Policing “Landscapes” for the Rule of Law and Public Protection: the State of Evidence on Organisational Policies, Structures, and Human Resources

What evidence is, or is not, available on organisational policies, structures, and practices in national policing institutional landscapes that enable a police service to promote the rule of law and protect the population from crime, disorder, and violations of human rights? Qualitative and quantitative research on police organisational policies, structures, and practices in areas such as human resources and recruiting, legal authorities, and reporting lines to governance. A review of the evidence in landscapes associated with higher or lower levels of indications of a rule of law and protection from harm across a range of high-, medium-, and low-income countries and across different regions. The paper seeks to assess the strengths, limitations, and gaps in the evidence and data for understanding this association. The evidence for institutions promoting better policing at the level of nation-states is largely qualitative or historical. Useful frameworks for different models of police landscapes have been developed, with some tentative conclusions observed from within-country studies. These include the negative impact on the rule of law of “militarizing” civilian police forces, the positive impact of checks and balances of external auditing by independent agencies, and the promising effects of well-implemented policies for community policing and body-worn video cameras. Yet all institutional landscapes are embedded in unique cultural traditions, and no one framework for police institutions is likely to work in most countries. Research advances over the past half-century have substantially improved scientific understanding of police institutions. Nations investing in those institutions have a direct stake in advancing that knowledge as part and parcel of an effective strategy for promoting police protection of the public, as well as its rule of law.


Introduction
This essay provides a review of the evidence on how national institutional "landscapes" for policing may affect the strength of either rule of law (ROL) or successful police protection of the public. It draws on published analyses drawn from a range of high-, medium-, and low-income countries and across different regions.
The purpose of this paper is to assess the strengths, limitations, and gaps in the evidence and data for understanding the effects of these landscapes. It also uses that assessment to make recommendations on the research designs, data, and tracking that international development organisations, such as the U.S. State Department, could use to guide and sustain their support for future development of the rule of law in democratic countries.
The topic does not easily lend itself to systematic review. A number of the areas covered have either not been explored in high-quality studies or present methodological challenges which may preclude such studies (such as judgments about the relative merits of different governance models). Where there are questions of effectiveness, the essay will have a preference for systematic review and experimental or quasi-experimental evidence. Yet evidence that can be drawn from those more rigorous research designs is sorely lacking for many issues and countries.

Definitions and Frameworks for Rule of Law and Police Reform
The United Nations (UN) provides the following definition of the "rule of law" (ROL): A principle of governance in which all persons, institutions, and entities, public, and private, including the state itself, are accountable to [domestic] laws that are publicly promulgated, equally enforced, and independently adjudicated, that are consistent with international human rights norms and standards.

3
A more detailed definition has recently been published by the Bureau of International Narcotics and Law Enforcement (INL). 1 That definition provides some apparent clarity for a contested and complex idea. The modern concept of ROL is usually credited to Dicey, whose formulation linked it inextricably to parliamentary sovereignty (Loughlin, 2010). This approach guided the UK's 1962 Royal Commission on the Police, which asserted that.
the proper criterion [for whether a state is a police state or not] is whether the police are answerable to the law and, ultimately, to a democratically elected Parliament (The Royal Commission on Criminal Procedure, 1962). Lord Bingham, a highly respected English judge, emphasised the need for accessible, clear law, applied equally to all and in compliance with international human rights (Bingham, 2007).
In seeking to turn ROL principles into a working tool, the US Agency for International Development (USAID) (U.S. Agency for International Development, 2008) have focused on a set of "essential elements," starting with the provision of order and security and underpinning this with legitimacy and fairness within a system of checks and balances (see Table 1): Each of USAID's essential elements requires a complex judgment about how to measure and how to relate them to the others. An attempt at this is provided by the World Justice Project (WJP) Rule of Law Index, 2 which utilises a slightly different set of factors sub-divided into criteria linked to a scale (see Table 2) designed to allow ROL comparisons between different jurisdictions. □Protection of basic human rights and civil liberties □Access to justice 1 "A principle of governance in which all persons, institutions, and entities, public and private, including the state itself, are accountable to laws that are publicly promulgated, equally enforced, and independently adjudicated that are consistent with international human rights norms and standards. It requires measures to ensure adherence to the principles of supremacy of law, equality before the law, accountability to the law, fairness in the application of the law, separation of powers, participation in decisionmaking, legal certainty, avoidance of arbitrariness, and procedural and legal transparency." (p. 44 There is a strong link between these ROL frameworks and models of police reform. In developing one of the most important models, Bayley (Bayley, 2006) drew on experience, observation, research, and case studies of international police reform. His starting point was the legal framework for the police and for their independent oversight. These were then supported by four themes: recruiting and retaining the right sort of people; developing the capacity of police executives to manage reform; making the prevention of crime as it affects individuals the primary focus of policing; and requiring legality and fairness. He found that police reform (and by implication the rule of law) was rarely, if ever, enhanced by focusing on reorganising, equipping, or training the police.
