Community Policing in the 21st Century: A Framework for Measuring Performance

Editor's note: The following is the second chapter of Urban Policy Frontiers: New thinking about how to improve housing, transportation, child welfare, and policing (2017), published by the Manhattan Institute.

A contentious narrative has developed that takes as its starting point the issue of race and policing, ranging from profiling, bias, inappropriate use of force, why, where, and how police focus their activities, as well as the safety of police officers. Recent events throughout the country do not represent the first flare-up of antagonism between police and African-Americans. In the background lurks a history of police as legal instruments supporting slavery, the Black Codes, Jim Crow, and de facto discrimination. Within memory are major eruptions during the 1960s; outbreaks of violence following the 1991 beating of Rodney King by Los Angeles police officers; and in the last few years, riots following police shootings in Ferguson, Baltimore, Milwaukee, and other cities. Several factors—a cluster of police killings of unarmed African-American men, over-imprisonment, and, arguably, the overuse of stop, question, and frisk—gave rise to the Black Lives Matter (BLM) movement.

Except for a small but noisy group of extremists (some of whom argue that African-American neighborhoods should be de-policed), police and reformers share at least one area of broad agreement: police must move, without equivocation, into community policing. For police, this is an acknowledgment of the need for renewal and revitalization of efforts to reshape policing conceptualized and initiated during the 1980s and 1990s.

For reasons beyond the scope of this paper, these efforts were sidelined during the first decades of the 21st century: the early movement toward community policing simply did not “stick” in many communities. The loss was significant, leading to the growth in mistrust between police and members of local communities that underlies so many of the violent events referred to above.

An important factor contributing to the marginalization of community policing was the failure of many departments to develop evaluation systems (of individual officers, units, and the department as a whole) that would have reinforced the strategic shift to this new policing strategy. Such processes and plans could have provided the tools for community members and police to work together to address the most salient issues facing them regarding crime and quality of life.

This paper¹ presents a framework for evaluating policing functions by which a community’s citizens, social and political leaders, and policymakers can hold police accountable for carrying out their duties in accord with legal and societal values and commensurate with local goals; for performing effectively and efficiently; and for achieving established outcomes, both crime- and noncrime-related. Implicit in the framework is the assumption that a police department must be able to demonstrate an understanding of local crime problems and concerns, knowledge of best practices in policing for addressing particular problems, and determination of their appropriate use in the local context. Police should also be held responsible for carrying out creative, effective problem solving to reduce and prevent crime and maintain public order and security.

The discussion that follows is organized into the following sections. The first section describes briefly how the business of policing developed in America over the past century and examines related attempts to develop independent performance measures for police through the 1980s. Despite their demonstrated inadequacy and problematic nature, many elements of policing from this period persist, and related performance measures are still utilized today.

The second section chronicles the maturation of community policing, beginning in the 1990s. This concept—which is built around establishing ties and working closely with citizens and community groups to better prevent and deter crime—required the development of new police tactics and strategies. These changes required rethinking how to measure police performance and accountability.

The third section proposes a framework for assessing and measuring performance based on policing functions (determined within a specific community), policing outputs (activities and best practices carried out to achieve the goals), and outcomes of policing.

A model for understanding the development of policing and measuring related police performance was developed by George Kelling (coauthor of this paper) and Mark Moore and is now widely accepted in academic as well as police circles.² The model divides the history of policing into three eras, each governed by a particular strategy: the political era (1840s–1920s), the reform (aka Progres- sive) era (1920s–1970s), and community policing (1980s–present).³ As used here, a strategy describes police organizations with reference to seven categories: source of authority; function or mission (the “business” of policing); organizational structure; relationships of police with the external environment; police efforts to manage the demand for their services; tactics (police activities and programs); and measurable outcomes. Performance measures were developed to reflect the prevailing strategy of policing during each era. In this section, we examine briefly the two strategies that governed American policing up to the 1980s, when the current strategy, community policing, began to develop.

Policing During the Political Era

When police were introduced in the U.S. during the mid-19th century, they were overlaid on the existing structures of local government. Unlike in England, where, for over a century, the national political and social elite debated how cities like London should be policed, the U.S. debates were conducted in the smoke-filled rooms of City Halls. Here, police were first established locally; with few exceptions, national and state police in America entered later, as early 20th-century developments. “From the outset,” historian Robert M. Fogelson notes, “most Americans had a firm belief that the police should be controlled by local officials and organized along municipal lines.”⁴ Just as cities were divided into wards controlled by local politicians, police departments were organized along district or precinct lines corresponding to those wards. Fogelson described these early American police departments as “adjuncts” to the political machines that dominated most cities from the late 19th century and into the 20th century.⁵ Ward leaders (“bosses”) selected district police captains, as well as most local police officers. It was therefore not surprising that, in Irish communities, most police were Irish; in Jewish communities, most police were Jews; in Italian neighborhoods, most were Italians; and so forth.

In terms of their functions, police departments during this period were catchall organizations providing the services that politicians and their constituents demanded, from housing the homeless to cleaning streets. Ward leaders handpicked police and local ward commanders and decided police priorities, which laws police were to enforce, and how order was to be defined. Police accountability was specific and strict: they were to please citizens, ensuring that ward leaders remained in office; failure for police likely meant loss of their patronage jobs. Certainly, police were expected to respond to crimes and maintain order, but the ultimate test of their efficacy was to assist ward bosses in holding on to their positions.

Police During the Reform (or Progressive) Era

Reformers, mostly outside policing and especially clergy, railed against police during the latter decades of the 19th century; but it was not until police leaders allied with political progressives early in the next century that the powerful ties between police and political machines were broken. For reformers, political influence and control were at the core of all that was wrong with American policing—corruption, inefficiency, and ineffectiveness. One way to free police from control by local politicians was to develop “scientific” measures of performance that police could use in appealing directly to the public for support. The Uniform Crime Reports (UCR), developed by the International Association of Chiefs of Police (IACP) in 1929, and newly created annual reports published by police departments that highlighted these crime statistics, were viewed as early means by which police would achieve this independence. Other mechanisms included instituting tenure for chiefs of police, civil service for employees, and tactics to put police out of reach of potentially corrupting citizens.

