POLICING AND CITIZEN TRUST IN KENYA: HOW COMMUNITY POLICING SHAPES LOCAL TRUST-BUILDING AND COLLABORATION

In contexts of high insecurity and mistrust in the police, how and why do local residents still choose to collaborate with the police, and what is the role of community policing in such considerations? Research on policing in Africa has emphasized the structural and macropolitical barriers to effective police reform, including institutionalized cultures of impunity and corruption. Less attention, however, has been paid to the contextual and relational dynamics that shape police-community collaboration. We argue that a relational perspective, which centres local residents’ interactions with police and community policing structures, provides novel insights into the challenges of policing reforms. This perspective also demonstrates how contingent and incremental trust can be built in very challenging circumstances. We study these dynamics in Karagita and Kaptembwo, two low-income urban settlements in Nakuru County, Kenya, that have experienced violent crime and repeated electoral violence. Despite considerable challenges of crime, police misconduct,


Introduction
Research on policing in Africa has documented high mistrust in and even fear of the police, particularly in urban areas, where residents often disengage from the police and pursue alternative security strategies. 1 In many cases, community policing has been introduced to increase citizen trust in the police and to encourage citizens to report incidents and collaborate with law enforcement. 2Nevertheless, globally, the implementation of community policing has often failed to strengthen community cooperation with the police. 3Studies on policing in Africa have emphasized the structural and macropolitical barriers to effective reform, including institutionalized cultures of impunity and corruption, 4 and uncovered the individual-level determinants of police trust, legitimacy, and collaboration. 5However, research has paid less attention to the contextual and relational dynamics that shape police-community collaboration.In this study, we shift focus to explore the micro-dynamics of police-community relations where insecurity is high, citizen trust in the police is low, and community policing is being implemented.
Kenya has adopted community policing to move towards more democratic policing.The challenges are significant: after decades of colonial policing, the police have been politicized and have continued to be used by the state for repressive purposes. 6While the police force is multiethnic, incumbents have secured important positions within the security establishment for their ethnic kin. 7Police corruption is widespread, and there is an institutionalized culture of impunity. 8Inadequate resources and logistical issues, including understaffed stations and lacking equipment, hamper effective police responses. 9Police reforms have been seen more as 'tactical concessions or gestures to donors on the part of political elites'. 10 Kenyans' trust in police is, consequently, staggeringly low: five surveys over the time period 2011-2022 show that more than 60 percent of the population report no or just a little trust in the police. 11espite entrenched mistrust and a history of police misconduct, however, residents sometimes choose to collaborate with police and form less negative perceptions of them.How can such instances be understood, and what do they tell us about the broader dynamics of police-community relations?We study these questions in Karagita (Naivasha) and Kaptembwo (Nakuru City) in Nakuru County, two low-income urban settlements that have experienced violent crime and repeated electoral violence. 12In these settlements, community policing has been presented as a solution to pervasive insecurity and police have formed Community Policing Committees (CPCs), whose members have received training from civil society on community policing.These forms of urban settings are particularly important to study for scholars and practitioners interested in how local insecurity can be addressed.First, while improved police-community relations are necessary across Kenya, the need is most urgent in urban areas: urban residents display lower trust in police than rural residents. 13Second, low-income, informal settlements are particularly crime-prone and under-protected, and such settlements are where the police have engaged in extrajudicial killings and profiling of male youth. 14e argue that a relational perspective, which centres local residents' interactions with police and community policing structures, provides previously neglected insights into the challenges of policing reforms and how contingent and incremental trust can be built in very challenging circumstances.By concentrating on how and when residents in high-insecurity areas choose to collaborate with police and the meaning they give to community policing, we complement existing work on policing in Africa.First, while research on African cases in part focuses on the everyday experiences of public security, most studies approach this question from the perspective of the public's turn to security provision by non-state actors, such as vigilantes. 15Furthermore, research on community policing as part of police reform is primarily focused on the barriers to functioning security structures, and few studies in the African context highlight when reform has the potential to work. 16wo main findings emerge from our study.First, while mistrust of the police is widespread, our analysis underlines the significance of everyday interactions between the police and communities in improving citizen collaboration and willingness to share information.By uncovering instances of collaboration and the underlying rationale residents offer for their actions, we find that improved police effectiveness is a primary motivation for why residents sometimes choose to collaborate with and form more positive perceptions of the police.Relatedly, we find that residents will trust individual police officers but not the police as an institution.Second, we find that community policing initiatives can shape the relationship between the community and the police in important ways.Specifically, the analysis shows that community policing structures and the community members who serve on the CPCs can become communication nodes and facilitate information-sharing when local residents fear reporting directly to the police.Despite considerable challenges of crime, police misconduct, and political interference in Karagita and Kaptembwo, our findings point to how community policing structures and police reform can contribute to incremental improvements in police-community relationships.