Bayley (Bayley, 2008) argued that his model had been substantially successful in the reform of the police in Northern Ireland-a judgment which was supported by Ellison's (Ellison, 2007) stocktake on progress to reform in Northern Ireland.
An alternative and complementary approach to Bayley has been adopted by Blair (Blair, 2021) in analysing the effectiveness of UN peacekeeping and law enforcement. Blair suggested "neutralising factors," which he associated with more effective outcomes: education and competencies of key leaders and managers in law enforcement; incentives and inducement such as funding for salaries Criminal justice An effective criminal justice system is a key aspect of the rule of law and this should take into consideration the entire system, including police, lawyers, prosecutors, judges, and prison officers 1 3 and infrastructure; beliefs and persuasion to shift mindsets in favour of reform; and accountability and oversight. Running through a number of these frameworks is a focus on legitimacy and procedural justice. Tyler's work has been influential in suggesting that there are links between citizen perceptions of legitimacy and compliance with the law (Tyler, 1990). The effectiveness of the mechanism of procedural justice proposed by Tyler has been questioned by some recent studies (Nagin & Telep, 2017). Yet it seems clear that an over reliance on the lawfulness of police powers, rather than paying attention to the way in which they are exercised, is likely to harm legitimacy (Meares & Neyroud, 2015) and therefore harm the ROL.
Bottoms and Tankebe's (Bottoms & Tankebe, 2012) "dialogic model" of legitimacy is helpful in bringing the apparently discrete elements of ROL together. The model positions many of the elements, including procedural justice and equality, within a constant dialogue between the power-holders -in this case the policeand the audience -the citizens and civil society. Bottoms and Tankebe drew on Crank and Langworthy's (Crank & Langworthy, 1992) application of institutional theory to policing, in order to emphasise the importance of the network of institutional sovereignty influencing policing.
Moreover, in the dialogue with institutional sovereigns, both Bottoms and Tankebe and Herbert (Herbert, 2006) suggest that police leaders need to balance subservience (complying with the priorities and policies of institutional sovereigns) with separation (maintaining independence in exercising discretion and decisionmaking). As such, the dialogic model reinforces the idea that "Checks and Balances," as a prominent feature of most ROL models, should not be viewed as static, fixed frameworks, but rather as part of a dynamic process of dialogue.
The dynamic process is particularly important in policing because of rapid changes in the challenges facing policing. den Boer (Boer, 2018) argues for the need to adopt a "living law" perspective as legal frameworks prove to be "rigid and immune to change," so policing systems have seen "hard law" substituted by "soft law" such as regulations, protocols, and rules of conduct.

Organisational Purposes, Structures, and Accountability
Comparative research on police organisation, its structures, and purposes has been developing as a discipline since Bayley's seminal comparisons of European and International police models (Bayley, 1975). Mawby (Mawby, 2018) proposes scholars focus on three areas: the historic and present basis for legitimacy; the structure for law enforcement and community safety; and the function, which can vary on a continuum between a strong focus on maintaining order to crime prevention and resolution of community problems. Mawby also suggested that it is important to understand the different legal frameworks. He distinguished seven types, including common law, civil law, and indigenous and religious-based laws. Mawby's analysis suggests that sustaining effective ROL interventions may depend on understanding the historical and embedded legal cultures and assumptions.
Emsley (Emsley, 1999), analysing the historical development of policing models, identified three broad types: state military, in which police were armed, stationed in barracks, and responsible to a central authority; state civilian, in which the police were independent of local authority but clearly distinguished from the military; and municipal civilian, recruited locally and largely under local control. He suggested that the boundaries between these types could be "fuzzy." However, a tendency towards one model or the other had implications for the ethos and purpose of the police:  (Reiner, 2010) in which local militia had used military force to disperse an unarmed public meeting.) Emsley's typologies and their implications for ROL remain relevant in the twentyfirst century. Weintraub and Blair have explored the renewed militarisation of street policing in South America in their experimental evaluation in Colombia (Blair & Weintraub, 2020). They suggest evidence for increased human rights abuses by both police and military, with no significant reductions in crime.
Flores-Macias and Zarkin (Flores-Macías & Zarkin, 2019) go further to argue that the militarization of the police in Latin America has undermined "citizen security, human rights, police reform and the legal order" (Flores-Macías & Zarkin, 2019).
Emsley's framework needs to be seen against the way that police organisations have developed and adapted to changing challenges. Brodeur (Brodeur, 2010) made a distinction between "high" and "low" policing: the expansion of "high policing" to tackle perceived threats from serious and organised crime and terrorism has presented very different challenges for oversight. The nature of the work frequently requires openness and transparency to be muted to protect operations.
