The power of police in any society is derived from the communities they serve. The public invests responsibility in its police to enforce the laws and, in some rare cases, to use force against citizens. This is an extremely important set of responsibilities that police must use with caution and care. In cases where police overstep their responsibilities and engage in misconduct, they can be removed from their positions and, in the most extreme cases, prosecuted for violating the law. The major way in which citizens ensure that police are exercising their responsibilities within the parameters set by the community is through police accountability measures.

Samuel Walker, one of the leading researchers of police accountability, describes two main levels at which the police can be held accountable: the individual level where police officers are held accountable for their actions and the organizational level where law enforcement agencies are held accountable for the services they deliver to their communities. Additionally, within each of these levels there are two main avenues for ensuring police accountability. Community members or other government actors may act through external mechanisms, while departmental leadership and supervisors may target internal influences (Walker 2005). A number of models or practices are being used in each area by law enforcement agencies across the country. In the following sections, we review some of the most promising of these models.

External Police Integrity

Citizen Involvement Models

Citizen involvement models refer to organizational accountability bodies, composed wholly or in part of citizens, that provide some level of oversight over police operations. Traditionally, citizen involvement models have focused on how police departments investigate complaints filed by citizens against officers. In recent years, however, some citizen involvement models have expanded their focus to include overseeing use-of-force investigations, issuing annual reports, providing outreach to citizens in the community, facilitating mediation between police and citizens, and providing policy analysis of local police policies.

There are roughly four different types of citizen review agencies that engage in varying levels of involvement in the police complaint investigation process (Bobb 2003; Finn 2001; Walker 2001). Although scholars and justice professionals vary in the ways in which they classify each model, the types of models generally include the following categories:

• External citizen oversight models. These organizations are generally separate from the police department, with offices in a separate location. They often take citizen complaints and perform their own investigations. Many of these organizations are composed entirely of private citizens, although some also have police employees play a role in the process.

• External auditors/ombudsman. These organizations are usually led by a single individual who is an employee of the city, county, or state and is not affiliated with the police. This individual has access to and may audit police investigations. In some cases this individual may participate in the investigation by asking additional questions or requesting to speak with different witnesses. The auditor/ombudsman evaluates the investigation as it unfolds and can request that the police conduct an additional investigation.

• Appellate review models. These organizations generally review the complaints that citizens file against the police once the internal affairs office has made a finding on the complaint. They serve as a way in which individuals who file a complaint can appeal the findings of the police department. However, their review is limited to the process of the investigation itself; they generally have limited powers and do not conduct their own investigations.

• Hybrid models. These organizations contain different elements from several of the models.

There is no one national model of citizen involvement. Rather citizen participation models are organic, emerging from a host of local variables, and the model is largely dependent on the police and the community and the history of the relationship between these two groups. Luna and Walker (2000) have posited that different cities throughout the United States have created models of citizen participation not from trying to replicate a specific model but through focusing on the local aspects of the community. They characterize the development of these other models as ”an ad hoc experimental” process that ”reflect[s] . . . the vagaries of local leadership and political compromise” (p. 88). Therefore, different models across the country have different levels of power and different functions that reflect local community expectations.

Consent Decrees

During the past decade, more than twenty police departments have entered into consent decrees with the U.S. Department of Justice after they were found to have fostered a ”pattern or practice” of civil rights violations. A consent decree is a settlement agreement entered into by the ”consent” of the party subject to its terms. The decree is submitted in writing to a court and gains legal authority once approved by a judge. These settlements are the result of federal investigations into allegations of ongoing civil rights violations by the departments. Consent decrees are intended to promote organizational accountability by providing agencies with the incentive to avoid litigation by implementing what are considered ”best practices” in the field. These can include such initiatives as enhancement of use-of-force policies and the implementation of an early intervention system.