The UCR included seven crimes: murder, nonnegligent homicide, forcible rape, burglary, aggravated assault, larceny, and motor-vehicle theft; arson was added in 1979. The U.S. Department of Justice, with IACP support, took over compiling and reporting the UCR in 1930, assigning the task to the Bureau of Investigation (later, under J. Edgar Hoover, the Federal Bureau of Investigation).

The UCR were meant to provide a baseline against which police departments could measure crime trends over time, as well as a basis for comparison among cities. Yet the UCR had, and have, shortcomings. First, they use self-reported data that are vulnerable to manipulation throughout police organizations. Second, citizens never report a large proportion of crimes. Third, the UCR record only serious crime; misdemeanors are ignored. Finally, what UCR data tell us is not always clear. Suppose, for example, departments improve how they handle the victims of rape. As this becomes well known, victims who previously would not have reported a rape might now be more inclined to do so. Thus, UCR data could reflect increases in reported rape even though the actual number of rapes might decline. Nonetheless, to this day, the UCR remain an important metric in evaluating police departments and units.

The move to reform police led to a major change in their mission: police became law-enforcement officers whose business was addressing serious crime. According to this model, incarceration, or criminals’ fear of getting caught, would deter crime or, at least, keep it under control. Police functions shifted from providing a broad array of services for citizens to identifying and apprehending criminals by arresting them during a criminal act or after a criminal investigation. As cars became more ubiquitous, police used them—first to go from beat to beat to patrol by foot, and later, patrolling in cars to create a sense of police omnipresence that supposedly would reassure citizens and deter criminals. With car radios and home telephones more common, rapidly responding to calls for service became a keystone of police. Response times, as well as the number of times patrol cars passed neighborhood “hazards” (saloons, schools, etc.), were added to the UCR and processing metrics (such as arrest) as benchmarks by which departments were evaluated.

Architects of these changes saw policing as a relatively simple set of tasks that resulted in straightforward and predictable actions by officers. After a crime was committed, police would go to the scene. If an offender was present, they would make an arrest. If an offender was not present, the attending officer would collect whatever evidence or information was available and turn it over to a detective for investigation. The detective would clear the case and, if it was strong enough, turn it over to a prosecutor. For O. W. Wilson, who was considered the last word on policing from the late 1930s to the 1960s, day-to-day police work was akin to that of a typist working from a manuscript: reflexive, simple, and routine.⁶

In sum, during the reform era, the tools that police used were ar- rests, citations, and clearances; and police were evaluated according to levels of serious crime, as well as the number of arrests, citations, and clearances. Arrests, citations, and clearances then became the second set of major metrics used to evaluate departments, units, and individual officers. Like the UCR, these measures had problems. Chief among them: the legal definition of arrest can vary widely among states and agencies, while definitions of case clearance can vary widely by organization. Moreover, strong emphasis on obtaining arrests and citations can lead to data manipulation and encourage overcriminalization of target groups or specific crime problems. Finally, these metrics are records of police activities, rather than outcomes.

By the mid-20th century, police departments and their overseers had developed five basic performance metrics: UCR, arrests, citations, clearances, and response time. All fit the strategy in place during this era: a focus on felonies and the deterrence tactics of preventive patrol by automobile, rapid response to calls for service, and criminal investigation. To this day, they remain important indicators of police performance and will continue to be important in the future.

In 1967, President Lyndon Johnson’s Commission on Law Enforcement and the Administration of Justice published “The Challenge of Crime in a Free Society,”⁷ an influential report that gave formal recognition and validation to the form and substance of policing described above and helped perpetuate it for decades. Three elements of the report are of special interest. First, the report put forward a theory of crime causation and prevention that would dominate criminology, criminal justice, and policing for at least 30 years and would also dominate much academic thinking about crime and criminal justice to this day. At its core was the idea that crime is caused by poverty, racism, and social injustice, and can be prevented only by ameliorating these problems. Second, the report led to start-up funding for academic criminal-justice programs throughout the U.S. that, to a great extent, perpetuated the commission’s thinking. Third, and most relevant, the report largely endorsed the law-enforcement view of policing: while police could improve in a variety of ways, especial- ly with the recruitment of minorities, the challenge was essentially to do better what was already being done. The commission’s view  of crime causation and prevention largely reflected the progressive law-enforcement strategy of police: crime is prevented through social engineering, and police respond when prevention fails.

Nevertheless, the law-enforcement strategy of policing as it was carried out in the U.S. was already beginning to collapse. The riots of the 1960s revealed pervasive resentment of the police in the black community. The strategy also revealed lack of preparedness on the part of police to deal with broad-based dissent. More generally, crime began an unrelenting surge in the 1960s that continued through the 1980s, threatening or destroying the quality of life in city after city.

Research into police tactics suggested that preventive patrol and rapid response to calls for service had little beneficial impact on urban life and little effect on citizen safety, fear of crime, or crime itself. Research into police functioning demonstrated that, although police identified themselves as law-enforcement officers, they actually did little law enforcement; that police at all levels had enormous discretion and used it regularly; and that police routinely provided  a wide array of public services, ranging from maintaining order to resolving disputes, even though most were unrecognized and unacknowledged.⁸ In short, by the end of the 1970s, American policing was struggling to find its identity—nothing seemed to work, and police were at odds with substantial portions of the public.

Early Stages of Community Policing

During the 1980s, police began reconsidering their strategy, and their efforts became identified with the transition to community policing. The new strategy urged police to reach out to various com- munities and institutions to gain, at a minimum, their consent to be policed; recognized that even within the same city, different neighborhoods have different problems; and adopted a new mission of policing far broader than its previous role as the front end of a criminal justice system focused on arresting and processing offenders.