Bowling and Sheptycki (Bowling & Sheptycki, 2015) have identified the complexity and, at times, fragility of the accountability networks for transnational policing. "High policing," transnational policing, and the policing of terrorism also tend to require a close relationship, if not a substantial overlap between police and security services. The scale of the potential interference in human rights, the consequences of failure, and the controversies around strategies attempting to de-radicalise individuals all create major gap in the landscape for ROL. Yet the evidence on accountability and oversight of high policing is weak.
One potential strategy to improve this has been opened up by the Campbell Collaboration/5RD programme 3 of systematic reviews on terrorism and radicalisation prevention (DHS Science & Technology Directorate, 2018). Letting the light of evidence (or the absence of evidence) shine on the strategies being implemented offers an opportunity for Ombudsmen and oversight scrutiny of, at least, the effectiveness of prevention.
Bayley developed a further important typology based on two dimensions of analysis: the centralisation of command and the number of commands. He suggested that these are often confused by an imprecise use of the concepts of centralisation and decentralisation: "The point is that decentralization creates multiple forces, but multiple forces are not always decentralized" (Bayley, 1975). Bayley found that initial centralisation was often a response to the demands of maintaining order in a fragile state. More recently, across Northern Europe, there has been trend towards centralised national police forces. These changes have tended to be justified by economies of scale and a need to enhance efforts to tackle organised crime, terrorism, and transnational crimes (Terpstra & Fyfe, 2019). The evaluations -comparison of the intended benefits to the outcomes after 3 years -suggest that the reforms were "naïve, almost utopian" (Terpstra & Fyfe, 2019).
The scale of the challenges in implementing structural reforms to policing (and also by implication, reform of policing by restructuring) needs to be placed alongside the literature on the effectiveness of different sizes of police force and different internal structures. The overall message from the research reviewed in the NAS report on Fairness and Effectiveness in policing could best be summarised as "it depends" (National Research Council, 2004). However, in considering a more sustainable approach to ROL, a more decentralised model capable of supporting community policing is one model which can be linked with enhanced legitimacy and, when linked to problem-oriented policing, reductions in crime and disorder (Neyroud, 2021).
The potential significance of community policing as a sustainable strategy for ROL highlights the critical importance of the context of institutions, accountability, and oversight structures within which policing is situated. A recent report from USAID reviewing the effectiveness of police accountability mechanisms and programmes found "extremely limited data on what works in police accountability" (U.S. Agency for International Development, 2020). The report's authors emphasised the importance of considering the whole system of accountability in any evaluation and, in particular, four, inter-connected dimensions: the horizontal system of checks and balances; the vertical system -internal mechanisms, processes, and procedures; external governance; and diagonal interactions with communities.
The USAID review used this four-dimensional model to frame an analysis that proposed three "promising" initiatives in accountability.
1. The first is the most radical: Tabula Rasa or the abolition and replacement of an existing police force or unit by a new one. This is clearly highly topical in view of the abolition/defund movement which was given new impetus by the circumstances around the death of George Floyd in the USA. Two examples from Peru (Karim, 2011) and (the nation of) Georgia are supported by mixed methods research, the latter with a comparison with two neighbouring states (Kakachia & O'Shea, 2012). In both cases, the data suggested reduced corruption and enhanced performance. Neither evaluation is entirely convincing on the challenges of achieving political support, implementation, or sustaining progress. Furthermore, neither Peruvian nor Georgian examples appear to have been a complete "Tabula Rasa." 2. A further cited example, the re-formation of the Camden police (Danley, 2020), appears much more akin to the force amalgamations that formed the centrepiece of the reforms in England and Wales in the 1960s and many of the ROL benefits that may have been prompted by police-community partnerships rather than abolition. It is just such police-community partnerships that provide the most promising approach. There are a variety of models that have been tested. The most recent and extensive programme has been the EGAP Metakata programme 4 of randomised trials in six medium-and low-income countries. The preliminary reports suggest that a combination of "town hall meetings" providing a local forum for discussion and agreement of priorities and local community policing teams can improve legitimacy and crime and disorder outcomes. The approach strongly supports a community policing strategy. 3. The third area proposed is technology and, in particular, the promise of Body Worn Cameras. The last decade has seen the rapid adoption of Body Worn Cameras (BWCs) by the police. There is a growing body of international evaluation research on its effects. The 2015 President's Task Force (Cooper, 2020) provided support for BWC as a potential method of enhancing policing legitimacy. The systematic review of the use of BWCs for the Campbell Collaboration (Lum et al., 2020) included studies from North and South America, Australasia, and Europe. It concluded that BWC did not have "a consistent or significant effects on officers' use of force, arrest activities, proactive or self-initiated activities, or other measured behaviors" (Lum et al., 2020). The largest factor in determining the strength of the findings lay in the deployment policies and the discretion given by many agencies to officers as when and whether the BWC was operational. The review's authors recommended that police leaders pay closer attention to the detail of the evidence and consider limiting the discretion that officers have with BWC use. They suggested that police could use BWCs more effectively for coaching, training, or evidentiary purposes. The review's authors held out the potential for BWCs to strengthen police-citizen relationships, internal investigations, or accountability systems.
The USAID review did not discuss one of the key issues in police accountability: the relationship between political control, operational decision-making, and the risks to ROL. This issue has been central to debates in England and Wales over the 2012 adoption of a system for directly elected Police and Crime Commissioner (PCC) to be responsible for the direction of local police forces and the appointment and accountability of the Chief Officer.
Cooper (Cooper, 2020), reviewing the operation of the changes over their first two electoral cycles, found that a PCC's ability to remove their Chief Constable could cause an instability in police leadership and a potential culture of compliance, as Chiefs -many close to retirement -chose to avoid conflict with their PCC. The power for a PCC to remove a Chief Constable, which contrasted with the complex process for removing a PCC, had resulted in a concentration of power at odds with the principles of good governance -a potential failure of checks and balances. Secondly, as a result of flaws in the process of appointment, some Chiefs were becoming overly indebted to their PCC, the average length of tenure for a Chief had fallen, and the pool of candidates coming forward diminished.
Cooper's analysis raises concerns. Yet it comes in the context of a policing system which has wider checks and balances, such as an Independent Police Inspectorate and an Independent Office for Police Conduct. Similar concerns over the effectiveness of checks and balances on political control can be found in many jurisdictions, including India, with one of the world's largest policing systems.
As a recent paper for the Indian Police Foundation has highlighted, reform there has had many false starts and has been driven most strongly by the judiciary (Lakhtakia, 2020). Comparing this to Bayley's model of reform, there has yet to be sufficient attention to the legal frameworks and independent oversight. The Supreme Court in the Prakash Singh judgment recognised that there was a significant dependency between ROL outcomes and the tenure, role stability, and operational independence of Indian Police Service (senior) officers (Singh & Others v., 2011) from partisan political interference.
There is a tendency for the accountability literature to focus on external oversight and governance mechanisms, but it is also important to consider the role that police unions and staff associations can play in advocating reform. Most studies have characterised unions as a barrier to reforms, but Marks and Sklansky (Sklansky & Marks, 2008) have argued that linking union support for reforms and better working conditions with participative management may be a promising direction of travel.

Organisational Policies and Practices
Lakhtakia (Lakhtakia, 2020) argued that, in the competitive world of Indian policing, evidence-based trials in States like Rajasthan (Banerjee et al., 2021) and Madhya Pradesh (Kruks-Wisner et al., 2020) can "produce an initial success story, which can be used to build evidence, bolster momentum and influence public opinion" (Lakhtakia, 2020). Given the emphasis on police effectiveness in the models of ROL and Bayley's reform model, a commitment to evidence-based policing (EBP) as set out by Sherman (Sherman, 1998), Weisburd and Neyroud (Weisburd & Neyroud, 2011), and Lum and Koper (Lum & Koper, 2017) is a key building block.
Over the last 20 years, research has produced a growing list of proactive police strategies that reduce crime and disorder (Academies & of Sciences, Engineering, & Medicine., 2018). Many of these, such as hotspot policing, problem-oriented policing, and police-led diversion, have been systematically reviewed (Sherman et al., 2014). Whilst the majority of the studies have been carried out in the Global North in high-income countries, there are a growing number of studies in mediumand low-income countries such as Trinidad and Tobago (Neyroud, 2019) and India (Neyroud, 2021).
In the context of strategies that sustain the ROL, an EBP approach has a further key benefit. As set out by Sherman (Sherman et al., 2014), EBP is targeted on reducing the harm within the most concentrated crime and disorder locations, the harm caused by the "power few" most harmful offenders, and the harm caused to the most vulnerable victims. Focusing the best-evidenced interventions on the greatest harms should also mean a less intrusive level of intervention against lower harms. The human rights compliance required by the ROL explicitly supports this targeted approach by encouraging police to adopt a proportionate use of police powers (Neyroud, 2019;Ramshaw et al., 2019). A targeted EBP approach provides a clear contrast with the almost universal incident driven policing "3 R's" model based on response, reactive investigation, and random patrol (Sherman et al., 2014).
