2

Practices Designed to Regulate and Control Use of Force

In its first report, this committee laid out a framework for evidence-based policing—the review, assessment, translation, and application of scientifically derived knowledge as well as processes to strengthen police departments’ decision-making, actions, and overall agency functioning. An evidence-based approach to policing requires: (1) a reliable body of knowledge about policing practices; (2) the ongoing practice of evidence-based and systematic targeting, testing, and tracking in policing; and (3) the institutionalization and implementation of knowledge in police practice (NASEM, 2021, p. 2). Concerning police use of force, it is envisioned that evidence-based policing practices and reforms will complement efforts through international law and human rights bodies to reduce the use of force and violence by the police as well as improving policing generally.

To aid in accumulating a reliable body of knowledge, the committee considers different efforts and research findings from studies of policies and practices aimed at minimizing use of excessive force by police. The different strategies described in the commissioned papers and workshop can be divided into six categories: laws and standards, external oversight bodies, police training, internal accountability and control mechanisms, technologies, and administrative incentives (outlined in Cano, 2021).

While insightful, many of the existing studies have methodological and contextual limitations that could affect their applicability to policing contexts in other countries. However, this should primarily be interpreted as an absence of rigorous evidence or lack of external validity, rather than evidence for a particular strategy’s inadequacy. Several policies and practices are promising in that they have strong theoretical groundings (and at

times—strong internal validity) and have shown some limited success in practice or through rigorous evaluation.

The committee emphasizes that these promising approaches require further testing and evaluation, as required by an evidence-based approach to policing. Promising approaches, when implemented, must be evaluated with attention to systematically measuring outcomes, as well as unintended consequences, across time (short-, medium-, and long-term). They need to be continuously assessed, and implementing agencies and funders need to be ready to adapt and change course based on those assessments if warranted. See further discussion on an evidence-based approach and evaluation of promising approaches to police training in the committee’s first two reports (NASEM, 2021; 2022). This chapter provides background on common approaches to minimizing police use of force, drawing heavily on studies and insights presented in the commissioned papers (Alpert et al., 2021; Cano, 2021) and the workshop discussion, and sets the stage for the committee’s conclusions and recommendations in Chapter 3.

LAWS AND STANDARDS

A basic way to control police use of force is through the law, specifically the application of the law by the courts against police officers who exceed reasonable standards of using force or who use force illegally. There are also international guidelines for policing that regulate police use of force and codes of conduct, and these have served as a framework for national and local standards and policies and for assessments of human rights violations. While the existence of laws that place boundaries on acceptable uses of force may or may not affect those uses in practice, creating a transparent definition of when and how force should, and should not, be used is a necessary first step. Such a definition is needed to distinguish between force that is excessive, such as firing into a crowd of peaceful protestors, from force that is both justified and promotes the rule of law (ROL), such as incapacitating an active shooter.

International Standards and Guidelines

Several declarations and principles support international human rights standards that can also be used by states to develop use-of-force policies. For example, the Universal Declaration of Human Rights, the International Covenant on Civil and Political Rights, and the Convention against Torture all provide guidance and benchmarks for how police and other law enforcement authorities should treat citizens. While the United Nations (UN) Basic Principles on the Use of Force and Firearms (hereafter, UN Basic Principles) acknowledge that in certain limited circumstances law enforcement officials

may need to use force, the principles also detail how in such instances force must be carried out in compliance with international human rights law (Biernert et al., 2015). Relatedly, the UN International Human Rights Standards for Law Enforcement on the use of force include the following:

• Everyone has the right to life, security of the person, and freedom from torture, cruel, inhuman, or degrading treatment and punishment;
• Non-violent means are to be attempted first;
• Force is to be used only when strictly necessary;
• Force is to be used only for lawful law enforcement purposes;
• No exceptions or excuses shall be allowed for unlawful use of force;
• Use of force is to be always proportional to lawful objectives;
• Restraint is to be exercised in the use of force;
• Damage and injury are to be minimized;
• A range of means for differentiated use of force is to be made available;
• All officers are to be trained in the use of the various means for differentiated use of force; and
• All officers are to be trained in use of non-violent means. (United Nations High Commissioner for Human Rights, 1997, p. 7)