Two major contributors to this evolution were Herman Goldstein’s work on problem solving;⁹ and James Q. Wilson and George L. Kelling’s “broken windows” thesis.¹⁰ Goldstein argued that the incidents to which police respond represent  problems such as drunken driving and domestic abuse (rather than discrete incidents) and should be treated as such. Wilson and Kelling  argued that neighborhood disorderly behavior and conditions are sequentially linked to fear of crime, citizen abandonment of public spaces, serious crime, and urban decay—hence, police should take disorderly conditions and behavior seriously and deal with them.  In sum, community policing emphasized work in neighborhoods, collaboration with public-and private-sector institutions and organizations to identify and solve neighborhood problems, and decentralized decision making regarding priorities and solutions.

The problem with early community policing, as practiced during the 1980s, was twofold. First, the constraining idea that crime could be prevented only through massive social change continued to dominate popular and professional thinking about policing. Second, community policing failed to capture the vision and commitment of line police officers. For them, community policing was “soft” or “feel good” policing, more akin to social work than the crime fighting they thought that they were getting into. All this changed in the 1990s, however, as community policing grew into a full-fledged new strategy that dominates police thinking today.

Community Policing Comes of Age

During the 1990s, America’s political and policing landscape changed considerably. In New York and other cities, “tolerating the intolerable,” to use Norman Podhoretz’s phrase describing urban disorder and crime, no longer was acceptable.¹¹ A demand for order expressed itself politically, resulting in the election of mayors like Rudy Giuliani—a conservative in an overwhelmingly liberal metropolis. Police departments unveiled a new strategy that produced crime declines unmatched in recent history. Overthrowing the previously accepted view that crime could be prevented only through massive social, economic, and political change, police could now claim to be more than law-enforcement officers whose lone responsibility was to respond to crime after it had occurred. By the end of the 20th century, a community-policing strategy that would replace the progressive/reform strategy emerged.

At the forefront of this transformation was a new generation of police leadership, most educated under the federal Law Enforcement Education Program. This program recognized the failure of the law-enforcement strategy, as well as the promise of ideas such as problem solving and “broken windows.” These new police leaders moved from reactive law-enforcement policing to crime prevention, developing new tactics to reduce crime.

When William Bratton took over the New York Police Department (NYPD) in 1994, he immediately demanded that precinct captains produce double-digit declines in crime. To facilitate and monitor this goal, he established CompStat, an interactive control mechanism in which captains met regularly with their superiors and peers to present and discuss their specific crime problems and plans for managing them. CompStat was essentially a crime analysis/accountability system that traced the progress of individual precincts in achieving substantial reductions in crime. It required mid-management to understand the nature of problems in specific geograph- ical areas and to craft creative responses to these problems; and it set consequences for mid-management’s achievements or failures.¹²

New York City’s experience broke the mold: NYPD’s actions represented new and renewed concepts in American policing, the importance of which is hard to exaggerate. Furthermore, NYC’s success suggested to prudent politicians and policymakers alike that police departments were shortchanging many U.S. cities and that police possessed untapped potential to provide more value to cities than they had during past decades. Other cities adopting the new strategy had similar results. In Boston, for example, police collaborated with other justice-agency partners in Operation Cease Fire to dramatically reduce youth gang violence in the mid-1990s.¹³ In San Diego, police took the national lead in developing a problem-solving methodology.

From such examples, it is clear that characterizing community policing as “soft” fails to recognize the inherent aggressiveness of police problem solving and crime-prevention activities, as well as the potential impact on felony crime.¹⁴ For good or ill, the progressive/reform model of policing was relatively nonintrusive in urban life: basically, police sat back, waited for something to happen, and then responded. Community policing, on the other hand, attempts to anticipate security breakdowns and crime opportunities and interfere with their progression; officers are in constant touch with citizens in local neighborhoods; and police work closely with partners in other justice agencies and in the private sector, with everyone bringing information, knowledge, and resources to bear on problems in particular areas. A range of crime-prevention tactics is available to police. Restoring order (using a broken-windows approach) to prevent crime, as was done in the New York City subways, is just one example.¹⁵ “Hot-spot” policing is another.¹⁶

Nonetheless, the NYPD success produced enormous controversy: To what extent was the NYPD responsible for the city’s remarkable crime drops? Were crime data manipulated? Was CompStat too rough on district captains? How replicable was the NYPD experience, including CompStat? These controversies still rage, and research on them continues. While a basic principle of CompStat—strict accountability for performance—lay at the heart of an incipient revolution in policing, it also opened up a new set of challenges for measuring police tactics and outcomes in the community-policing strategy. The NYPD’s experiences brought issues of performance, goal setting, “bottom lines,” “stretch goals,” accountability, benchmarking, and balanced scorecards to the forefront of police strategic management.

The best of American policing today incorporates into the community policing strategy a wide array of tactics and practices effective in preventing and reducing crime. Community policing itself comprises three basic elements. One is a geographic, rather than functional, organization;¹⁷ analysis and management of problems such as carjacking or aggressive panhandling within a geographical and social context; and assumption of joint responsibility by police and community interests for setting problem-solving priorities and managing, resolving, and solving problems. An important caveat: no particular program or tactic—say, foot patrol—constitutes community policing; rather, community policing is a department-wide strategy.¹⁸ Within this framework, police can, and should, be held accountable for their knowledge and use of best practices, as well as their demonstrated record in achieving and maintaining minimum levels of security.

With the dominant paradigm in American policing today based upon a community policing model, a framework for evaluating police performance must begin by identifying and assessing the relationship between the local community and its police department: What does the community expect of its police? How is police accountability to citizens in the community to be ensured, thereby maintaining the legitimacy that police need to perform their work? Equally important yet often not identified so clearly is the burden that citizens themselves share with police for maintaining the quality of life and safety.

