In this chapter, we test the effects of community policing in the Sorsogon Province of the Philippines. The intervention generated a four-fold increase in police-citizen interactions in treated villages, but consistent with meta-analysis of all six sites in this volume, we found no effects of the intervention on crime rates or citizens’ attitudes about public safety. To disaggregate the effects of different aspects of community policing, we sequenced the implementation of community engagement (CEP) and problem-oriented policing (POP) but found no effects on the harmonized outcomes of either CEP on its own or the combination of CEP and POP. Finally, we present suggestive evidence of positive impacts on the specific types of crimes that barangays’ problem-oriented policing teams elected to focus on, indicating that while community policing cannot address all of a community’s problems en masse, it may improve specifically targeted issues.

In this chapter, we test the efficacy of community policing in thirteen districts throughout rural Uganda. As in many authoritarian regimes, police in Uganda serve the dual role of providing security to citizens on the one hand and quelling dissent and opposition on behalf of the regime on the other. Community policing may help citizens delink the political arm of the police from less politicized local officers. The community policing initiative we study was locally designed and funded by the Ugandan police. Our evaluation combines administrative crime data from the Uganda Police Force with surveys of thousands of Ugandan citizens, local leaders, and police officers. While the initiative we study succeeded in increasing the frequency of interactions between citizens and the police in these far-flung villages and improved citizens’ understanding of the criminal justice system, we find no evidence that it reduced crime, enhanced perceptions of safety, improved attitudes towards the police, or strengthened norms of cooperation with the police. These results are consistent with other chapters in this volume and point to the potential limitations of community policing in low-income countries.

Between the early 1990s and the mid-2010s, citizen security in Medellín dramatically improved and police violence declined. But residents’ trust in police stagnated. We evaluate a police-led effort to build trust through town-hall-style police–community meetings. In 174 treated neighborhoods – but not in 173 control neighborhoods – the police held more than 500 such meetings over a period of nine months. We find that the meetings induced small positive changes in perceptions of the police, though they did not alter trust in police per se – or crime reporting behavior, much less crime itself. We interpret these findings as evidence that voluntary informal contact between residents and police officers is a weak but not irrelevant policy for reshaping police–community relations.

This chapter describes patterns of crime and insecurity in the six study countries where the coordinated randomized trials took place and how these places compare to other countries. It then provides a theoretical framework for understanding the causes of crime, why crime rates often diverge from citizen perceptions of insecurity, and how crime and perceived insecurity can be reduced. The chapter concludes by surveying the set of policy tools used to reduce crime, and how policing fits into those tools.

This chapter lays out the theoretical foundations of community policing and highlights evidence gaps in evaluations of community policing’s effectiveness. Community policing is a law enforcement strategy that centers around building trust between police and citizens as well as promoting citizen engagement with authorities in order to advance public safety. The chapter describes the origins of community policing as well as the logic of how it might render the police more effective, primarily through improved information provision from citizens. Despite substantial support for community policing, a systematic review detailed in the chapter reveals significant evidence gaps in evaluations of the effectiveness of community policing interventions such as beat patrols and the police engaging in town hall meetings. The review finds that the evidence gaps are particularly acute with respect to evaluations in Global South communities.

This chapter discusses how to interpret the findings from six randomized experiments on community policing, and the implications for policymaking and police reform. The bottom line is that locally appropriate increases in the strength of community policing practices do not generate the changes to trust in the police, citizen cooperation, or crime reduction that we hypothesized or that its advocates claim. The evidence suggests, at a minimum, that caution should be exercised in advocating for the adoption or continuation of community policing in the Global South. New evidence may emerge that shows community policing can be effective in a different type of context, when implemented in response to demands from a social movement of citizen groups, with a different set of institutional preconditions, or in combination with other reforms, such as citizen accountability boards. Until it does, we suggest that it be deprioritized in the list of policy levers to reduce crime and build trust in police in the Global South.

This chapter introduces a research design to study the effects of community policing. The chapter introduces the Metaketa model of multi-site trials, which are used to answer questions relevant to policy using coordinated experiments in which the same intervention is randomly assigned to units in multiple contexts and the same outcomes are measured to estimate effects. In specific, the chapter introduces how the six countries were selected for study and describes their characteristics in terms of crime and policing and then how the interventions were selected and harmonized across the settings and how they compare to community policing policies in the world. The remainder of the chapter details the experimental design, from how police beats and units are sampled, how community policing intervention was randomly assigned, how outcomes were measured and harmonized, how effects were estimated for each site and then averaging across sites, and how we planned to address threats to inference.

What is the effect of community policing in settings where the rule of law is weak and communities gripped by crime turn to vigilantism to deter and prevent crime? In this chapter, I test the effectiveness of the Liberia National Police’s model of community policing, which focuses not only on building citizens’ trust and cooperation but also on providing communities with an alternative to vigilantism via its Community Watch Forum initiative. Drawing on large-scale crime surveys and administrative data, I find that the program led to moderate improvements in perceptions of police intentions, norms of cooperation, and perceptions of police capacity. I also find that the program increased community contributions to local security groups, reduced support for mob violence, and reduced reports of actual mob violence incidents by 39 percent. Despite these improvements, the program had no significant effect on other forms of crime victimization, crime reporting, crime tips, or residents’ sense of security.

This chapter summarizes the findings from our study, based on the meta-analysis averaging across the effects from the six experiments. We found that increases in locally appropriate community policing practices led to no improvements in citizen–police trust, no greater citizen cooperation with the police, and no reduction in crime. Despite a strong commitment from leadership in each context at the outset, the police implemented the interventions unevenly and incompletely. Although citizens reported more frequent and robust exposure to the police in places where community policing was implemented, we have limited evidence of police action in response to citizen reports.

This chapter describes the emergence of policing as an institutional mechanism for maintaining order in increasingly urban social contexts. Through a review of prior literature, we identify three impediments to police effectiveness – autonomy, capacity, and principal–agent problems – and explain the ways in which poor performance undermines citizens’ trust in police and willingness to cooperate. Then, using data from citizen and officer surveys, we illustrate the ways in which a lack of trust between citizens and the police undermines effective policing across the six countries that are the focus of this study.

This chapter studies the effect of Rede de Vizinhos (RdV or “Neighbor Network”) community policing program in Santa Catarina, Brazil, which aims to improve public safety and trust between citizens and police by facilitating real-time information about crime and public safety through dedicated WhatsApp instant messages groups with the participation of a police officer. We randomly allocated neighborhoods that would see the policy implemented into treatment and control groups, with the former being exposed to an information campaign through which we publicized induction meetings using Facebook. Despite reaching roughly 10 percent of Santa Catarina’s population, our study does not find a differential participation rate in the RdV campaign regions. Further, we don’t find evidence of increased perceptions over the police or improvements over criminal rates in treated neighborhoods. Our results suggest that despite their promise, (technology-enabled) community policing programs may fail to deliver substantial impacts given saturation dynamics and diminishing marginal returns – our baseline surveys indicated that 52 percent of respondents had heard already about the RdV program with 13.1 percent actively participating.