Once a police agency enters into a consent decree, it is obligated to take specific steps involving policy and practice, as described in the decree, so that the agency comes into compliance with the law. For example, the consent decree the New Jersey State Police entered into in 1999 mandated policy changes to specifically proscribe the use of race, national, or ethnic characteristics to make traffic stops; to document traffic stops and provide supervisory review of individual traffic stops and of patterns of conduct; to conduct comprehensive investigations into allegations of misconduct; to improve training for recruits and incumbent troopers; to submit to auditing by the New Jersey attorney general; to issue regular reports on certain law enforcement activities; and to be evaluated by a court-appointed monitor. Other departments that have entered into consent decrees during the past ten years include the Pittsburgh Police Bureau, Pennsylvania; Steubenville Police Department, Ohio; and Los Angeles Police Department, California (U.S. Department of Justice 2006b).

Use-of-Force Reporting

Among topics of police accountability, officer use of force has attracted the most attention. The authority to use physical coercion and deadly force distinguishes the police from other types of organizations. Despite the consequences for citizens when this authority is misused or abused, use of force has only been subject to meaningful organizational constraints since the 1970s, before which most police agencies did not have policies governing officers’ use of force or the collection of data on use of force.

Only relatively recently have police received training and guidance in how they use force on the job (Walker 2005). The policy developed in 1972 by the New York City Police Department (NYPD) provided an original model for other departments. The NYPD made several important decisions, which included a clarification of the rules for when the use of force is appropriate, changing the standard from when a felon is fleeing to when force must be used in defense of the officer’s or another person’s life. Rules about discharging weapons were also restricted. Officers could not fire warning shots, shoot to wound, or shoot at or from a moving vehicle (Walker 2005).

In an overview of prior studies of use of force, McEwen (1996) notes that ”the research over the last 30 years . . . consistently calls for improved data collection at the local and national level” (p. 26). The Violent Crime Control and Law Enforcement Act of 1994 represents the first effort to do so nationally, providing the attorney general with the authority to bring litigation against agencies engaging in a ”pattern or practice” of civil rights violations. This legislation also requires the attorney general to ”acquire data about the use of excessive force by law enforcement officers” and ”publish an annual summary of the data acquired.”

In response to this mandate, in 1995 the National Institute of Justice (NIJ) and Bureau of Justice Statistics (BJS) convened a workshop to discuss how best to collect data on excessive use of force. Participants, which included police practitioners, lawyers, researchers, and civilians, noted that there was no single accepted definition of what constitutes ”excessive force.” The legal test, which comes from the 1985 U.S. Supreme Court case Tennessee v. Garner, limits the use of deadly force to when ”it is necessary to prevent the escape and the officer to believe that the suspect poses a significant threat of death or serious physical injury to the officer or others.” However, the policies and procedures developed to comply with these requirements differ widely.

Another important issue raised concerned the lack of a single source of information (e.g., court records, citizen complaints, police reports) that would provide the complete picture (McEwen 1996). Additionally, participants agreed that data collection should include incidents of police use of force in general rather than the limited scope of excessive force (McEwen 1996). Following the workshop, the NIJ and BJS jointly funded the International Association of Chiefs of Police (IACP) to develop a National Police Use of Force Database. Because of the sensitivity of the subject, data contributions would be voluntary and anonymous to promote participation and accurate reporting (Henriquez 1999).

As agencies continue to implement use-of-force policies, these policies have developed in four important ways: Written policies cover an increasing range of actions, the content of policies has become increasingly detailed, the review of use-of-force reports is the subject of greater attention, and increasingly departments are analyzing use-of-force reports in the aggregate to determine if patterns exist (Walker 2005). In particular, report and review has become the best practice for critical incidents, defined by Walker (2005) as ”any police action that has a potentially adverse effect on the life, liberty, or dignity or a citizen” (p. 44).

