Can Tougher Misconduct Investigators Improve Police Behavior Without Causing De-Policing?

This study examines whether stricter misconduct investigators reduce future police misconduct without decreasing productivity. Using quasi-random assignment of civilian complaints to investigators in Chicago from 2006–2018, the authors measure the impact of being assigned to supervisors more likely to obtain affidavits. They find that officers investigated by high-affidavit supervisors receive 15% fewer complaints in the first 3 months after the complaint and, before the 2015 Laquan McDonald scandal, make 5.5% more arrests. However, after the scandal, these officers reduce arrests by 4.4%, indicating de-policing. The findings highlight that oversight improves behavior in ordinary times but can backfire amid public scrutiny.

Full Citation and Link to Article

Cheryl L. Phillips, Brent D. Mast, and Michael H. Becker (2024). “The Impacts of Misconduct Investigators on Police Officer Behavior,” Journal of Policy Analysis and Management (Advance online publication). Available at https://doi.org/10.1002/pam.70002

Extended Summary

Central Research Question

This study investigates whether stronger oversight of police misconduct—specifically through more rigorous civilian complaint investigators—can reduce future police misconduct without leading to a reduction in officer productivity. The researchers focus on whether being assigned to a misconduct investigator with a higher propensity to obtain sworn affidavits (required to open formal investigations) affects the future behavior of officers. They also examine whether these effects differ in periods of normal oversight versus periods of public scrutiny, such as after the release of the Laquan McDonald shooting video in 2015. The study ultimately asks whether formal complaint investigations can be a constructive tool to improve officer conduct without unintended consequences like de-policing.

Previous Literature

The study contributes to a growing body of literature on police oversight and misconduct. Prior research has shown that police misconduct disproportionately harms Black individuals and communities. Much of this literature examines large-scale reforms triggered by public scandals (e.g., Devi & Fryer, 2020), finding that such events can lead to “de-policing,” where officers reduce their proactive engagement to avoid scrutiny. Other studies, such as Rozema and Schanzenbach (2020), demonstrate that officers with sustained misconduct complaints tend to improve behavior, but these studies cannot isolate the causal role of formal investigations themselves.

Additionally, a body of research on “judge instruments” (e.g., Aizer & Doyle, 2015) uses quasi-random assignment of cases to decision-makers with varying stringency levels to assess treatment effects. This study applies a similar strategy to the context of police discipline, offering a new approach that avoids confounding from officer self-selection or department-wide shocks. It also builds on the broader literature on monitoring and incentives in public service delivery (Duflo et al., 2013; Olken, 2007), where better oversight often leads to improved outcomes.

Data

The study uses complaint, personnel, and arrest data from the Chicago Police Department (CPD) between 2006 and 2018. The primary dataset comprises civilian complaints filed with the Independent Police Review Authority (IPRA) and later with the Civilian Office of Police Accountability (COPA). The analysis sample includes 2,049 complaints assigned to 11 supervising investigators. The focus is on allegations eligible for IPRA review, primarily excessive force and civil rights violations, excluding cases like internal affairs investigations, domestic violence, and shootings.

To measure officer behavior, the authors merge this data with arrest records and complaint histories. Arrests are categorized by total arrests and “Type 1” serious arrests, such as robbery or assault. They also use administrative data on officer assignments and attendance, though this data is only available for part of the sample. A key feature of the dataset is the quasi-random assignment of complaints to supervising investigators during most of the sample period, which provides the basis for causal inference.

Methods

The empirical strategy relies on the quasi-random assignment of civilian complaints to supervising investigators with varying tendencies to pursue complaints aggressively—measured by their propensity to obtain affidavits. The authors construct an instrument at the supervisor-year level that reflects each supervisor’s average affidavit acquisition rate, excluding the case in question (leave-one-out mean). This instrument is used in a reduced-form design to estimate the impact of being assigned to a “high-affidavit” supervisor on subsequent officer outcomes.

The main outcomes analyzed are monthly complaints, total arrests, and Type 1 arrests, each measured in three time frames: within 3 months after the complaint, months 4–12, and the second year after the complaint. The first period captures behavior during the early investigation; the latter periods reflect later stages of the investigation or its conclusion.

Robustness checks include verifying the balance of covariates across supervisor assignments and testing for variation in supervisor behavior over time. The authors avoid a Local Average Treatment Effect (LATE) interpretation due to possible violations of exclusion restrictions—e.g., supervisors might influence outcomes via channels other than affidavits.

Findings/Size Effects

The study’s key finding is that being assigned to a high-affidavit supervising investigator causes officers to receive fewer complaints in the short term without reducing arrests—at least under normal conditions. Specifically, officers under high-affidavit supervisors experience a 15% reduction in new complaints during the first 3 months after a complaint is filed. This effect is temporary: there is no statistically significant effect on complaints beyond the first year.

Crucially, officer productivity does not appear to suffer. Prior to the 2015 Laquan McDonald scandal, officers assigned to tougher investigators actually made 5.5% more arrests in the months following the complaint. The authors interpret this as increased effort—officers may be working harder or more cautiously in response to perceived scrutiny. There is no evidence of de-policing or disengagement in this period.

However, the story changes after the McDonald scandal, which led to heightened public scrutiny and a major overhaul of police oversight in Chicago. In this post-scandal period, officers assigned to high-affidavit investigators made 4.4% fewer arrests—evidence of de-policing. The reduction in productivity is not accompanied by a corresponding drop in complaints, suggesting that the effect is driven by disengagement rather than improved behavior.

These results indicate that oversight can have different effects depending on the broader political and institutional environment. In normal times, enhanced investigation seems to encourage improved conduct and even more proactive policing. In times of public crisis or perceived hostility toward police, the same oversight tools may backfire, causing officers to pull back from their duties.

Additional analyses support these findings. Investigators’ affidavit propensity is not correlated with case characteristics or officer demographics, confirming the plausibility of random assignment. The first-stage relationship between supervisor stringency and actual affidavit acquisition is strong (coefficient ~0.93), supporting the instrument’s relevance. The authors also test whether certain supervisor-investigator team dynamics drive the results but find that results hold across specifications.

Conclusion

This study provides important causal evidence that formal complaint investigations—specifically those led by more rigorous supervising investigators—can improve police officer behavior, at least in the short term and under typical institutional conditions. By reducing future complaints without dampening arrest activity, strong oversight seems to create beneficial behavioral responses among officers. These findings support the idea that personnel oversight, when structured effectively, can serve as a valuable policy tool to improve public service delivery.

However, the study also shows that the effects of oversight are context-dependent. During periods of scandal and heightened public attention, increased scrutiny may trigger officer withdrawal, leading to fewer arrests without any gain in accountability. This de-policing response highlights the need for thoughtful implementation of reforms that account for the institutional climate and officer perceptions.

From a policy standpoint, the results suggest several actionable takeaways. Agencies may benefit from promoting the practices of high-affidavit supervisors and expanding investigative capacity. At the same time, reforms must be designed to avoid creating a hostile or overly punitive environment that leads to disengagement. Policies such as relaxing affidavit requirements, improving training, or using body-worn camera footage may strike a better balance between accountability and performance.

In sum, the study contributes to a nuanced understanding of how police oversight mechanisms function and how their impact can vary depending on the broader social and institutional context. It highlights both the potential and the limits of misconduct investigations as a lever for reform, emphasizing the importance of context-sensitive strategies in public sector governance.