Working with the community does not mean that police should cater to every community or sector whim. Rather, police must learn to manage demand. When communities, particular neighborhoods, or local groups espouse values and priorities that are trivial or even alien to constitutional, legal, and moral principles—as they surely do at times—police must be able to say no to requests for police action that departs from those principles or that lies clearly outside appro- priate policing functions. Nevertheless, police responses to com- munity interests and involvement in working with them to address quality of life and crime problems should be, insofar as possible, encouraging, positive, and receptive.

With community concerns paramount as community policing has come of age, police leaders face new questions: How do we measure crime prevention? Fear reduction? The quality of police problem solving? The effectiveness of police collaboration with citizens or other justice-agency partners? Answering such questions has gained urgency, given the challenges to police that have arisen as a result  of the cluster of police shootings and recent riots in Ferguson, Balti- more, Milwaukee, and elsewhere. Properly evaluating police performance and measuring outcomes should enable a community and its police department to determine current levels of police effectiveness; suggest how and where the police might improve; and develop, or restore, citizen confidence and trust in the police.

Measuring Outputs: What Should Police Do and How Should They Do It?

In light of the strategic shifts that constitute community policing, Mark Moore of Harvard University and his colleagues set out  to identify the functions of police that are valued by a community.¹⁹ They propose seven dimensions of police performance and suggest seven related outcomes, with corresponding measures:

• Reduce crime and victimization
• Call offenders to account (initiate justice processes)²⁰
• Reduce fear and enhance personal security
• Guarantee safety in public spaces
• Use financial resources fairly, efficiently, and effectively
•  Use force and authority fairly, efficiently, and effectively
• Satisfy citizens’ demands and achieve public legitimacy

The first four represent goals and outcomes that police seek to achieve—and essentially define what are appropriate police functions. The last three pertain to specific resources that police require (force, finances, credibility, trust, and legitimacy) to carry out their work.

This list does not imply that every police department will look or perform the same, nor does it mean that the outcomes that all communities demand and expect from their police departments will be identical. The needs and expectations of various cities, neighborhoods, and communities will not be monolithic—and it is ultimately the community that should identify and prioritize police functions. Even such basic features as minimum acceptable levels of security and maximum tolerable levels of violence will vary: within one city or county, demands may vary considerably from neighborhood to neighborhood (from reducing street prostitution, to street-corner drug dealing, to illegal parking, to gang activity and homicides), and from police district to police district. Therefore, leaders and citizens within a particular city, district, or precinct will prioritize by emphasizing certain perceived dimensions of police performance and encouraging or insisting that police establish related policing goals and performance benchmarks and outcomes over (or even to the exclusion of) others.

To the work of Moore and others, we add an intermediate step that we consider critical to improving policing: identifying an organized set of benchmarks and best practices that should constitute policing outputs. Outputs comprise those activities and services performed by police as they attempt to effect outcomes or goals. They include routine activities plus innovative problem-solving efforts, and they should employ best practices recognized in policing today.

Police are still in the early stages of developing many best practices. Throughout the 1970s, police and researchers learned more about what didn’t work than about what did. Regarding crime prevention, a core function, only in the mid-1990s did policing begin to create a portfolio of tactics and activities that offered the promise of predictable results. Even then, some of what was considered the best research provided results that could not be replicated. The most egregious example was the Minneapolis Domestic Violence Experiment, which found that arresting an assailant produced outcomes superior to those obtained from offering advice and counsel or asking the assailant to leave for eight hours.²¹

The Minneapolis experiment had an enormous impact on public policy. Yet three subsequent replications of the study failed to support its findings. Consequently, in our attempt below to identify best practices as benchmarks, we counsel readers to be cautious. Some of these practices are solidly supported by experimental research; others by correlational studies; others still by reflected-upon experience. It is beyond the scope of this report to weigh the value of all these practices. Police departments will need to do so before adopting them.

In what follows, we identify and discuss some benchmarks and best practices related to each performance function and corresponding outcome. It is not an exhaustive list; the field is changing as we write. Instead, it is a select inventory of available outputs applicable to one or more functions of police. We organize them according to categories that Moore and his colleagues have set out as dimensions of police performance and desired outcomes.

Police have available at least six methods for reducing crime and victimization. Although they are distinct, effective crime prevention often includes a combination of them working in tandem.

• PRESENCE. A sense of strong police presence is established through foot and bicycle patrols, regular participation in neighborhood and community activities, and other activities that increase the number of police contacts with citizens.
• PERSUASION. Perhaps the best example of successful persuasion of offenders to desist from their criminal behavior by police has been the work of David Kennedy in Boston and other communities, known as “cease fire” or “pulling levers.”²² This approach has been tailored for work with chronic offenders, gangs, drug dealers, and gang “wannabes.” It emphasizes joint police, prosecution, and community confrontation of repeat offenders to spell out consequences for continued predatory behavior—and forceful moves to hold offenders accountable if they persist.
• MAINTAINING ORDER. This practice is most commonly identified as “broken windows.” It is based on the idea that a causal relationship exists among disorderly behaviors and conditions, breakdown of community controls, and serious crime.
• PROBLEM SOLVING. Problem solving routinely involves police in partnerships and collaboration with representatives of other justice agencies, private-sector groups, and private citizens, working together to identify and understand the contours of specific problems in their particular community and crafting a combination of law-enforcement and extra-law-enforcement solutions.
• LAW ENFORCEMENT. Law enforcement overlaps with the next broad category, “initiating justice processes” (calling offenders to account). Still, it is a basic preventive measure that operates through incarceration and primary (an imposed punishment will deter an offender’s future crimes) and secondary (the punishment of others as an example) deterrence.
• REMINDING OTHERS OF THEIR RESPONSIBILITIES. Because  police operate 24 hours a day and are distributed throughout cities, they are in a position to identify problems for which other agencies are responsible and should take action, collaborating with police where appropriate—for example, on issues such as zoning, liquor control, probation and parole violations, private security, and health-and safety-code violations.

Although some of the police mechanisms that we identify may appear to be outcome measures, the focus here is on processes or activities involved, rather than numbers attained.