The article proceeds with a section that positions our study in relation to previous research, and outlines key research gaps and our theoretical points of departure.Next, we present our method, case study sites and data collection.The main analytical section outlines and discusses four main themes that emerge from our respondents' accounts of police-community relations in the two settlements.A final concluding section brings out the findings from our study, and discusses these in relation to the existing work on police-community relations in Kenya and beyond.

An ambiguous relationship between community policing, trust, and communication
Community policing policies emerged against a backdrop of a confidence crisis in the police, and its practices and policies were initially developed in the USA and the UK. 17 As part of police reform, community policing aims to transform the police force into a police service that conducts policing in cooperation with the people rather than as an externally imposed security force.Community policing encompasses a range of activities and actors, and there are important differences in how community policing has been implemented in the USA and UK (and other Western states): US policing has been under local control and community policing as a result carried out locally with great variation, while community policing in the UK has been part of a broader crime prevention strategy.Yet, the similarities are enough for scholars and practitioners to refer to a 'Western model' of community policing, which with varying success has been adopted in the Global South. 18e understand community policing as state-driven policing aiming to identify and address crime and security issues through interactions and partnering with local communities. 19Typically, community policing initiatives establish formal structures for police-community partnerships. 20e conceptually distinguish community policing from security provision by non-state actors, such as private security companies and youth groups, but note how it often operates in a context of multiple security providers, where a wide set of non-state and state actors provide public security. 21ommunity policing aims to improve public perceptions of the police and increase effectiveness in curbing crime.However, research on the effects of community policing as a strategy and as reform has yielded mixed results 22 and highlights several challenges with the implementation of community policing, including communities that are overtly resentful due to fear or bad experiences with police in the past. 23xisting work on community policing has mainly focused on longstanding Western democracies (notably the USA, the UK, and Australia), while challenges and needs may be even more significant in contexts of political instability and with a history of colonial, repressive policing. 24esearch identifies lack of legitimacy, leadership commitment, high levels of inequality, and marketization of policing as key barriers to police reform in East Africa. 25A series of experimental studies from the Global South-conducted in Brazil, Colombia, Liberia, Pakistan, Philippines, and Uganda-found that community policing failed to reduce crime and did not increase citizens' cooperation with police. 26Other studies highlight that at least minimal levels of trust and cooperation are a precondition for community policing to be successful, 27 implying that it tends to be 'least effective where the need is greatest'. 28Research on policing in Kenya similarly charts the problematic nature of community policing reform, 29 its association and conflation with other security providers, 30 how colonial legacies have had a lasting impact on police-community relations, 31 and how it has 'fail [ed] to address or deliberately ignore[d] the wider political context'. 32e advance existing work on police-community relations in Africa in two ways.First, previous research on policing in Africa emphasizes the overarching problems with the police, particularly its colonial legacies and connection to macropolitical dynamics.Several studies allude to how these issues translate into challenges associated with state-driven community policing reform.Our study complements existing research by focusing on the microdynamics of how contextual and relational factors influence police-community relations.Studies addressing similar questions have identified key challenges in police work, including policing practices and organizational models shaped by colonial and war-related legacies, and the heavy reliance on policing by non-state actors. 33Research on citizens' incentives to collaborate with police primarily focuses on individual-level determinants based on survey data.Such studies have highlighted how socio-economic characteristics, as well as individual experiences of police effectiveness and fairness, shape attitudes to the police and willingness to collaborate. 34Complementing these bodies of research, our analysis provides insights into the contextual and relational dynamics of how police effectiveness and community policing shape the residents' perceptions of and collaboration with the police.
Second, our relational focus advances knowledge on how communication and trust interact in police-community relations.Community policing reforms revolve around improving communication channels by establishing an everyday presence in neighbourhoods: the police create new local 28.Rukus police posts, and the police officers patrol and talk to business owners and residents, host community meetings, sponsor citizen advisory councils, and launch information campaigns. 35In turn, communication and trust are intricately linked. 36Transparent and mutual communication increases trust, but such communication also requires trust in the first place.This creates a seemingly insurmountable dilemma in situations where mistrust is high and improved community-police relations are most needed.Previous research has identified unidirectional information flows, socio-linguistic challenges, a rationalistic approach to policing, and identity-based bias or underrepresentation in the police force as barriers to communication in socially disadvantaged neighbourhoods. 37In contexts where the police have mainly acted in the interest of dominant groups, marginalized communities are likely to fear that sharing information may put them at risk, or perceive community policing as a way for the state to extract information. 38Despite these challenges for communication and trust-building, a few studies show that interventions to improve public knowledge about the police and its function can change citizens' perceptions of the police even when initial mistrust is high, pointing to the potential for incremental or contingent trust-building. 39ur study builds on these insights.We acknowledge that trust can involve the citizens' confidence in individual police officers and the police service as an institution. 40In contexts characterized by insecurity, violence, and conflict, we do not expect trust to develop easily: contingent and limited trust is a suitable way of conceptualizing citizen trust in police-community relations in Kenya.It is trust with a limited span in terms of who is to be trusted, to what extent, and the areas to which it is extended. 41Our study thus probes the factors favouring the establishment of this minimal level of citizen trust in the police and whether community policing plays a role in these processes.