Practitioners and researchers increasingly understand police-citizen encounters as fluid and composed of multiple stages that can be analyzed and prepared for. De-escalation techniques such as ”verbal judo” and training and policies on responding to people with mental disorders are examples of efforts to control proactively the outcome of situations. Today, departments use a force continuum to train officers to apply force in a way that approximately matches the level of resistance offered by the suspect and changes incrementally based on how the interaction progresses. Some researchers (e.g., Alpert and Dunham 1999; Terrill et al. 2003; Terrill 2005) are working to improve the analytical and training capabilities of the force continuum.

Internal Police Integrity COMPSTAT

COMPSTAT, which stands for computer-driven statistics, has been described as ”a goal-oriented strategic management process that uses computer technology, operational strategy and managerial accountability to structure the manner in which a police department provides crime-control services” (Walsh 2001). After developing the fundamentals of the system while chief of the New York City Transit Police, William Bratton introduced COMP-STAT to the NYPD when he became commissioner in 1994. Many departments throughout the country have implemented COMPSTAT-type systems. COMPSTAT has since rapidly diffused through larger police departments (Weisburd et al. 2003).

COMPSTAT employs geographic information systems (GIS) and crime analysis to provide a data-driven and visual representation of a particular problem in a timely manner so that an intervention can be crafted in real time. The system is meant to promote accountability through the focused use of these data in regular police staff meetings. In these meetings, those in leadership positions are held accountable for addressing problems within their area of geographic responsibility. The emphasis of COMPSTAT is on assisting mangers to deploy rapid follow-up to the problems in their patrol beats and to then address those efforts in these regular meetings.

Departments using COMPSTAT may note the difference in where COMPSTAT locates responsibility for problem solving as opposed to its locus in community policing. Whereas community policing seeks to empower the patrol officer because of his or her proximity to the street, many departments using COMPSTAT programs give responsibility for problem prioritization and strategy to supervisors and managers, who, its proponents argue, possess the experience and resources to marshal change. It is then up to the local manager to identify the best approach to solving the problem that has been identified.

Some observers (e.g., Scott 2000) have noted that COMPSTAT’s effectiveness is highly dependent on how it is practiced. Overreliance on technology, to the detriment of both analytical thinking and selection by commanders of problems from a ”limited and conventional set” of responses (e.g., increased patrol), is not the problem-solving capability touted by the system’s proponents. Some police practitioners note that their daily routine has been completely co-opted by COMP-STAT. Departments with strong unions may face obstacles when implementing meaningful accountability measures, although some police executives have come up with creative alternatives. Anecdotal evidence also indicates that COMPSTAT meetings must find a happy medium between an accountability-free staff meeting and a gratuitously harsh or demeaning atmosphere that many COMPSTAT programs have embraced (Dewan 2004). Although the pressure to produce results is great, COMPSTAT’s advocates argue that accountability should address the quality of the efforts made in targeting a problem, not its amelioration.

Early Intervention Systems

Early intervention (EI) systems are computer programs that provide police administrators with the ability to monitor officer performance across a variety of fields.

The two types of EI systems are the Comprehensive Personnel Assessment System (CPAS) and the Performance Problem System (Walker 2005). The first system collects information from a wide range of fields and, as a result, provides a window into a broad variety of performance issues. However, these systems need a ”sophisticated technological infrastructure and an enormous amount of administrative oversight” (Walker 2005, 105). The Performance Problem System, on the other hand, collects information from a smaller set of fields and provides a more limited view of officer performance. This system is easier to manage, is less expensive, and requires less technological support than the CPAS.

The origin of the EI system derives from the long-held, but previously little tested, belief that a small number of officers are responsible for the majority of complaints made against a police department (Walker 2005). During the 1970s, some departments began to track officers who used force on the job, especially those involved in shootings. Several controversial incidents between police officers and racial minorities also served as precipitating events (Walker 2005). One such event occurred in Miami-Dade, Florida, in 1980 when four police officers beat to death an African American citizen, Arthur McDuffie, and significant community violence ensued. Following the violence, a local ordinance directed Miami-Dade to create the Employee Profile System to allow for a detailed examination of each officer (Walker 2005).