The UN Code of Conduct for Law Enforcement Officials, adopted in 1979, says officers can only use force as a last resort, and that force should be proportional to the legitimate goal of the officer (United Nations, 1979). The Code of Conduct also emphasizes that police officers should uphold the human rights of all citizens and seek medical attention for all persons as necessary. Similarly, the European Convention on Human Rights permits force only when “absolutely necessary” and ensures that citizens have the right to peaceful assembly without restrictions (European Court of Human Rights, 2021). The UN Basic Principles specify that law enforcement officials shall not use firearms against persons except in self-defense or the defense of others against the imminent threat of death or serious injury. These principles are also outlined in the UN Guidance on Less-Lethal Weapons in Law Enforcement.¹ In a workshop presentation to the committee, Flores (2021) summed up these international standards by cataloging them under four overarching principles: legality, necessity, proportionality, and accountability (see Box 2-1).

UN human rights treaty monitoring bodies can provide guidance and recommendations to member states for relatively technical modifications to regulations. These modifications can, in theory, provide a legal basis for

___________________

¹ Available: https://www.ohchr.org/sites/default/files/Documents/HRBodies/CCPR/LLW_Guidance.pdf.

pushing practice in a direction to align policing with human rights principles. The UN Human Rights Council also conducts a Universal Periodic Review, which includes a review of human rights records for all member states (United Nations Human Rights Council, 2020). This review is a universal mechanism that can serve as an accountability mechanism for countries with human rights violations within police forces.² Similarly, the

___________________

² For example, the Committee Against Torture (CAT) at the UN High Commission on Human Rights, in relation to ongoing investigations of police torture of more than 100 African American men in Chicago during the 1970s and 1980s, instructed that the United States “should promptly, thoroughly and impartially investigate all allegations of acts of torture by law enforcement personnel and bring perpetrators to justice” (UN Committee Against Torture, 2006).

Inter-American Commission on Human Rights monitors human rights violations by the police on a more case by case basis (see Box 2-2).

These international structures serve to call attention to pervasive issues that may go unaddressed at individual jurisdiction or precinct levels. Here we highlight three examples of issues regarding policing impact on minority communities, lack of swift investigation, and shortcomings in sufficient and independent investigations of police-involved killings.

In 2021, the UN Human Rights Council (2021) released a report calling for policing reforms to address systemic racism affecting people of African descent worldwide.³ This report highlighted that racism was not simply an individual officer problem but could be found within policing systems. Thus, systemic racism within policing cannot be remedied by training or correcting individual officers but only by reforming those systems in policing that can create disparate or discriminatory outcomes.

In 2009, the European Court of Human Rights ruled in the case of Iribarren Pinillos v. Spain that Spain had violated the prohibition of inhuman or degrading treatment and the right to a fair trial within a reasonable time. The Court considered that the use of a smoke bomb thrown by national police created risk to physical injury and death to those present and that a determination on whether the use was necessary and proportionate in the circumstances was never established. The Court found that failure of the Spanish judicial system to conduct an effective and timely investigation into the incident violated both the European Convention on Human Rights Article 3, prohibition of torture, and Article 6, right to a fair trial within a reasonable time, given that the proceedings had lasted almost 12 years.

In 2017, the Inter-American Court of Human Rights ruled on Brazil for the case of Nova Brasilia, police operations in a neighborhood in Rio de Janeiro in 1994 and 1995 that resulted in a number of citizen deaths. The court found that early investigations did not guarantee independence and impartiality. In its judgment, the Court determined that: (1) Brazil had to conduct effective investigations into these events and publish annual reports on deaths in police operations with information on the resulting investigations; and (2) the State of Rio de Janeiro had to introduce policies and goals to reduce deaths in police encounters (Inter-American Court of Human Rights, 2017). In June 2021, the same court determined that the State of Venezuela must also publish similar annual reports, as part of its sentencing in the case of Guerrero, Molina and others v. Venezuela (Inter-American Court of Human Rights, 2021). While these actions suggest that there may be scope for international norms to regulate police use of force, the impact of UN evaluations or these court orders on police use of force has not been empirically evaluated.

___________________

³ See https://www.ohchr.org/en/NewsEvents/Pages/DisplayNews.aspx?NewsID=27296&LangID=E.

National Laws Related to Use of Force

Several countries have introduced new laws or guidelines that define acceptable uses of force by police. Peru, for instance, issued a legislative decree in 2015 to regulate the use of force by the National Police, which contains and develops relevant international principles. Likewise, the Brazilian government published inter-ministerial guidelines in 2019 to the same effect, though only binding for federal police forces. In South Africa, the Marikana Panel Report, produced by an official commission after review of deaths in a police encounter in Marikana in 2012, recommended that the state consider the Model Bill for Use of Force by Police and other law enforcement agencies in South Africa, which had been proposed by civil society organizations, as a suitable starting point for new legislation (Panel of Experts on Policing and Crowd Management, 2018, p. 132). In the United States, the 1985 Supreme Court case Tennesee v. Garner held that laws authorizing police use

of deadly force to apprehend fleeing, unarmed, and nondangerous suspects were unconstitutional.

The U.S. ruling is notable because, following this decision, homicides committed by police officers nationwide fell by approximately 16 percent, with evidence suggesting that the reduction is due not only to a reduction in fleeing felons but also to a general reduction in police shootings (Tennenbaum, 1994). This case illustrates the effect that national policy decisions can have on police behavior.

Many countries have adopted laws that reflect the UN General Assembly principle of Necessity, summarized in Box 2-1. The following examples illustrate this concept according to various police codes:

• Cyprus: Police Code of Ethics, 2003 Art. 32: “The Police use force only when strictly necessary and only to the extent required to obtain a legitimate objective”;
• Kenya: National Police Service Act No. 11A, 2011 Sixth Schedule, A(1), p. 62: “A police officer shall always attempt to use non-violent means first and force may only be employed when non-violent means are ineffective or without any promise of achieving the intended result”; and
• Greece: Code of Police Ethics, 2004 Art. 2(e): “The Police personnel ... shall use non-violent means while maintaining and enforcing law. The use of force is, in theory, permitted only when absolutely necessary and to the extent envisaged and required for law enforcement. The use of force shall always respect the principles of necessity, adequacy and proportionality” (Biernert et al., 2015).

However, the committee notes that simply adopting UN guidelines as local or national laws is not always feasible or desirable. This is particularly true when strong norms or practical on-the-ground realities conflict with the UN standards as written. As a result, guidelines may need to be adjusted to fit the needs of the community at hand. Box 2-3 provides an example of how international standards can be translated to local context with some success.

Moving beyond a focus on necessity, some countries and municipalities around the world offer specific guidance on when the use of force can be justified by on-duty officers, including situations where the deployment of weapons, notably firearms, could be limited, such as in the following examples:

• Mexico: Law on the use of force by law enforcement in the Federal District, 2008 Art. 9, lists the following situations in which force may be used: “in the submission of a person resisting arrest following the order of a competent authority or after having committed a breach of a law or regulation, in fulfilling a duty or lawful order

given by a competent authority, in the prevention of unlawful acts in the protection or defense of legally protected interests, or as legitimate self-defense” (Amnesty International, 2015, p. 23).

• Indonesia: INP Regulation No. 8/2009 Art. 47 provides that “[t]he use of firearms shall be allowed only if strictly necessary to preserve human life” (Biernert et al., 2015).

International standards and guidelines attempt to put boundaries on police use of force and define when that force may be considered excessive, while national laws codifying a reasonable amount of force and when it can be used vary by country. How standards and guidelines are translated into regional or local laws and directives, and how these translations are in turn effective in controlling police use of force, both could benefit from further examination. The effectiveness of laws and standards ties directly to the extent to which they are enforced. Obstacles to enforcement—whether

they be lack of resources to implement legal requirements, lack of training on the standards and laws, or purposeful neglect—can be investigated and addressed by the oversight and accountability strategies discussed below.

EXTERNAL OVERSIGHT BODIES

The committee was also presented with information on different structures for external oversight. Informative examples include: the Independent Commission of Investigations (INDECOM) in Jamaica, Police Complaints Authorities (PCAs) in India, Police Ombudsman Office (Ouvidoria da Polícia) in Brazil, the Independent Police Investigative Directory (IPID) in South Africa, the Independent Policing Oversight Authority (IPOA) in Kenya, and consent decrees and civilian review boards in the United States.

The INDECOM is a good example of an external oversight body. Jamaica instituted it in 2010 in response to an infamous case of multiple killings by police and the army in the Tivoli Gardens neighborhood in Kingston (INDECOM, 2020). INDECOM was tasked to independently investigate serious allegations against police. In principle, INDECOM had full control over investigations of police shootings and deaths in custody. INDECOM claims that, as a result of its work, deaths in police shootings diminished slightly at first, from 733 in the three-year period before its creation to 687 in the first three years of its existence, and then declined sharply to 327 in the following three years, after it established working protocols and awareness sessions. Even though a causal relationship between the establishment of INDECOM and reductions in police shootings cannot be drawn from existing evaluations, and other factors, such as a reversion to previous practices after an abnormally violent few years, may play a role, the declining trends in shootings warrants further study.

Hu and Conrad (2020) found that court-ordered accountability institutions can decrease unremanded deaths in police custody.⁴ A 2006 judgment by the Supreme Court of India required states and districts to establish

___________________

⁴ The Hu and Conrad study referenced two types of police custodial deaths: (1) deaths remanded to police custody, and (2) deaths not remanded to police custody. The former refers to the death of individuals sent back to police custody by an Indian court for the purpose of furthering an investigation; the latter refers to the death of individuals taken into custody by police officers prior to a court appearance. When an individual is remanded to police custody by an Indian court, one can expect the court to pay heightened attention to the physical security of that individual. As such, the police are likely to be more constrained in their use of violence against these individuals. As expected, the count of custodial deaths not remanded to police by the court is larger than the count of custodial deaths remanded to police custody following legal action. Because police violence is more prevalent in cases where individuals are taken into initial police custody than when they are remanded to police by an Indian court, this study uses a count of unremanded police custodial deaths as the main measure of police violence (Hu and Conrad, 2020).

local PCAs. These court-ordered, regional bodies were created to allow citizens to report allegations of police abuse, by which police officers can be monitored—and potentially held accountable—for abuses of power. Using a difference-in-difference research design that uses state-level variations when PCAs were adopted, Hu and Conrad (2020) found that from 2001 to 2015, Indian police officers were less likely to engage in behaviors that violated human rights when these oversight bodies were created. These findings align with general human rights literature on what works for national human rights institutions: formal institutional safeguards can be effective even in authoritarian and transition regimes, and complaint-handling mandates help institutions build broad bases of support (Linos and Pegram, 2017).

Conversely, however, the creation of external monitoring bodies in Brazil, South Africa, and Kenya appears to have had little effect on the use of force by officers. In 1995, the state of São Paulo, Brazil, created a Police Ombudsman Office (Ouvidoria da Polícia), which received confidential complaints against the police from the general public and addressed them to the relevant bodies: the Internal Affairs Unit (Corregedoria) or the Prosecutor’s Office. The ombudsman had no investigative powers. Many other Brazilian states followed suit during the late 1990s and early 2000s (Lemgruber, 2002). Though there are no robust impact evaluations of these oversight bodies, in Brazil the creation of Ouvidorias did not appear to coincide with a reduction in police abuses, which Lemgruber and colleagues (2003) hypothesize is due to the Ouvidorias’ lack of investigative capacity and reliance on previously existing internal controls.

In South Africa, the IPID came into operation on April 1, 2012, and was mandated to investigate serious cases of police misconduct and all civilian deaths in which officers were involved. If allegations were substantiated, the case either transferred to the National Prosecuting Authority, if criminal in nature, or to the police with disciplinary recommendations (APCOF, 2017). The IPID’s predecessor, the Independent Complaints Directorate (ICD), came into operation in 1997. The ICD was only required to investigate deaths in police custody and deaths as a result of police action, so the IPID was designed as a replacement (Bruce, 2020). While the IDC also had a general responsibility to receive complaints against the police, it could refer these to the police to allow for internal investigation by the police. Muntingh and Dereymaeker (2013) and Bruce (2020) argue that the creation of the IPID, while more powerful than its predecessor and with obligations to investigate serious allegations of criminal offenses, provided the South African Police Service with an apparent justification for not being internally accountable for its use of force (i.e., for relaxing its own internal controls, given that there would be an external body charged with oversight).

In Kenya, IPOA is charged with investigating deaths and serious injuries caused by police actions (Gandhi et al., 2021). It can refer criminal cases, its main focus, to the Office of the Director of Public Prosecutions. IPOA is also mandated with monitoring the Internal Affairs Unit investigations and can take them over if it believes the IAU is being delayed or unreasonable. In other words, it has the additional mandate of overseeing internal controls. However, this depends on the cooperation from the police, and although the police are legally bound to inform IPOA about deaths in police actions, in practice many cases are not reported, which seriously compromises its oversight capability (Gandhi et al., 2021).

Recent research has pointed to the influence the judicial system has on policing practices. Magaloni and Rodriguez (2020) show empirically through statistical analysis of data from the National Survey of the Population Deprived of Liberty, collected by the Mexican National Statistics Agency, how greater judicial oversight of police and prosecutors introduced through the abandonment of the inquisitorial system in Mexico led to a reduction in torture of suspects by police. This paper supports the notion that norms and incentives within the criminal justice system can affect police use of force and that adjustment to policing behavior requires institutional design or reform. While more research is needed to trace the progression of jurisprudence and the behavior of courts on procedural rights in criminal trials, the research by Magaloni and Rodriguez (2020) advocates that the introduction of stronger judicial checks on prosecutors and police is essential to restrain abuse. In another example, a study examined a supreme court order to halt police operations in the favelas in Rio de Janeiro and found a 66 percent decrease in persons killed by police, accompanied by decreases in police shootings and injuries resulting from shootings; a secondary effect of the order on police operations was a sharp decrease in homicides and shootings between citizens (Bullock, 2021).

In the United States, a consent decree is a form of federal oversight of local police—a court-ordered agreement between law enforcement agencies and government to address a problem. There is some correlational evidence that consent decrees and other forms of structural reform litigation may have been effective in reducing misconduct (see e.g., Chanin 2015, 2017; Goh, 2020; Harmon, 2009; Rushin, 2014, 2015). After reviewing the evaluation research conducted on consent decrees, Chillar(2021, p. 7) concludes that the timing of many studies (e.g., while an agency is still under the decree or recently released from it) does not allow for understanding the long-term impacts of consent decrees. Almost all of the cited studies note that, in the absence of continuing oversight, the organizational changes that occurred during consent decrees were at high risk of reverting back to previous practices.

Civilian review boards have also gained popularity in the United States as another type of external monitoring body. The Council on Criminal

Justice’s Task Force on Policing (2021) reviewed a variety of models of civilian oversight and concluded that rigorous empirical research and relative merits of oversight models do not exist. Specifically, their report reviews the most influential studies and notes many methodological problems, including the lack of accounting for other policy changes during the research periods. Thus, while there are good theoretical reasons to assume that civilian oversight may offer promising aspects such as improving public relations and transparency, the committee is unaware of any robust empirical evidence evaluating the effectiveness of these oversight bodies on police use of force.