Insecurity and policing in Nakuru County: case selection and methods
Nakuru-and Kenya more broadly-is a suitable context for analysing how residents in high-insecurity and low-trust contexts relate to the police, why they sometimes choose to collaborate with the police, and the role of community policing in shaping police-community relations.Community policing has been implemented on a project basis in parts of the country since 1999, but it gained momentum when it became a core part of nationwide police reform under the 2010 Constitution.Guidelines formulated at the national level now serve as the overarching framework for organizing community policing at the local level.The reforms were much needed: nationally, the performance of the police in addressing insecurity has consistently been judged as poor, unaccountable, and unprofessional, 42 many residents in urban informal settlements feel mistreated and harassed by the police, and these dynamics plus insecurity have given rise to support for vigilantism. 43For instance, according to a study on violence in Nakuru County, 40 percent of respondents felt that the police use unnecessary force when responding to incidents of violence, and 60 percent expressed mistrust in the police due to a culture of corruption and impunity. 44n 2017, the national government released Kenya's community policing guidelines, which recognize the local community's voluntary participation in maintaining security and the need for police to be responsive to the communities.These guidelines emphasize the need for joint identification of problems and solutions, 'while respecting the different responsibilities that police and the public have in crime prevention and maintaining order'. 45s part of this strategy, CPCs are formed at the local level.These are comprised of representatives of the local community (with a civilian chairperson and a police officer as vice chairperson), who serve as a key communication channel between the police and the community.Other members of the CPC are the chief, village elders, youth, and women leaders. 46In Nakuru, the CPCs are defined by the area covered by a local police station or police post. 47Together with the police and other key stakeholders, the CPCs convene broader Community Policing Forums to 'identify and solve problems in their areas'. 48ithin Nakuru County (Figure 1), we selected two urban settlements where CPCs have been implemented to improve security and policecommunity relations.Kaptembwo is in Nakuru West Constituency in Nakuru city, with most of the 79,480 49 residents belonging to Kalenjin, Luo, and Luhya ethnic communities.The informal settlement is close to Nakuru Central Business District (see Figure 2), where many residents work. 50Karagita is one of the informal settlements in Naivasha, a town  on Lake Naivasha centred on tourism and the floriculture industry, where fishing is an important source of income.It is administratively located in Hellsgate Ward, Mirera sublocation (see Figure 3).Kaptembwo and Karagita are low-income, informal settlements with poor housing conditions and inadequate drainage, water and sanitation systems, access roads, and street lighting.They have also been settings of post-election violence in the past and were identified as potential 'hotspots' ahead of the 2022 national elections. 51In the 2007/08 post-election violence, youth from different ethnic communities were split along party lines and attacked each other. 52In Kaptembwo, the Mungiki militia 53 51.Mary Wambui, 'Kenya: NCIC warns five major towns possible poll violence hotspots', Daily Nation, 17 February 2021; Kirera Mwiti, 'Security agencies map out violence spots in Naivasha', People Daily, 31 July 2021.52.Ethnically-based political coalitions are typically formed to win elections, as no group is large enough to gain power on their own.These ethnic coalitions have shifted over time, but national politics has generally been perceived by Kenyans as a rivalry between Kikuyu, Kalenjin, Luhya, Luo and Coastal groups.See Daniel N. Posner, 'Regime change and ethnic cleavages in Africa,' Comparative Political Studies, 40, 11 (2007)  was implicated in displacing residents who were not from the Kikuyu ethnic group and forcefully circumcising and killing men of the Luo ethnic group. 54Kikuyus living in Naivasha engaged in retaliatory violence against mostly Luos, Luhyas, and Kalenjins (who had attacked Kikuyus in different parts of the country) to expel them from their rented residences. 55ue to the volatility of the area, Karagita has become a 'transient space' for labour migrants, whose experiences with insecurity have kept them highly mobile. 56

Methods and empirical material
Our analysis relies on extensive qualitative data collected in Karagita and Kaptembwo informal settlements.We set out to understand the police's efforts in mitigating insecurity, the functioning of community policing structures, residents' perceived security challenges, and their relationship with police and other security providers.We conducted key informant interviews (KIIs) and focus group discussions (FGDs) in three phases: June, July/August, and October/November 2022 (i.e.before and after the 9 August general elections).In collaboration with local research assistants, the research team conducted 25 focus groups and 37 KIIs with 240 research participants.A full list of interviews and focus groups (pseudonymized) is provided as Supplementary Material in an Online Appendix, which also contains more details about the field research, the empirical material, and the data collection.