Over the last half a century, since team policing was first tested in the field (Sherman et al., 1973), community policing has developed from a local policing approach to an international development strategy. Originally concerned with engaging police with communities and partnering for prevention, community policing in the twentyfirst century has been stretched to encompass rebuilding law enforcement in postconflict societies and preventing terrorism and radicalisation. Yet, at heart, there remains a relatively simple formula: a shift from a reactive law enforcement model to consultation or collaboration between the police and local citizens for the purpose of defining, prioritising, and/or solving problems. That formula has been shown to work to enhance police legitimacy and combined with evidence-based approaches to problem-solving and crime reduction, can help tackle a wide range of harms, crimes, and disorder (Neyroud, 2021). The evaluation evidence on the effectiveness of community policing has been systematically reviewed using Campbell Collaboration methods by Gill et al. (Gill et al., 2014). They concluded that "our findings suggest that community-oriented policing strategies have positive effects on citizen satisfaction, perceptions of disorder, and police legitimacy, but limited effects on crime and fear of crime" (Gill et al., 2014).
The eligible studies were selected on the basis that there was a "policing strategy that involved, at minimum, some type of consultation or collaboration between the police and local citizens for the purpose of defining, prioritizing, and/or solving problems" (Gill et al., 2014). On that basis, the review provided evidence of the benefits of community policing for improving perceptions of the police. However, the authors found that there were problems with explaining the mechanisms that translated partnership and community engaged priority setting into the measured outcomes, such as improved citizen satisfaction, and in identifying the policing strategies that benefited most from community participation. A subsequent review by Colover and Quinton (Colover & Quinton, 2018) for the College of Policing focused on the components of effective community policing and was able to identify a clear framework of strong governance, leadership, accountability, and support to maintain a focus on delivery and effective implementation.
However, time and again, whether it be in high-, medium-, or low-income countries, the challenge for police and communities is not to understand the philosophy of community policing; it is the complexity of initial and, then, sustained implementation in the face of competing demands, professional resistance, and, in many contexts, corruption and poor governance systems (Neyroud, 2021).

Operational and Legal Standards for the Use of Police Powers
When the police abuse the rule of law, the potential harms include wrongful conviction of suspects, torture and physical and mental ill-treatment of suspects, insufficient protection for vulnerable suspects and witnesses, flawed identifications, disproportionate use of stop, frisk, and arrest, and illegal searches and seizure of property (Chalmers et al., 2014). One of the main policy responses to these harms has been legal regulation of police powers and, especially, the pre-charge processes.
As a result of two sets of scandals involving false confessions in England and Wales, two Royal Commissions recommended radical changes to the regulation of pre-charge investigatory processes. First, in 1981, the Royal Commission on Criminal Procedure (RCCP) (The Royal Commission on Criminal Procedure, 1981) led directly to the Police and Criminal Evidence (PACE) Act of 1984, which provided a statutory code governing all aspects of the custody and investigation process, including audio-recorded interviews. Then in 1993, the Royal Commission on Criminal Justice (RCCJ) (The Royal Commission on Criminal Justice, (1993)) recommended a requirement for video-recording and strict rules on identification parades (called "lineups" in the USA). Following RCCJ, amongst key reforms including the creation of the DNA Database, the police committed to the PEACE model for cognitive interviewing of suspects, victims, and witnesses (Memon et al., 2010).
Most of the key elements of the England-Wales PACE statute have been adopted in high-income countries across Europe, North America, and Australasia and within UN international standards for policing (Tiratelli et al., 2018), but PACE still stands out as an integrated model of regulation (Chalmers et al., 2014). Brown's (Brown, 1997) detailed Home Office review of the effectiveness of the PACE regulation of police detention and powers indicated that PACE had been largely effective in regulating the environment inside the police station, but much less so regulating what happens outside those stations.
Dixon, whose work has focused on the legal regulation of the police, endorsed that judgment with the caveat that "the question 'can law control policing?' must not be reduced to the traditional question 'can rules control discretion?'" He drew on Braithwaite's work on regulation to argue that rules can provide a "framework and a focus" but would not be sufficient on their own without understanding the "nature of the law, type of policing and the social and political contexts." To these considerations, for medium-and low-income countries, could be added the very significant investment in training, technology, and buildings required to implement regulatory frameworks like PACE.
Outside the police station, the regulation of powers such as police use of Stop and Search (Stop/Question/Frisk) has been less effective (Quinton, 2011;Tiratelli et al., 2018). Thematic reviews by the Police Inspectorate (Her Majesty's Inspectorate of Constabulary & Fire Rescue Services, 2021) and Home Office research (Tiratelli et al., 2018) has identified that a significant proportion of stops and searches did not meet the threshold test in the legislation. Even so, Murray's (Murray, 2014) analysis of the far less regulated stop and search processes in Scotland has demonstrated that codified regulation of the power is better for ROL than informal policies.
Beyond the street powers of the police, the regulation of the covert investigation and surveillance practices of the police has been a major area of controversy and complexity (Loftus, 2019;Marx, 1988). This area of policing, which has now been generalised well beyond "high policing" by the importance of communications data and the omnipresence of surveillance technologies, has produced specialist oversight bodies in a number of countries but little by way of credible evaluation. In some ways, covert policing is an area that, by the very nature of its operational requirements, happens under the radar until it goes, usually spectacularly, wrong.
Two other aspects of legal regulation may also be important in sustaining the ROL. The first is the scope of the legislation governing the role of the police. The legislation which supported the reforms to police in Northern Ireland did not just provide a definition of the role, powers, and governance of the police but also, drawing on the Patten Report (Independent Commission on Policing for Northern Ireland, 1999), framed the way that the police were intended to perform their role, embedding a partnership and community-oriented approach.
The second issue is the potential importance of codified professional practice in which international standards such as the UN Principles on the Use of Force and Firearms are set out and interpreted to guide practice (United Nations, Eighth United Nations Congress on the Prevention of Crime & the Treatment of Offenders, 1990). There has been limited research on the effectiveness of such professional guidance in areas such as police use of firearms, policing of public order, and major crime investigation. However, where such manuals or codes have been developed by (and therefore "owned" by) police experts and leadership, where they draw on tested evidence, and where they are transparently published and are reviewed and revised as lessons are learnt, they seem likely to be an important part of the framework that sustains ROL.

Leadership, Management, and Human Resources
In the model of regulation of policing proposed by the UNODC, there are three elements: • Policy input and direction setting before operations and action; • Supervision during operations; • Review and evaluation after operations and actions (Handbook on police accountability, oversight & integrity., 2011).
The model appears to place its reliance on "supervision" rather than leadership, although it does mention the importance of a "working culture that promotes transparency and evaluation" (Handbook on police accountability, oversight & integrity., 2011). Police leadership is accorded the rather passive task of "monitoring actions and operations" (Handbook on police accountability, oversight & integrity., 2011).
Yet, for Bayley, two of the key tasks of police reform were to get the right people and prepare leaders to be able to manage reform. Leadership, not just supervision, is frequently proffered as a key element of effective policing practice, but the quality of the research on effective police leadership does not, generally, match the perceived importance of the issue. Most of the studies are characterised by qualitative surveys, often of relatively small samples with low response rates, using bespoke surveys and exploring a "new" leadership model. Yarlagadda et al. (Yarlagadda et al., 2017), for example, explored the potential for "purposeful leadership" and Zheng et al. (Zheng et al., 2019) the impact of "authoritarian leadership." Both studies were restricted to a single UK police force with similarly low survey response rates. The first study found that "purposeful leaders" could provide a sense of direction. The second found that authoritarian leaders could suppress the raising of concerns. Both are potentially helpful prompts for the sustainability of ROL approaches. Neither are robust enough designs, nor have they been replicated.
In carrying out a Review of Police Leadership and Training for the UK Government, the present author commissioned a rapid evidence review of police leadership development (Campbell & Kodz, 2010). The review found that there were few high-quality studies on the effects of leadership on police performance. A subsequent systematic review (Pearson-Goff & Herrington, 2013) did identify a set of key characteristics: "ethical behaviour, trustworthiness, legitimacy, being a role model, communication, decision making, and critical, creative, and strategic thinking ability" (Pearson-Goff & Herrington, 2013). The authors also found five key activities that were important for leaders to engage in: "creating a shared vision, engendering organizational commitment, caring for subordinates, driving and managing change, and problem solving" (Pearson-Goff & Herrington, 2013).
In relating these characteristics of police leadership to strategies that might sustain ROL, the clearest connection could lie between two dimensions: (1) the combination of ethical behaviour, legitimacy, and engendering organisational commitment and (2) building internal procedural justice and its relationship with improved policing outcomes.
Quinton et al. (Quinton et al., 2015) summarised the evidence from their own fieldwork  by stating that "fairness and respect, internally within police organisations, can have a similar effect [of encouraging co-operation and compliance with the law] on the attitudes and behaviour of the workforce. Fairness at a supervisory and senior leadership level was associated with officers 'going the extra mile' without personal gain, following work rules, valuing the public, feeling empowered, and supporting ethical policing." Staffing the police with the right people, as one of Bayley's 6 main principles, led to several case studies. They focused on finding the "right" balance between continuity and renewal in failing or failed agencies. He concluded that the criteria for and process of recruitment are important for ROL. There are significant gaps in the research base, much of which has been focused on evaluating the screening out characteristics deemed to be incompatible with a police officer (Hughes & Andre, 2007). However, such screening tools are not so useful in screening in the traits of a good police officer. That task requires a clear framework for what good policing looks like.