• ARRESTS. Arrests are the first step in criminal-justice processing. They are associated with booking, the formal police method of processing offenders. We expect that arrests should be reasonable and based on probable cause. Booking should be done rapidly and thoroughly, to be fair to those arrested and to get the arresting officer back on the street as soon as possible.
• CLEARANCES. Clearance rates indicate crimes solved by police. They can be an indicator of police productivity and a basis for holding police units and departments accountable. Clearance rates vary considerably by crime type, with homicide usually being the crime most often cleared and burglary the least often cleared.
• CONVICTIONS. Convictions (achieved by plea bargaining or a jury verdict) are an indirect measure of the degree to which police provide prosecutors with cases that have been investigated legally and that are thorough enough to meet the threshold of probable cause. This measure should include a determination of whether investigators are attempting to employ best practices, as well as the extent to which detectives are involved in CompStat-like practices to identify local problems and develop solutions rather than using a case-by-case approach. It should also consider if detectives routinely share information with patrol and special-unit officers.

Strong evidence suggests that five sets of activities (often overlapping) reduce fear. The first four—presence, maintaining order, problem solving, and reminding organizations and citizens of their responsibilities—have been noted. But there is one other that should be mentioned:

• INCREASING  SELF-DEFENSE  CAPACITY.  A substantial body of research, going back to the 1930s, suggests that even poor neighborhoods that are effectively organized can contribute to crime prevention, order, and the reduction of fear. Some neighborhoods have always been able to maintain low fear levels; others need help organizing but have a latent capacity to maintain order and safety. Still others need massive police assistance to remove criminals and restore order; but once this is achieved, they can protect themselves.

Public spaces include parks, streets, sidewalks, commercial areas, malls, schools, public-transit facilities (train and bus stations), and roadways. Activities to achieve the goal of public safety include police patrols; programs aimed at reducing street prostitution, drug use, or graffiti; partnering with business-improvement district representatives and private-security forces in commercial areas (as is common in midtown Manhattan and Seattle); and traffic enforcement on major thoroughfares.²³ Dangerous driving practices are a major problem in many neighborhoods, and traffic enforcement can be used for a variety of purposes, including guaranteeing public safety, especially in residential neighborhoods where traffic enforcement is often ignored by patrol officers.

While many budget practices are routine, carrying them out properly is an important indicator of excellence in policing.²⁴ They include, but are not limited to:

• DEPLOYMENT. The allocation of personnel and resources to neighborhoods or geographical areas is determined by various factors, including crime levels, calls for service, population patterns, geographical characteristics (rivers, expressways, or other boundaries), and determination of different neighborhoods’ capacities for self-defense, including the availability of private security.

The current best deployment practice has at least two characteristics that, at times, conflict. First, it is flexible and quickly changing, depending upon shifting problems and other criteria. Second, it attempts to retain as many permanent patrol assignments as possible to ensure that officers remain in areas long enough to become familiar with them and be familiar to residents and users of the area. At times, both values are sought by having special units deployed flexibly and patrol officers deployed permanently.

• BUDGET COMPLIANCE. Agencies should use cost-control measures to stay within their budgets—an oft-ignored administrative process. Since personnel costs constitute the overwhelming portion of any police budget, the best practice is monthly reporting on expenditure levels against the portion of the budget year that has passed.
• OVERTIME. Misuse of overtime is widespread across police agencies, and it is often used for activities with little impact on agency goals. The best practice is to assign approximately 75% of overtime funds to geographic (district and precincts) and unit (tactical) commanders, who are held accountable through realtime overtime expenditure monitoring, often through CompStat.
• CIVILIANIZATION. Many police departments use police officers to do jobs for which they are not qualified, such as developing computer systems. The best practices are to civilianize all staff positions for which police powers are not required, which lowers costs and allows the use of specialized skills.

These best practices can serve as internal benchmarks (among patrol units within a department) and external benchmarks and as outcome and output benchmarks. We propose:

• VALUE-BASED GUIDELINES THAT SHAPE OFFICERS’ DISCRETIONARY DECISIONS ABOUT WHEN AND HOW TO USE FORCE. This highly discretionary police activity, while rarely used, is still in need of value-based guidelines.
• AN EASILY ACCESSIBLE CITIZEN COMPLAINT SYSTEM. Complaints must be courteously and promptly accepted in locations accessible to, and easily identified by, citizens. Having a complaint system available via the Internet is an essential part of any serious attempt to make such a system easily accessible.
• MECHANISMS FOR INFORMAL AND FORMAL RESOLUTION OF COMPLAINTS AGAINST POLICE. Many, if not most, complaints have to do with impolite or caustic police behavior, and most citizens would be happy with a simple apology. Care must  be taken, however, to ensure that police departments do not apply pressure to avoid formal complaints and that those received are handled professionally.
• SPEEDY RESOLUTION OF ALL COMPLAINTS. Delays will lead to decay of citizen confidence that they are receiving fair treatment. For officers, pending complaints often result in bad assignments or delays in promotion.
• TRAINING. Officers need the verbal and tactical skills to defuse conflicts whenever possible; training to do so should be linked to departmental values and guidelines.
• QUALITY DEBRIEFING. Police departments have been reluctant to debrief their experiences in handling crisis events. Yet they have much to learn by doing so. For example, lessons learned from debriefing the 1999 Columbine school massacre taught police that they couldn’t wait for special units in an active shooting situation.
• MONITORING TROUBLESOME OFFICERS. Evidence shows that a small number of officers are responsible for a large percentage of cases in which charges of police brutality and abuse are brought. Departments should set up a monitoring system to identify such officers, attempt to find means through which their behavior can be changed, assign them to low-conflict jobs, or terminate their police employment.

Since the 1960s, a critical issue for police in servicing citizens’ demands has been dealing with 911 calls. The evidence regarding 911 is strong: rapid response to calls for service provides little benefit in solving problems or preventing crime. (This does not refer to rapid response for fire departments or emergency medical service but only to police service.) Nonetheless, rapid response has been institutionalized as a police service, and managing it is essential for all police departments.