In brief, we used a semi-structured guide in English and/or Kiswahili to conduct the interviews and focus group discussions.Research participants were recruited through a snowball approach with multiple entry points to cover different perspectives.Identity dimensions such as age and gender shape the type of interactions people have with police, so we ensured the inclusion of both men and women, as well as youth and elders.The majority (at least 75 percent) of respondents were local residents, but we also interviewed key stakeholders such as police and local government representatives, community leaders, and civil society representatives working in the settlements.Thirty-seven of the respondents were active or previous members of CPCs, and these individuals lived in the local community.
We took several steps to ensure the safety and well-being of the research participants.We discussed ethical challenges and security risks during the preparation and conduct of the study.Interviews were only conducted based on informed consent and we obtained ethical and research clearance from the Swedish Ethical Review Authority (2019-03777) and the National Commission for Science, Technology and Innovation in Kenya (NACOSTI/P/22/16,840).All interviews were transcribed, manually coded, and thematically analysed using NVivo software.We identified all instances in the material where residents and security actors described general perceptions of and relations with the police, changes over time, and instances of collaboration and confrontation.Based on this analysis, we inductively identified the main reasons residents described for collaborating (or not) with the police and the role of community policing therein.

The microdynamics of police-community relations in Karagita and Kaptembwo
This section lays out four key themes that emerge from our respondents' accounts of police-community relations in the two settlements.The first theme provides a general overview of the main security challenges local residents identified in the respective area and the reasons for mistrust in the police.For the next three themes, we analyse the rationale residents of Karagita and Kaptembwo offer for their collaboration, or their noncollaboration, with the police.The primary themes concern perceptions about police (in)effectiveness, how communication and trust interact in Downloaded from https://academic.oup.com/afraf/advance-article/doi/10.1093/afraf/adae018/7753529 by Dennis Hejhal user on 10 September 2024 police-community relations, and the importance of personal interactions and individualized trust-building.

Police-community relations: the challenges
A first theme in the empirical material identifies the security challenges of each settlement and why police-community relations were fraught.Kaptembwo and Karagita both faced high crime and related insecurities at the time of our data collection, but the dynamics differed.In Karagita, residents reported high levels of interpersonal violence, sexual and gender-based violence, and violence against children.While these forms of insecurities were also common in Kaptembwo, the settlement is notable for the presence of organized criminal gangs, such as Confirm and Gaza, whose members have committed crimes, including rape and homicide. 57Residents in Karagita also mentioned the presence of gangs, but not to the same extent as in Kaptembwo. 58In both settlements, respondents narrated being victims of muggings, stabbings, and house break-ins by criminals operating in the area. 59In a 2017 survey, 36 percent of respondents in Karagita noted that violent crime was prevalent in the settlement, while the figure in Kaptembwo was even higher, 55 percent.For political and ethnic violence, the survey results were similar in the two areas (45 percent in Karagita and 44 percent in Kaptembwo area). 60But, as we will return to, in 2022, the political dimensions of crime and insecurity were more pronounced for Kaptembwo than in Karagita.