White (White, 2008) used academy performance as a basis for measuring "good" policing because there were well-established metrics. His analysis found a number of variables that appeared to be significant, but he then posed the question as to whether academy and street performance were strongly related. Henson et al. (Henson et al., 2010) compared academy and field performance to a set of proposed predictive variables. They found that almost none was predictive of positive outcomes.
Diversity and equality strategies are clearly important for ROL in policing. There are a number of studies that have tried to estimate the impact of a more representative police service, with increased female and minority officers. These studies are largely correlational studies. Miller and Segal (Miller & Segal, 2013) compared female representation in US police forces with force-level crime victimisation data. They found that as female representation increases amongst officers in an area, violent crimes against women in that area, and especially domestic violence, are reported to the police at significantly higher rates. Using a similar comparative method with Indian data, Amaral et al. (Amaral et al., 2019) explored the impact of "all women police stations" on reported crime against women. They found that the opening of police stations increased the reporting of crime against women by 22%. This included kidnappings and domestic violence. They concluded that, overall, "the inclusion of women in this traditionally male occupation can improve women's access to justice and help deter future crime." Amaral et al.'s findings are supported by a study of women's police stations in Brazil, in which such stations in large cities were associated with reduced domestic homicide rates (Perova & Reynolds, 2017).
Hong (Hong, 2017) took a similar correlational approach to the representation of Black and Ethnic minority officers in policing in England and Wales. The study found an association between an increase in police force ethnic minority representation and a subsequent reduction in police misconduct. However, such correlation studies do not provide strong evidence. When a systematic review on police service diversity was carried out for the College of Policing, it found the quality and heterogeneity of the studies precluded a meta-analysis (Bullock et al., 2017). The available evidence made it difficult to determine the effectiveness of Black and Ethnic recruitment either on crime or on public satisfaction. The authors concluded that the studies suggested that Black and Ethnic Minority recruitment did not worsen arrest or crime rates or negatively influence public satisfaction with police.
Despite gaps in the evidence of improved outcomes from diversity strategies, the potential impact of a more representative workforce may be significant for perceptions of police legitimacy. Riccucci et al. (Riccucci et al., 2014) found increased legitimacy appeared to follow from increases in female officer representation in a domestic violence unit.

Complaint Investigations, Corruption Prevention, and Independent Oversight
There is "little information as the effectiveness of either internal affairs/professional standards (vertical accountability) or external civilian oversight bodies (external accountability) as methods of enhancing police accountability" (U.S. Agency for International Development, 2008). However, Borges et al. (Borges et al., 2017) provided an overview of the effectiveness of what works in combatting corruption by government officials. They found that "in contrast to the lukewarm assessments of ethnographies, case studies, and laboratory experiments, from a policy impact evaluation perspective, top-down monitoring is arguably the most thoroughly and rigorously studied anti-corruption policy to date-and the findings are nearly unanimous in the ability of audits to significantly reduce corruption at least in the short term, although questions remain about their long-term effectiveness." Whilst this finding relates to audits and inspections in areas such as public hospitals, rural development, and sanitation, there were also studies included which covered criminal justice.
A Campbell Review of anti-corruption interventions also found that combinations of control and deterrence measures were the most effective approaches (Mugellini et al., 2021). The evidence in that review and in Borges et al. would also appear to offer support for the potential effectiveness of an oversight model such as the Police Inspectorate in England and Wales, in which regular and thematic audits and inspections are conducted and published. In 2020, HMICFRS (Her Majesty's Inspectorate of Constabulary & Fire Rescue Services, 2021) covered areas such as the disproportionate use of stop and search, the service to victims, and the use of pre-charge bail, all of which are central to ROL outcomes.
Whilst there are weaknesses to the evidence base on discipline and conduct models, one area that has received more attention than most is the development of early warning systems to identify higher risk officers and intervene to prevent misconduct and corruption (Walker et al., 2001). However, it is one thing to identify that a small group of officers contribute to a disproportionate level of misconduct and corruption, quite another to design successful interventions to reduce harm (Worden et al., 2013).
An alternative approach to the traditional deterrent models of cracking down on misconduct is to encourage compliance and positive behaviours by investing in a procedural justice model within the police organisation. As we have seen above, investing in procedural justice within the organisation is a promising strategy. One potential strategy that could be used to support it is the resolution of police complaints with restorative justice models. A study in Thames Valley, 5 although small scale, was promising and the evidence for restorative justice is robust and systematically reviewed (Strang et al., 2013).

Discussion
This review has sought to cover the most promising evidence and some of the major gaps on the landscape for ROL and public protection (particularly policing). This is a complex field, with many challenges to experimental research designs. Many of the research designs reviewed were flawed. Many studies were single-site and not replicated. The strongest evidence related to the effectiveness of targeted policing strategies.
