to objectively measure sleep, as well as a battery of validated surveys to measure sleep quality, sleepiness, depression, anxiety, and PTSD symptomatology.
Results: Study results revealed that the training significantly improved employee sleep (+ 18 min per 24-h period) and reduced rates of depression, anxiety, PTSD symptomatology, and likelihood of falling asleep at the wheel.