The second issue of crucial importance today is establishing police legitimacy with crime victims, offenders, and the broader community. Police have found many avenues for developing credibility with citizens. Nevertheless, establishing legitimacy with citizens has been a special problem in minority communities. Part of this issue is historical, part a residue of cultural tradition. More recently, two problems have complicated it: (1) police use of force, especially deadly force; and (2) claims that too many people, especially members of minority groups, are incarcerated, along with the role that many believe police have played in these incarcerations.

Regardless, we have many examples of police departments that were once seriously at odds with neighborhoods and communities but that now enjoy supportive, relatively harmonious, relations with diverse communities. Los Angeles is an example of such a turn-around; Boston is another.²⁵ We have learned from Los Angeles that police can restructure their relationship with communities while aggressively working to lower crime rates. Indeed, the good news is that reducing victimization and restoring order are a prerequisite for establishing police legitimacy.

Benchmarks for achieving legitimacy with the public include:

• VALUE STATEMENT. A clear set of value statements that emphasize understanding, patience, and helpfulness toward the public without officers being manipulated to pursue inappropriate goals or actions.
• CALL-MANAGEMENT SYSTEM. First, although officers should rush to emergency calls, the tradition of staying in automobiles so that officers can immediately respond to calls for service should end. Riding around in cars and waiting for calls is not good police work. Second, the idea that good police response is responding to all calls by sending a car is wasteful. In Milwaukee, for example, officers on light duty (because of injury or illness) handle a substantial portion of calls by telephone. Citizen approval of police service delivered in this fashion is quite high.
• SHAPING CITIZEN DEMANDS. Police should actively educate citizens about the services they offer, as well as alternative public and private services, such as mental health or addiction clinics. Police should not focus on sloughing off police responsibilities but should offer a means for citizens to obtain better or more appropriate services more quickly.
• TRANSPARENCY. Frequent contact with citizens and opening up police business, to the extent possible, is key. In Los Angeles, for example, portions of many CompStat meetings were open to neighborhood residents and interested citizens (discussions of confidential matters, such as suspects, were not open). Officers can regularly update citizens on activities in their neighborhoods at association meetings. And the Internet offers many more opportunities for transparency.

In policing, process often should, and often does, take precedence over outcomes. Put another way, good police work is work conducted properly. This emphasis is especially significant because American police operate within a constitutional and legal framework that appropriately constrains their exercise of power and authority. If, for example, we examine the value of calling offenders to account, how an arrest is made or an investigation conducted is ultimately more important than obtaining a specific outcome.

Measuring outputs presents particular problems in policing. We discuss the issue only briefly here. Take foot patrol in mixed-use neighborhoods (residential and small businesses that serve local residents). Research shows that fear is substantially reduced when officers patrol on foot during one shift per week. We also know that fear of crime increases when foot patrol stops.²⁶ Yet we have no idea of what benefits, if any, would result if foot patrol were to be increased beyond tested levels; nor do we know how far we could reduce the “dosage” of foot patrol before seeing a loss in impact. Thus, almost without exception, we can define or describe outputs—and usually measure their impact—with relative certainty. But we can say little quantitatively about their dimension or scope.

Ongoing documentation of outputs by police departments and making them available to public scrutiny is important, whether the activity is routine and familiar, or is a new best practice being adopted and implemented for the first time. Particular problem-solving projects often are worthy of a formal evaluation to determine whether they should continue and what adjustments are appropriate along the way.

Police should also regularly debrief their operations and make the results public. Most departments are extremely reluctant to do so, especially when operations go bad, preferring instead to deny problems or play the blame game. Yet learning from mistakes and making them public, so that others can learn from them, too, is as important as learning from successes.

Measuring Outcomes

Over time, police have accrued sets of police-outcome measures: originally, pleasing citizens and politicians; later, clearly defined law-enforcement metrics; and, most recently, complex and subtle measures, like fear reduction and creating feelings of public safety and security. Each set reflects a particular policing strategy. All have flaws, too. Some, such as UCR, are readily available but  can be manipulated and are hard to interpret. Others, such as victimization surveys, provide a relatively reliable picture of what they measure but are expensive and require skills to administer that are not typically found in police departments. Nonetheless, these are the tools available, and we must do the best that we can with them.

Further complicating the issue is the variation in problems that police address across and within cities, as well as in the priorities of citizens. Different cities not only have different problems but different tolerance levels for certain kinds of behavior. Comparing Milwaukee and San Francisco with regard to levels of disorder (as a possible outcome measure for reducing fear and guaranteeing safety)  is simply not feasible. Milwaukee has traditions of orderliness quite alien to San Francisco’s traditions of freedom of expression. Likewise, comparing districts within a city is problematic.

To measure outcomes, we must begin with citizen priorities about what is important in a city or district; add to this the problems that the area/district confronts based on additional sources (police data can reveal problems that citizens may not be aware of); identify the means used to deal with the problem (best practices); and, finally, select outcome data sources that pertain to the problem(s) of the area and that are feasible, given the resources available. We will end up with a mix of outcome measures particular to a city or district, all of which are likely to have shortcomings. Still, the mix of measures allows for crossverification (“triangulation”) and greater confidence in the reliability of the indicators. A few years ago, a dispute arose when a researcher charged that NYPD precinct commanders altered UCR data to get positive results. Several observers were quick to point out that victimization data correlated highly with the UCR data in critical dimensions, thus cross-verifying the findings.²⁷

Next, we present some measures and measurement issues associated with policing outcomes, again organized according to the dimensions of valued police functions/outcomes set forth by Moore et al.²⁸
 

• UNIFORM CRIME REPORTS. Two indicators, homicide and car theft, are generally considered reliable and accurate and could be used as benchmarks across police departments. Otherwise, UCR have limitations for measuring performance. They measure only reported and recorded crime and are vulnerable to manipulation. Likewise, an increase in certain types of offenses could indicate that more people are willing to report crimes like rape because of the improvement of police handling of such crimes.
• VICTIMIZATION SURVEYS. Victimization surveys (surveys of a random sample of a given population) provide a more accurate picture of crime levels and also provide a check on the UCR. They are, however, expensive to conduct and have shortcomings as well (under- and overreporting).