Another controversial incident occurred in Los Angeles in 1991 when, at the end of a pursuit, police officers beat Rodney King, an unarmed motorist who resisted arrest. Examining the Los Angeles Police Department (LAPD), the Christopher Commission found that forty-four officers accounted for a high percentage of complaints and that all of these officers could be identified through readily available forms of data. Despite the fact that a system could be constructed to track these data, the LAPD did not develop an EI system. The next year the Kolt Commission found exactly the same situation with sixty-seven deputies in the Los Angeles County Sheriff’s Office. In this case, however, the Sheriff’s Office did develop an early intervention system in response.

Originally, EI systems were referred to as early warning (EW) systems. The original terminology bothered many officers, because it had ”a negative connotation” (Walker 2005, 102), making it sound as if these systems focused on singling out officers for discipline. The name change also emphasizes that the EI system is intended as a tool for supervisors to use to spot officers who may be engaging in potentially problematic behavior and intervene before there is a larger problem—not to impose discipline. Specifically, the EI system is ”intended to highlight poor officer performance and provide a system to correct behavior, thus potentially saving a career rather than destroying one” (Davis et al. 2002, 25). These systems track police performance in a variety of areas in addition to use of force. Moreover, some of these systems have the ability to examine officers who are performing their jobs better than their peers in addition to those officers perhaps engaged in troublesome behaviors (Davis et al. 2002).

The EI system uses a two-step process: identification and selection. Identification occurs when an officer is flagged by the system. At this point, the system has merely identified the officer for the supervisor to take a closer look at so the supervisor can determine whether there is reason for concern. This occurs when an officer reaches a threshold in one or a combination of the fields (e.g., citizen complaints, use-of-force incidents, line-of-duty injuries). The supervisor looks at official records of the tasks that the officer does during his or her shift, discretionary decisions that involve controversial actions (e.g., use of force), and personnel records (e.g., how many sick days the officer has taken). As an extra level of supervision, the administrator examines ”the context of an officer’s work assignment and performance history” to identify possible counter-explanations for the officer being selected by the EI system (Walker 2005, 111). The second step, selection, refers to the act of the supervisor’s decision to intervene. At this phase, the supervisor determines that there is a reason to intervene and take action with the officer. Importantly, it is the supervisor rather than the EI system making the determination of whether some sort of intervention is necessary.

Officers who are selected enter into the intervention stage. The intervention stage is important because it is ”where the department delivers the ‘treatment’ designed to improve an officer’s performance” (Walker 2005, 116). Despite its importance, however, the intervention stage has received scant attention from researchers and practitioners (Walker 2005). Once the intervention is administered, the department often monitors the officer.

The EI system has two additional functions besides merely identifying problematic behavior from officers (Walker 2005). First, EI systems communicate to officers that their behavior on the job is being monitored and that the department will hold them accountable for their work performance. Second, in some EI systems, officers have access to observe their score. In these systems, officers can observe their performance and take action in advance to enhance their performance and bring it into compliance with the threshold of their fellow officers.

In-Car Video

The introduction of in-car video (ICV) units into police departments represents a potentially exciting advance in police supervision. The ICV system is a video camera system mounted in the patrol car for the purpose of filming police-citizen contacts—including criminal incidents, car accidents, and drunk driving arrests—and for offering increased protection to both officers and citizens. The ICV also provides supervisors with access to the decision making that officers conduct during the delivery of police services to citizens.

The literature traces the first experimental uses of cameras to record police-citizen contacts to the mid-1980s (Kubiovak 1997; Schrest et al. 1990). Schrest et al. note that in 1984 police administrators began to consider outfitting patrol cars with cameras after an analysis of Uniform Crime Reports (UCR) data found that four out of every five officers attacked (both fatally and non-fatally) had been on uniformed patrol. The ICV unit was suggested as a reactive tool to serve as a means of capturing police assaults on video so that the assailant could eventually be held criminally responsible (Liquori and Perry 1988, 35, in Schrest et al. 1990).