There are a number of major handbooks and international standards, notably those published by the UN and international NGOs. Some of these, such as the UNODC Handbook on police accountability, oversight, and integrity (Handbook on police accountability, oversight & integrity., 2011), provide wide coverage of the potential legal, institutional, community, and professional strategies and actions that UNODC recommend as "good practice." From the evidence reviewed in this paper, some but by no means the majority of the recommended practices are supported to some extent by research evidence. Given the complexity and variety of the different strategies, it would be timely to consider an evidence and gap map (EGM) 6 in order to provide a more comprehensive and systematic baseline assessment of the current state of the evidence and the gaps.
Drawing on an EGM would be a stronger basis for assessing the integrity and effectiveness of any nation's policing system. However, in advance of this, a structured implementation assessment tool drawing on these studies of ROL would be a significant next step. The UNODC has commissioned checklists like this for areas such as community policing (UNODC, 2021). Given the support for audit and inspection models identified in this review, it seems likely that a field audit, with local police force involvement and leadership, supported by a checklist, would, in itself, be a productive strategy to support the development of more ROL supportive approaches.

Data Each Nation Can Collect to Build Better Knowledge
Many kinds of data at the national level could help to build better knowledge about improving the rule of law and police effectiveness at crime prevention. Unfortunately, the scope of current measures is quite limited. Those studies that have looked at the performance indicators for transparency in accountability and oversight in policing, for example, have tended to focus exclusively on the area of police complaints (Filstad & Gottschalk, 2011).
A more comprehensive approach to ROL would need, as Bayley suggested for effective police reform, to focus on at least three levels of analysis: local neighbourhood; local command; and police agency. For many countries, a fourth level, national policing, would also be necessary.

Conclusions and Recommendations
This review of national policing landscapes suggests that there are a number of areas where the evidence is strong, some where there are promising approaches, and quite a number where the evidence is either weak or suggesting that approaches could backfire rather than benefit. Finally, quite a number of significant gaps were found for offering any evidence at all. These gaps can be summarised as follows: What works: There are no strategies for sustaining ROL which are clearly evidenced at a nation state level. Two strategies have clear supporting evidence but have not yet been evidenced scaled up for a whole state: • Using an evidence-based policing approach to reduce harm by targeting, testing, and tracking high-risk places, people, and the most vulnerable victims. • Community policing, linked to problem-oriented policing as a strategy to sustain police legitimacy. Community policing has a slightly stronger claim to have been implemented at a national level with programmes such as the England and Wales National Neighbourhood programme. However, as Innes et al. (Innes et al., 2020) have shown, the programme was not sustained in the face of austerity, raising questions about the challenges of sustaining community policing at a state level.
What is promising: • The legal regulation of the exercise of police detention, video-recording of police interviewing, and identification processes. • The use of cognitive, investigative interviewing models in police investigation. • The deployment of Body Worn Cameras, provided that attention is paid to the lessons of effective implementation. • The use of focused audits by an independent agency such as the HMICFRS in England and Wales. • Procedural justice for management within the police organisation supported by a style of leadership that reinforces it. • Restorative justice as a model for resolving complaints and internal grievances.
What is weak or suggestive of backfire: • The militarisation of policing appears to backfire with human rights abuses and a weakening of police reform and legal order. • The legal regulation of street powers such as stop and search/stop, question, and frisk and the regulation of covert policing.
• The benefits of different screening approaches to recruitment on identifying the positive characteristics for ROL policing. • The benefits of a representative police service -although the research on increasing the representation of women appears to more persuasive than research on minority representation. • The tenure and appointment of Chief Officers -but the case studies from England and Wales and India suggest that short tenures and direct political control without checks and balances may backfire. • The benefits of different models of policing and different internal organisations, except in so far as they may support a community policing approach. • A "Tabula Rasa" approach to abolition replacement of policing provision. • Early warning systems to identify officers with a higher risk of misconduct: the EAW systems may be able to identify the officers, but the evidence on effective interventions is not persuasive.
Significant gaps in the evidence: • There is "extremely limited data on what works in police accountability" (U.S. Agency for International Development, 2020). • The effectiveness of different models of internal investigation, professional standards, or external civilian oversight bodies.
These conclusions reflect the best evidence available as of early 2021. At the current rate of growth in policing research, it is certain that more evidence will be available by the date of publication of this essay in 2022. Yet there is no guarantee that the key gaps in the evidence will be filled. Thus the major conclusion is for the nations promoting development of policing institutions in the Global South to invest more in the research issues discussed above. The world can only hope to promote the rule of law and public protection more effectively than it has done in recent decades with more knowledge in future years.