Generally, residents in Kaptembwo and Karagita perceived the police as corrupt and inept.For instance, respondents reported that police misuse their powers in order to harass, extort, and threaten citizens.Allegedly, some police officers had been involved in extra-judicial killings, arbitrary arrests, and leaking of confidential information to criminal suspects. 61 frequently cited issue was police profiling of mainly young men in 57.Kaptembwo 4 FGD (community leaders); Kaptembwo 6 FGD (boda boda); Kaptembwo 16 KII (community leader); see also Wairuri, Chemlali, and Ruteere, Urban violence in Nakuru County, Kenya. 58.E.g.Karagita 5 FGD (youth).59.See also Judy Ngina, Gender implications of household inaccessibility to safe water and improved sanitation in Kaptembwo, Nakuru County (Egerton University, unpublished MSc dissertation, 2017).60.The survey report does not define 'political' and 'ethnic', but notes that in the Kenyan context, politics and ethnic mobilization are intertwined, and polical and ethnic violence often escaltes during election periods and politically contententious moments.The question posed in the survey was: 'How prevalent is political and ethnic violence in your area?'.See Wairuri low-income areas as criminals. 62The profiling, common in similar neighbourhoods in Nairobi, 63 made it difficult for youth to share information and interact with the police.Many residents who were not youth also reported how the profiling made them hesitant to collaborate with police. 64hile our respondents observed some improvements in policecommunity relations, many did not personally trust the police. 65A boda boda (motorcycle taxi) driver in Karagita illustrated this contradiction: 'I see a good relationship.Police do their job.They also give us jobs.They don't mistreat anyone.I have never seen the police mistreat anyone'.Yet the boda boda driver also stated that although he had been robbed, he had not taken the case to the police. 66A resident of Kaptembwo, when asked about police-community relations, asserted: 'A sane citizen in this area cannot positively relate to the police'. 67Respondents from both settlements raised concerns that police upheld security unevenly, harassed citizens to collect bribes, and colluded with criminal actors. 68hile respondents alluded to a mix of possible motives for alleged police collaboration with criminals, such perceptions clearly made residents fearful that sharing information with the police would put them at direct risk.However, the dynamics varied across the two locations, with implications for how people communicated with the police.Comparatively, respondents in Kaptembwo to a greater extent cited political dynamics as a reason that information would be compromised, whereas residents in Karagita placed more emphasis on inefficiency and corruption. 69When asked explicitly if politicians interfered with police work, respondents in Karagita said no. 70 In Kaptembwo, in contrast, there was a strong perception that an influential local politician was linked to and was protecting members of particular gangs. 71A village elder in Kaptembwo reflected: We have no guarantee that the confidential information we share cannot be used against us.If someone witnesses a crime/incident, they would rather walk away from the scene or ask someone else to go and report it because there is no guarantee of their safety … In Nakuru West, as things are, you cannot go and report an incident to any office and trust that you will be safe.The area MP [Member of Parliament] has a huge network of dangerous people, and you could be reporting to one of his people, so we rather not take the risk. 72spite entrenched mistrust, there is variation in the data as to whether respondents would consider turning to the police and how they perceived such interactions.Below, we discuss how these dynamics are linked to police effectiveness, communication and personal relationships with police.

Perceptions of police (in)effectiveness
A second theme in the empirical material helps us to understand how police effectiveness and ineffectiveness shape residents' willingness to interact and collaborate with the police.In both settlements, there was a widely held perception of the police as ineffective, which many respondents cited as a primary reason for not reporting suspected criminal activity. 73The police responded slowly, failed to make arrests, or only responded if given financial incentives (bribes), according to respondents.In the words of one boda boda driver in Kaptembwo: 'It doesn't make sense to go to the cops to report robbery of Ksh [Kenyan Shillings] 3000 [about $23 USD] and give a similar bribe for them to start taking action'. 74In addition, respondents in both settlements claimed that police officers often leak confidential information to suspected criminals, enabling suspects to 71.Kaptembwo 3 FGD (community members); Kaptembwo 5 FGD (community members); Kaptembwo 6 FGD (boda boda); Kaptembwo 9 FGD (community members/leaders); Kaptembwo 12 FGD (community leaders).72.Kaptembwo 3 FGD (community members).73.Kaptembwo 5 FGD (community members); Kaptembwo 6 FGD (boda boda); Kaptembwo 7 FGD (community leaders); Kaptembwo 10 FGD (boda boda); Kaptembwo 11 FGD (community members); Kaptembwo 12 FGD (community leaders); Kaptembwo 14 KII (community youth leader); Karagita 2 FGD (community members); Karagita 3 FGD (boda boda); Karagita 4 FGD (peace committee); Karagita 5 FGD (youth); Karagita  elude apprehension. 75While inefficiency was variously ascribed to corruption, laziness, or the police being under-resourced, the implication was that many respondents saw little point in reporting incidents.Against this backdrop, however, some respondents cited improvements in police effectiveness as a reason for collaborating with them or a change that led them to form a more positive perception of the police.Notably, in Kaptembwo, upgrading of the police station in 2019/2020, from a wooden structure to a permanent building enclosed in a perimeter wall, was considered to have made certain forms of issues easier to report, such as gender-related crimes. 76Before, the gender desk was under a tree, which meant there was no privacy for the victims to report such crimes.A community leader in Kaptembwo describes: 'Now, there are more police officers deployed there; they have a good gender desk, no longer held out in the open as before.It is no longer as congested.They have created some trust between the community and police'. 77Community members also reported that crime had gone down due to deliberate efforts and collaboration between the community and the police. 