Despite the interest in equipping patrol cars with cameras, police departments still faced barriers when trying to implement ICV units. In particular, ICV systems were unfeasible due to the technological limitations of the time—large cameras that recorded poorly (Schrest et al. 1990). Police departments also faced financial constraints (Schrest et al. 1990). One agency, the Georgia Department of Public Safety, funded their ICV unit through a partnership with the Drug Enforcement Administration that provided them with federal money and money from drug seizures (Johnson 1992). Improvements in technology and the decrease in cost have now made ICV units a viable tool for many police departments.

The ICV assists law enforcement officers in the performance of their job by recording evidence, police-citizen interactions, and criminal incidents or investigations that result in arrest. The units are ideal for creating an ”objective” record of drug seizures and automobile accidents to supplement the written police report (Cook 1993; Pilant 1995). Johnson (1992) reports that Georgia had a 100% conviction rate for drug-related cases after implementing their ICV system. Additionally, should the police incident report be called into question, police departments can provide the video record of the original incident (Vetterli 1996). Ultimately, this ”objective record” of the incident serves as a check on the accuracy and veracity of a police report.

In particular, the ICV unit is also well suited for capturing officer-citizen interactions during traffic patrol (Pilant 1995). Police officers can record traffic incidents in two ways (Maxfield and Andresen 2001). First, when a police officer turns on his patrol lights to signal a motorist to pull over to the side of the road, the camera automatically begins filming. The camera records the officer-citizen interaction until the officer turns off the camera. This allows a supervisor to examine the officer’s manner of interacting with a citizen in a particular stop. Second, a police officer can film the offender actually committing the traffic offense by manually activating the camera before turning on his police lights. In recording a traffic offense, an objective record of the infraction is created for the officer and supervisor to examine later. In certain cases, officers have taken the additional step of providing voiceover narration as the event unfolds, thereby indicating in word and on video his or her reasons for stopping the motorist (Maxfield and Andresen 2001).

It is important to remember that the ICV unit is still only a tool, useful in specific situations. Though the ICV unit shows promise for improving certain aspects of police work, it is clearly inappropriate for some police operations. For example, since the camera is mounted, or at least located, in the patrol car, officers on foot patrol cannot use an ICV unit (Pilant 1995). Also, the fixed location of the camera makes the ICV unit ineffective for incidents that occur off of roadways, such as those that are the result of police calls for service and where officers leave their cars and travel outside the view of the camera and out of range of the microphone. Finally, initial experiences with these systems have found their use as a training tool to be limited. Few agencies with ICV systems have supervisors regularly monitor the tapes of their reporting officers, so to date many of the ongoing training opportunities associated with ICV have been unfulfilled.

External versus Internal Models

Many law enforcement agencies utilize both external and internal models of ensuring police accountability. While effective internal systems reduce the demand for external models of law enforcement oversight, it appears to be the case that a blend of internal and external models offers an agency the best approach for monitoring its officers and ensuring members of the community that they are receiving enforcement efforts in the most fair and equitable manner.


The key to the effectiveness of any of the police accountability models cited above is the role of the supervisor. The first-level supervisor must be trained and inclined to use the information that comes from these various models to influence the behavior of her or his officers. Initial training is essential because in many law enforcement agencies training in employee supervision is extremely limited when officers are promoted to supervisory positions. In many cases officers simply try to replicate the behavior of their prior supervisors whom they respected.

In the area of police accountability, the role of the supervisor is critical. When supervisors ignore difficult employees or fail to effectively intervene in cases of employee misconduct, the resulting situation not only affects the future behavior of that employee but negatively affects the attitudes and orientation of all employees, who quickly identify a double standard between what an organization says are its values and policies and what happens to employees who fail to carry out these policies or reflect the organizational values.

Next post:

Previous post: