Brief 87: How Does Community Policing Affect Police Attitudes? An Experimental Test and a Theory of Bureaucrat-Citizen Contact
EGAP researchers: Dotan Haim, Nico Ravanilla
Other researchers: Matthew Nanes
Key takeaways: A seven-month community policing program in the Philippines substantially improved officers’ understanding of citizens’ public safety concerns, but did not increase their trust, empathy, accountability, or views on corruption toward citizens. These attitudinal effects depended heavily on context: community policing improved attitudes only among officers who were not already embedded in the local community, and it reduced trust among officers working in areas with active insurgent activity.
Geographical Region: Philippines
Type of study: Pre-registered field experiment
Preparer: Danielle Bohonos
Executive Summary
This study asks how community policing affects the attitudes of the officers who implement it, a question largely overlooked by a literature that focuses almost exclusively on citizens’ attitudes toward police. The authors conducted a randomized field experiment in Sorsogon Province, Philippines, in which police officers were assigned to “One Sorsogon,” a community policing program involving joint problem-solving meetings with village leaders, citizen-facing patrols, and town halls. Contrary to the conventional wisdom that increased contact with citizens should improve officers’ trust, the authors find that officers came to share citizens’ public safety concerns but showed no improvement on the other attitudinal outcomes. Building on this puzzle, the authors develop an inductive theory grounded in intergroup contact theory, showing that community policing’s effects on officer attitudes depend on officers’ pre-existing embeddedness in the community and on whether the local security context heightens perceived threats to officer safety.
Policy Challenge
Community policing — informal, non-enforcement contact between officers and citizens intended to build mutual trust and understanding — is one of the most recommended reforms for improving strained police-citizen relations around the world. However, existing research on community policing has focused almost entirely on citizen-facing outcomes such as citizen trust in police, crime reporting, and crime rates. Far less is known about how these same programs affect officers themselves, despite theoretical and practical reasons to expect that officer attitudes are just as consequential for the quality of policing as citizen attitudes are.
This gap sits within a broader literature on street-level bureaucrats, who exercise substantial discretion in how they implement policy. Existing work in this area has largely treated bureaucrats’ attitudes and local knowledge as fixed traits shaped by pre-existing characteristics such as personality, shared identity with citizens, or local embeddedness, rather than as outcomes that can be shaped by the governance strategies and social contact bureaucrats experience on the job.
Context
The study takes place in Sorsogon Province, a semi-rural province in the Philippines with a population of about one million people and a poverty rate slightly above the national average. Public safety is provided jointly by the Philippine National Police (PNP) and barangay-level “tanods,” civilian safety officers who handle minor disputes and help route more serious issues to the PNP. Prior to the intervention, PNP officers had very limited contact with communities: officers rarely conducted town halls or met with community leaders, and more than 45% of surveyed residents reported seeing an officer once a month or less. Sorsogon is also a longstanding hotspot for the New People’s Army (NPA), a communist insurgency active in the province since the 1960s, though the study sample was restricted to barangays with limited NPA presence for safety and ethical reasons. The “One Sorsogon” program was initiated by the Provincial Police Chief as a service-oriented alternative to the violent tactics associated with the national “War on Drugs,” to improve the PNP’s local reputation even though little drug-war violence had occurred in the province itself.
Research Design
From a pool of 705 eligible officers (holding the ranks of Police Officer or Senior Police Officer), the authors randomly selected 198 officers, blocked by police station and rank, to participate in community policing teams in 99 randomly selected barangays; 47 additional officers were later added as randomized replacements for officers who were reassigned, bringing the total treated group to 245 for the intention-to-treat analysis. The seven-month program began with a two-day training on community policing best practices, followed by six monthly “Problem-Oriented Policing” (POP) meetings between assigned officers and barangay leaders to identify and address local public safety concerns, along with ongoing citizen-facing patrols and a culminating town hall in each barangay. Control officers continued their normal duties without participating in any study activities.
The authors surveyed officers before and after the intervention to construct five pre-registered indices: sharing citizens’ public safety concerns, trust in citizens, empathy for citizens’ concerns, perceived organizational accountability for citizen-facing misconduct, and views on the seriousness of corruption. Treatment effects were estimated via linear regression controlling for baseline outcomes and block fixed effects, with standard errors clustered at the police station level. The authors report that treatment officers displayed high compliance (attending 84% of POP meetings and over 90% of town halls) and that the design is well powered, balanced across treatment and control groups, and robust to a range of alternative explanations for null findings, including compliance, ceiling effects, attrition, and spillovers to control officers.
Results
Consistent with the authors’ first hypothesis, community policing produced a large and statistically significant 24% increase in the number of public safety concerns that officers shared with local citizens. However, the intervention had no detectable effect on any of the other four pre-registered outcomes — trust, empathy, organizational accountability, or views on corruption — and the study is well powered enough to rule out even modest positive effects. The authors show these null results are not easily explained by low statistical power, poor compliance, ceiling effects in baseline attitudes, differential attrition, or spillovers between treatment and control officers.
To make sense of this puzzle, the authors turn to exploratory, non-pre-registered heterogeneous-effects analyses informed by qualitative observations from field staff who attended every program meeting. They find that community policing substantially improved trust and empathy among officers who were not originally from Sorsogon Province, essentially closing the gap with locally embedded officers by the end of the program, while producing no improvement among officers who were already embedded in the community. Separately, the effect of the program on officer trust depended sharply on local security conditions: community policing modestly increased trust in areas without NPA insurgent presence but significantly decreased trust in areas where the insurgency was more active, even though the citizens officers interacted with in these meetings were themselves law-abiding and cooperative.
Lessons
The central lesson for policymakers is that community policing is not a uniformly effective tool for improving how officers view the citizens they serve — it can build shared understanding of local concerns without changing officers’ underlying trust, empathy, or sense of accountability, and its effects on these deeper attitudes are highly conditional. Programs are more likely to shift attitudes among officers who are not already embedded in the community, since embedded officers typically already have the informal citizen contact, local knowledge, and shared identity that community policing is designed to build. In areas with active security threats, contact with citizens can backfire by raising the salience of danger and hardening officers’ suspicion of the broader population, even when the specific citizens involved are cooperative — suggesting that departments should not assume community policing will “overcome” a dangerous neighborhood context and may need to pair it with complementary reforms, such as perspective-taking trainings, in higher-risk areas. More broadly, the authors argue that other frontline state agents who exercise similar discretion — soldiers, social workers, healthcare workers — likely face comparable dynamics, meaning governance reforms aimed at improving bureaucrat-citizen relations should be designed with attention to bureaucrats’ pre-existing local ties and the safety context in which contact occurs, rather than assuming that more contact is uniformly beneficial.