78A member of the Karagita CPC underlined how concrete results in policing strengthened continued collaboration: There are so many drugs in the area that are abused.Through our joint effort with the police, chang'aa [illicit brew] dens have been raided, and the sellers stopped.Openly smoking marijuana has halted, and the number of young men who dropped out of school has reduced.Many young men and teenagers of school-going age are loitering around here, sometimes getting mixed up with the wrong company.Still, we reported them to the police, who then worked hand in hand with their guardians/parents and local school principals to ensure they were back to class. 79milarly, a young boda boda driver in Karagita noted: There was a particular road that was notorious for mugging and pickpocketing.… We reported this to the CPC, who collaborated with the area 75.Kaptembwo 5 FGD (community members); Kaptembwo 6 FGD (boda boda); Kaptembwo 7 FGD (community leaders); Kaptembwo 9 FGD (community members/leaders); Kaptembwo 12 FGD (community leaders); Karagita 4 FGD (peace committee); Karagita 25 KII (community leader).76.Kaptembwo 20 KII (local police officer); Kaptembwo 24 KII (CSO representative); Kaptembwo 26 KII (local police official).The CPC allegedly played a central role in putting together the proposal for such upgrading and lobbying politicians and NGOs for the funding, which came from the Nakuru West Constituency Development Fund (Kaptembwo 22 KII (community leader/CPC)).77.Kaptembwo 24 KII (CSO representative).78.Kaptembwo 21 KII (CSO representative); Karagita 1 FGD (CPC committee); Karagita inspector and police.As we speak, that is the most secure road around.We work late; therefore, safety is an essential feature of our work.A year ago, the situation was so bad that it was impossible to work late at night or in the wee hours of the morning because of the danger the road presented. 80 several of the interviews and focus groups we held after the general elections in August 2022, the police were identified as a key actor in upholding security during the election period, thereby improving locals' trust in them.Given the history of electoral violence in the settlements, elections are critical moments when residents form perceptions about the security actors' effectiveness or ineffectiveness.Respondents described how the police were collaborating constructively with other security actors, were visible within the settlements, and kept calm without an aggressive attitude toward residents. 81For instance, a resident of Karagita said: The police were perfect-no problem this time.They were patrolling and acted very peacefully.This is another thing that reassured us. 82spondents also alluded to how the police had worked more proactively in the 2022 election, started the preventive measures earlier, and also had been present to calm the political campaigns. 83

Communication and the importance of community policing in promoting collaboration
A third theme in the empirical material is how community policing has improved communication.Several respondents who held that the relationship between the community and the police had improved attributed this change at least partly to how community policing enhanced information flows and mutual understanding. 84This narrative was recurrent among security officials and civil society organizations (CSOs) actively promoting police-community relations, who claimed community policing had improved cohesion, collaboration, and reporting of crimes to police. 85here were also numerous examples where residents from both settlements gave similar descriptions of changing dynamics and attributed improvements to community policing. 86For instance, when asked if relations with police were good, a resident of Karagita, responded: Respondent: Oh yes, not long ago, when someone saw the police, they would run; nowadays, you can see them mingling and talking to the police, they are friends … We treat them like one of us now.
Interviewer: Why so?Is it that they have done a lot of meetings?Have you attended any?
Respondent: Yes, a lot of meetings with CPC have been done.I attended one, I found it very good, it was for IPOA [The Independent Policing Oversight Authority] and I witnessed how the police are trying to relate with the common man.I found it good because it was engaging everyone.… Interviewer: have you seen any changes that have led to increase of security?
Respondent: It's like that one example for IPOA, that relationship, and those ones for CPC where even residents are allowed to speak about their challenges. 87veral CPC members also alluded to how their relationship with the police improved over time: 'At first we did not have a good relation between us, the police and the community.We were just there in between … It was not a good relationship, but when it turned to be a policing service, it's better'. 88Another CPC member said: 'In a period of a decade, the relation has changed and is, let's say, 5 of 10 on a scale.There's still work to do'. 89rucially, CPCs and their members had become a communication node between residents and police, allowing residents to transmit information confidentially.This is a key in a context where many residents fear the negative consequences of engaging and sharing information with the police, as described above.In the words of one village elder in Karagita: 'The people report cases to them, and the CPC follows up and reports to the police.Reporting crimes to the police can be risky, and many people don't want to get themselves in danger.CPC has bridged this gap by being the face of crime reporting in the community, and this is a role much respected and appreciated'. 90Notably, respondents from both Karagita and Kaptembwo indicated that the CPCs had facilitated the reporting of cases of genderbased violence. 91However, in both locations, several respondents were unsure about who the CPC members were or if they were operating in the settlement. 92Nonetheless, local administrators and police officers claim that CPCs have improved the information flow, assisting the police in preventing crimes. 93everal respondents also indicated that CPCs have facilitated collaboration among a broader set of security actors on the local level.In addition to the police, the area chief (local government appointee) is responsible for maintaining order and preventing crime. 94Another community security structure, Nyumba Kumi (Swahili for ten houses), is also commonly referred to as a form of community policing.It operates under the National Government Administration Officers docket of the president's office and was launched to enhance information-sharing on crime, especially with the chiefs, who in turn report to the police. 