• ARRESTS. We discuss, above, the problems with using arrest as an output indicator. The same concerns arise in using it to measure outcomes. Definitions of arrest vary among jurisdictions. Using arrest as a sign of productivity can lead to overcriminalization, especially of minority populations. This characteristic can weaken the value of arrests as a benchmark for comparing different police organizations. With proper guidance, however, arrests can be an important internal benchmark (within departments).
• CLEARANCES. Clearances are vulnerable to the same definitional problems as arrests. Moreover, clearances can be manufactured by officers or units—for example, if a unit or officer offers to trade lessening the charge or recommending leniency in sentencing in exchange for the offender accepting responsibility for additional crimes (such as burglaries). Clearances are probably more reliable for internal (within departments) rather than external (between departments) benchmarking.
• CONVICTIONS. Convictions and other forms of case handling, such as plea bargaining, have potential as outcome measures. However, given that such processes are largely under the control of prosecution—and that many prosecutors are unwilling to take cases to court that are not a near-certainty to win—convictions can be more reflective of prosecutorial policies than police performance.

• ATTITUDINAL SURVEYS. Like victimization surveys, attitudinal surveys can provide information about levels of fear in communities that could be used for internal and external (if the surveys and methodologies coincide) benchmarking. Although expensive, attitudinal surveys cost less than victimization surveys. The former can measure reported attitudes and behaviors (purchase of weapons, etc.).
• FOCUS GROUPS AND OTHER FEEDBACK SOURCES. These include neighborhood associations, crime-watch groups, and small-business owners.
• SOCIAL AND ECONOMIC INDICATORS. There is some overlap here with category 4, below. Measures might include real-estate data indicating the number of people moving in to a neighborhood, as opposed to leaving it; the number of businesses and financial institutions opening and closing; and whether businesses such as restaurants, athletic clubs, recreation facilities, and grocery and drug stores stay open in the evening.

• COUNTS OF PUBLIC USAGE. Observers can count, and revenues can register, the increased or decreased use of public spaces (public transportation, parks, zoos, public toilets, sidewalks, malls, etc.).
• TRAFFIC RECORDS. Traffic records can provide data about collisions, deaths, injuries, and other damage.
• PROPERTY VALUES AND RENTAL COSTS. Real-estate, tax, and other records can be used to determine the impact of crime and fear (or lack thereof) on property and commercial interests.

Department data could be reviewed to determine if desirable outcomes were achieved for the various benchmarks spelled out in the previous section, including cost per citizen, deployment efficiency and fairness, scheduling efficiency, budget compliance, overtime expenditures, and civilianization.

• ANALYSIS OF CITIZEN COMPLAINTS. Such an analysis would examine the substance, numerical trends in, and promptness with which complaints are handled.
• OBSERVATIONS OF COMPLAINT PROCESS. One method of evaluating such a process is to walk several people, perhaps actors, through the complaint process and record their experiences.
• SETTLEMENTS IN LIABILITY SUITS
• POLICE SHOOTINGS
• REVIEW OF GUIDELINES AND TRAINING MATERIAL
• REVIEW OF RECORDS OF DEBRIEFINGS

• RESPONSE TIMES. Departmental data are readily available. However, given our understandings about the efficacy of rapid response, such data provide only limited information about the effectiveness of the police response.
• EVALUATION   OF   ALTERNATIVE   RESPONSES    TO    CALLS  FOR SERVICE. Follow-up telephone interviews can be conducted relatively inexpensively with citizens who have called for service.
• SURVEYS AND INTERVIEWS WITH POLITICAL AND ORGANIZATIONAL ELITES. These surveys would provide information about the level of credibility of, and trust in, the police department. Departments could also use surveys of citizens’ attitudes toward the police and police practices.

We have presented performance measures for police that have accrued over the decades, with most data flawed or at least having obvious weaknesses. Our proposed solution to this problem is to use multiple data sets for the purpose of triangulation (cross-verification). Moreover, we recognize the typical pluralism and variation by district and city in urban problems that citizens and police confront. The solution to this problem lies in tailoring outcome measures to localities. Our strongest recommendation, however, is to use outcome measures in combination with data gathered to assess policing outputs and their relationships with citizens and partners in the community.

We propose that cities and their police departments develop:
(1) a research and development capacity for searching out the best policing practices and outcome measures in light of the problems that they identify and seek to address; and (2) a formal measurement process, as well as a capacity for ongoing documentation, monitoring, assessing, and feeding back information for adjustment purposes during implementation of programs and processes. Constant monitoring and feedback are essential to facilitate achievement of goals and improvement in police performance, and they allow for timely, ongoing adjustment of priorities and processes.²⁹

These are important capacities—ideally, in-house—for police departments to possess. Some police departments have formed successful partnerships with universities or research organizations to carry out the second of these functions, as well as to assist periodically in conducting community surveys and formal evaluations of policing activities on various scales.³⁰

Police leaders now enjoy a rapidly developing inventory  of best practices that can be drawn upon to improve the performance of the men and women charged with policing their community. Ultimately, police leadership must be answerable for the conduct of police performance, for instituting a measurement system that ensures police accountability to the local community, and for the achievement of established goals/outcomes.

1. The authors and the Manhattan Institute thank the Goldwater Institute for authorizing a redrafting of a report on measurement of police performance that it originally commissioned in 2016. Nick Dranias, then a program officer at the Goldwater Institute, made substantive contributions to the original report.
2. George L. Kelling and Mark H. Moore, “The Evolving Strategy of Policing,” U.S. Department of Justice, National Institute of Justice, Perspectives on Policing 4 (Nov. 1988).
3. Some have argued, especially advocates of “evidence-based,” “intelligence-led,” or “predictive policing,” that we are now at the end of the community era and entering a new policing paradigm. We believe this to be mistaken: all such empirical approaches simply add new analytical techniques to the problem-solving methods that are integral ingredients of community policing.
4. Robert M. Fogelson, Big-City Police (Cambridge, MA: Harvard University Press, 1977), pp. 13–14.
5. Ibid.
6. O. W. Wilson, “Basic Police Policies,” The Police Chief (Nov. 1956): 28–29.
The Police Chief is a publication of IACP.
7. “The Challenge of Crime in a Free Society,” President’s Commission on Law Enforcement and Administration of Justice, Feb. 1967.
8. For a summary of this research, see George L. Kelling and Catherine M. Coles, “The Failure of Past Policing Strategies,” in idem, Fixing Broken Windows: Restoring Order and Reducing Crime in Our Communities (New York: Free Press, 1996), pp. 70–107.
9. Herman Goldstein, Problem-Oriented Policing (New York: McGraw-Hill, 1990).
10. James Q. Wilson and George L. Kelling, “Broken Windows: The Police and Neighborhood Safety,” The Atlantic, Mar. 1982, pp. 29–38.
11. Norman Podhoretz, “My New York,” National Review (June 4, 1999), pp. 2–8.
12. For a full description of the CompStat process and implementation, see Jon M. Shane’s three-part article: “Compstat Process,” FBI Law Enforcement Bulletin 73, no. 4 (Apr. 2004): 12–21; “Compstat Design,” no. 5 (May 2004): 12–19; and “Compstat Implementation,” no. 6 (June 2004): 13–21. For a description of a CompStat-like process in the private sector, see Robert Simons, “Control in an Age of Empowerment,” Harvard Business Review (Mar.–Apr. 1995), pp. 80–88.
13. David M. Kennedy, “Pulling Levers: Chronic Offenders, High-Crime Settings, and a Theory Prevention,” Valparaiso University Law Review 31, no. 2 (1997): 449–84.
14. George L. Kelling and William H. Sousa, Jr., “Do Police Matter?: An Analysis of the Impact of New York City’s Police Reform,” Manhattan Institute (Dec. 2001).
15. In early 1990, after an earlier failed attempt to restore order in the city’s subways, the New York City Transit Authority hired William Bratton to head the Transit Authority Police Department—then an independent police agency of approximately 3,600 officers. Order was restored, and crime began to decline in the subway almost immediately after Bratton took charge. For a detailed account, see Kelling and Coles, Fixing Broken Windows, chap. 4.
16. Anthony A. Braga, “Hot Spots Policing and Crime Prevention: A Systematic Review of Randomized Controlled Trials,” Journal of Experimental Criminology 1, no. 3 (Sept. 2005): 317–42.
17. Functional police organizations are centralized and structured around tactics—patrol, criminal investigation, special units, administration, etc. Priorities and methods are centrally controlled. Geographical police organizations are decentralized and structured around city geography, with each area (precinct/ district) providing complete police services based on local problems.
18. Kelling and Moore, “The Evolving Strategy of Policing.”
19. Mark H. Moore et al., Recognizing Value in Policing: The Challenge of Measuring Police Performance (Washington, DC: Police Executive Research Forum, 2002);Mark H. Moore and Anthony Braga, The “Bottom Line” of Policing: What Citizens Should Value (and Measure!) in Police Performance (Washington, DC: Police Executive Research Forum, 2003). Their list does not imply that every police department will look or perform the same, nor does it mean that the outcomes that all communities demand and expect from their police departments will be identical.
20. Calling “offenders to account” is a misleading phrase. Taken literally, it implies that police should move beyond investigation and arrest to achieve justice on their own. Clearly, this is not what Moore and his colleagues intend. We think that the phrase refers to initiating the full process of holding offenders accountable for their actions. Through the rest of this paper, we therefore substitute the phrase “initiate justice processes” for “calling offenders to account.”
21. Lawrence W. Sherman and Richard A. Berk, The Minneapolis Domestic Violence Experiment (Washington, DC: Police Foundation, 1984).
22. See Kennedy, “Pulling Levers.”
23. Kelling and Coles, Fixing Broken Windows, chaps. 4 and 6.
24. We thank Robert Wasserman, a police consultant, for his advice on this section.
25. See Christopher Stone, Todd Foglesong, and Christine M. Cole, “Policing Los Angeles Under a Consent Decree: The Dynamics of Change at the LAPD,” Program in Criminal Justice Policy and Management, Working Paper Series, Harvard Kennedy School (May 2009); Anthony A. Braga and Christopher Winship, “Partnership, Accountability, and Innovation: Clarifying Boston’s Experience with Pulling Levers,” in Police Innovation: Contrasting Perspectives, ed. David Weisburd and Anthony A. Braga (Cambridge: Cambridge University Press, 2006), pp. 171–87.
26. The Newark Foot Patrol Experiment (Washington, DC: Police Foundation, 1981).
27. See “Just Like in ‘The Wire,’ Real FBI Stats Are ‘Juked,’ ” Mother Jones, June 19, 2012.
28. To restate, we use Moore’s seven dimensions of the police performance and some of his elaborations on them. As readers will note, we have added materi- als to his original conceptualizations.
29. Mark H. Moore and Anthony A. Braga, “Measuring and Improving Police Performance: The Lessons of Compstat and Its Progeny,” Policing: An International Journal of Police Strategies & Management 26, no. 3 (2003): 439–53.
30. See, e.g., Beth A. Sanders and Marc L. Fields, “Partnerships with University-Based Researchers,” The Police Chief (June 2009): 58–61. One of the best- known partnerships has taken place with the Chicago CAPS program. See Wesley G. Skogan and Lynn Steiner, “Community Policing in Chicago, Year Ten: An Evaluation of Chicago’s Alternative Policing Strategy,” Chicago Community Policing Evaluation Consortium and Institute for Policy Research, Northwestern University, Jan. 2004; and Wesley G. Skogan, Police and Community in Chicago: A Tale of Three Cities (New York: Oxford University Press, 2006).