95Nyumba Kumi is separate from the CPC structure, which operates under the police but has a similar function to improve collaboration at the neighbourhood level. 96Chairpersons of Nyumba Kumi blocks are expected to sit in the CPC. 97The chiefs also work with the village elders, who are frequently members of the Nyumba Kumi structure and/or the CPCs. 98n both settlements, many respondents indicated that collaboration among the security-related actors-CPC, Nyumba Kumi, and the chiefoverall worked well. 99 structures.… I've seen them working harmoniously'. 100However, local community leaders, Nyumba Kumi and CPC members also allude to tensions between these structures. 101A CPC member in Karagita pointed to the role of the local chief in facilitating or obstructing cooperation; '[a former chief] could not let Nyumba Kumi and CPC work together, creating a rift between us.He often said he is to deal with the Nyumba Kumi business, and CPC is to work with the OCS.We often collaborate for the good of the community, but the chief would undermine our work.' 102A similar dynamic was also reported in Kaptembwo. 103These dynamics suggest that if local authority figures like the chiefs feel undermined by the introduction of the CPC, they may hamper collaborative relationships between different security structures.As reflected in other work on Kenya, the impact of community policing structures will be shaped by the pre-existing security and community conflict management structures. 104

Personal interactions, individualized trust-building, and police rotation
A fourth theme from the empirical analysis relates to how trust-building between police and the local communities, to a large extent revolved around residents' relationships with individual officers.To compensate for institutional inefficiencies, residents of Kaptembwo and Karagita established personal contacts with officers as guarantors of security.These strategies suggest that trust was built at an individual rather than an institutional level.A resident of Karagita (a local fisherman, who was also an elder in the community and CPC member) described: 'In general, there is low trust in the police, but people do trust specific police officers.Of the ten police they have at the police post-I trust two'. 105While police officers previously lived in police housing inside the police compound, now more officers reside within communities.Respondents in both Karagita and Kaptembwo cited this change as important for relation-building: Previously people couldn't talk to a police officer.It is not perfect but at least we have seen change.In the new system, police can live within the community-this encourages people to feel that they can report.Before they would only live in the compound. 106ainst this background, one issue that complicates efforts to improve relationships between citizens and security actors is the frequent transfer of police and administrative officers.On the one hand, frequent rotation of officers is intended to prevent corruption, and several respondents argued that officers should be transferred often not to become integrated into local criminal networks. 107On the other hand, transfers undermine relationshipbuilding with CPCs and ordinary residents and the networks that have been established.As stated by a community leader in Karagita: 'There are police officers who used to work in Karagita and made regular patrols, especially at night.Those who attend night keshas [night vigil prayers] appreciated this because we could stay out late.Then there was a mass transfer; some relocated, and others retired.The new batch of officers who came are very lax and lenient'. 108When the officers operating in a specific locality were transferred, residents found it difficult to anticipate what kind of policing would prevail in their next encounter with the police.Furthermore, some respondents perceived it was mainly the officers doing good work that were quickly transferred: 'Political interference prevents the police from doing what they are supposed to do.They are strong-armed to release criminals under instructions from politicians, and failure to oblige will lead to them getting transferred'. 109he transfer of police leadership is of particular consequence for the work of the CPCs.In Karagita, the transfer of the Officer Commanding Station (OCS) was highlighted as a challenge, as one CPC member highlighted: Yes, we have the CPC, but each time the OCS is transferred we have to start over.They do things differently rather than pick up where the other left off.So the reform takes too slow.When OCS is brought from other parts of the country they may not be engaged, just do what they have to do to get the salary.Might not at all be engaged with the communities.… When transfers come and go the progress gets stalled. 110long the same lines, another CPC member stated: 'When one is transferred, things go back.… When they don't know me, they can't get good information'. 111In contrast, the OCS in Kaptembwo had been stationed there unusually long.In 2022, the OCS had stayed in the station for at least 6 years.According to many respondents, the Officer was seen as too closely affiliated with, and also intimidated by, local politicians.The closeness with politicians was also what some respondents believed was keeping the OCS in place, as alluded to by a FGD participant: 'The OCS, however, is working with [the MP].She has been here since 2014 and has three officers that she walks with.Whenever there are transfer letters for these four people to leave, [the MP] blocks them'. 112While CPC members we interviewed agreed that the Kaptembwo OCS area might have overstayed, they also noted that she had organized the community policing structures, and had been instrumental in ensuring the construction of a new police station with better facilities and private space for police matters. 113verall, these dynamics underline the dilemma concerning officer rotation.While many residents perceive such rotation as necessary to prevent corruption, our observations also suggest that the transfers may undermine improvements in police-community relations, where the intermediary channel of the community policing structures plays a constructive role.Thus, despite the apparent autonomy at the street level, the success of community policing could be disrupted higher up in the hierarchy and by national-level political and bureaucratic processes.

Concluding discussion
Police-community relations face considerable challenges in locations fraught with insecurity, a history of authoritarian rule and politicised policing.The relational perspective, which centres local residents' interactions with police and community policing structures, offers important insights into the challenges of policing reform and how contingent trust can be built.Our study has centred the experiences of residents of Karagita and Kaptembwo, and the contextual and relational factors that shape their decisions to collaborate with the police or not.The analysis yields several insights of relevance for the literature on police-community relations in low-income urban areas.
An important overarching finding is that community trust in the police remained fragile.Like previous studies in Kenya, fear was a key consideration: residents feared that sharing information with the police would put them at risk. 114However, despite a challenging context and mistrust in police, we found evidence of some progress in police-community relationships.Perceptions of improved police effectiveness and increased local security, and experiences of police preserving security without mistreating locals, were primary motivations for why residents sometimes chose to collaborate with and formed more positive perceptions of the police.However, the nature of these dynamics creates a dilemma in that much trust was vested in individual police officers and not the police as an institution.This leaves police-community relationships vulnerable to staff rotations, which otherwise is an important mechanism for preventing corruption.Like existing work on community policing, our study identifies turnover in police officers tasked to carry out community policing as problematic because time is needed to establish the required rapport between local communities and local leaders to get the relationship to work. 115e also uncover some ways that community policing shapes these dynamics and inform research on how communication between communities and police can be improved.In particular, the CPCs and their members acted as transmitters of information when mistrust hampered direct interaction between the police and the community.If such information-sharing also leads to improved police effectiveness, as our respondents indicated, it can gradually build trust.This insight resonates with research from Ghana, which found that assessment of police effectiveness was the most important factor shaping urban residents' willingness to cooperate with police. 116Similarly, the community policing reform, at least in part, helped to improve everyday interactions with the police and reduced perceptions of them as merely repressive, aspects that our analysis suggests are a key to trust-building.Overall, our observations align with Spalek's conclusion that in contexts of high mistrust, community policing reforms must initially build 'contingent trust' through reciprocal relationships and demonstrating that community concerns are considered and acted upon. 117ur attention to how everyday interactions matter resonates with findings from Johannesburg, where people's practices in dealing with the police originate from their daily engagement with specific police officers. 118Such insights underline the important agency of individual police officers.In Kenya, despite the predominant focus on police brutality and misconduct, studies also show that many police officers have undertaken their duties during election crises and saved countless lives. 119Research also underlines that police officers 'are more than political tools' and have multiple identities and interests (ethnic, social, economic, and political) that shape how they exercise power in everyday interactions with citizens. 120lthough research has shown how expectations of police behaviour and trust often follow ethnic lines, 121 ethnic bias was not a prominent theme in our interviews.However, respondents emphasized local political dynamics, particularly in Kaptembwo, where collusion with criminal gangs is perceived to run high up the political hierarchy, affecting perceptions of police and impeding trust-building and accountability.This dimension was particularly pronounced in the run-up to the election and less so in the interviews conducted after the election, likely because the August 2022 post-election period was relatively peaceful in Nakuru County, specifically, and in Kenya, generally. 122e expect the findings from the two settlements in two of Nakuru's urban centres-Nakuru City and Naivasha-to be relevant across Kenya, but most so in areas which similarly are urban, cosmopolitan, low-income and densely populated.In particular, we expect police operations and police interactions with local communities to be different in rural areas.The security challenges manifesting in rural areas are partly different and in the Kenyan context involve, for example cattle theft, banditry, and land tenure conflict.In some areas of the country, the terrorist threat from Al-Shabaab has been prominent.The prospects of effective policing and local resident's everyday interactions with the police will also be different in rural areas, where police cover very large jurisdictions and geographical distances.In rural settings, formal policing is also relatively less important compared to urban areas, because other institutions for managing local security exist, are strong and have authority, including traditional and religious institutions, to deal with local dispute resolution, insecurity, and crime.
To conclude, our study highlights how community policing can lead to incremental improvements in police-community relations.We underline that such improvements remain small and fragile: structural and historical conditions continue to hamper police reform and imply that mistrust in police among residents in urban settlements like Kaptembwo and Karagita

Figure 1
Figure 1 Nakuru County is located northwest of Nairobi.Kaptembwo is in Nakuru city, and Karagita is in Naivasha town.Map created in QGIS based on Open Street Map.

Figure 2
Figure 2 Kaptembwo ward within Nakuru City.Map created in QGIS based on Open Street Map.

Figure 3
Figure 3 Karagita in Hells Gate ward, on the southern fringe of Naivasha town.Map created in QGIS based on Open Street Map.
A local government official stated: 'In other places where I have worked, [different community security structures] had differences.Here, they do not.… Some of the members are also in all the