There is some evidence from different countries to suggest that when external monitoring bodies are given investigatory and disciplinary powers and strong internal monitoring mechanisms, they may be more effective in reducing the use of force than bodies that lack these characteristics. As with other approaches to reducing the use of force, there is not yet a body of empirical evidence to support the effectiveness of external monitoring bodies. Future research could be beneficial for developing a better understanding of the role of external oversight bodies and whether they can support internal efforts to train police officers and provide reliable accountability mechanisms.

POLICE TRAINING AND EVALUATIONS

In the second report (NASEM, 2022), the committee reached a consensus on five connected principles of police training that are grounded in an evidence-based approach and that can support the ROL and the protection of the public. Training that prepares officers to reasonably use force when necessary must dissuade officers from resorting to excessive force. The five principles, which are interconnected, are as follows:

1. Training must do no harm. This first principle is fundamental to developing training that controls police use of force. Training must not only focus on the evidence about what is effective in protecting the public and promoting the ROL, it must also not contribute to greater abuses, harm, or other negative consequences at a population level. The committee understands that legitimate use of force by police can injure or kill individuals posing threats but underscores that training that leads to excessive use of force can damage whole communities without crime-reducing or public order benefits. The uses of training must be monitored for positive outcomes, negative consequences, and misuses. The remaining principles operationalize this goal.
2. The content of training should be based on the best available evidence. The policing strategies, tactics, knowledge, and skills that

officers are trained in should be supported by evidence showing that they are effectively linked to both supporting the ROL and protecting the public. Existing scientific knowledge and scientific approaches can help identify effective policing strategies and activities in which officers should be trained.

3. Training must also use evidence-based methods. The educational methods and styles of teaching used to deliver training should also be selected based on evidence of impact. Finding effective ways for officers to learn, apply knowledge, and update that knowledge is essential in linking training to everyday practices, behaviors, and outcomes.
4. Police agencies must continuously gather new evidence about the impact of training content and methods by tracking, testing, and evaluating ongoing training efforts for implementation and outcomes.
5. The delivery of training needs to be flexible and contextualized given the resources, cultures, and capacities of different police agencies that INL supports.

There is some evidence from the United States that police training may reduce the rate at which police use force in the field. Some of this evidence is based on the procedural justice model of policing, which emphasizes transparency, responding to community concerns, and explaining police actions (Wood et al., 2020). Procedural justice training teaches officers to give community members an opportunity to explain their actions and to describe the legal bases for officers’ responses to those actions (Council on Criminal Justice, 2021). Procedural justice training promotes respectful encounters between community members and police officers with the goals of enhancing police legitimacy and perceptions of fairness, and building community trust in the police. Continual training programs focused on procedural justice have been evaluated in Seattle (Owens et al., 2018) using a randomized controlled trial (RCT), as well as in Chicago (Papachristos, 2015) and Camden (Goh, 2021) using credible quasi-experimental variation in the roll-out of training programs. In all these evaluations, the programs appear to have reduced the rate at which officers used force. A recent RCT conducted in 120 crime hot spots within three U.S. cities examined the effect of procedural justice training on police behavior (Weisburd et al., 2022). The study found a reduction in arrests for officers who received the training compared to those that did not. Further, through pre- and post-training surveys to households, it found that citizens in areas patrolled by officers who received the training were less likely to perceive officers as using unnecessary force.

There is also some evidence that de-escalation training⁵ may reduce use of force incidents. Most de-escalation training emphasizes strategies for the prevention and management of aggression, including training in early intervention, verbal and non-verbal interaction or communication styles, the selection of appropriate responses in potentially violent encounters, and the use of physical intervention techniques, all designed to reduce conflict, aggression, injuries, and violence during police encounters (Engel et al., 2017). Engel and colleagues’ evaluation of the Police Executive Research Forum’s Integrating Communications, Assessment, and Tactics (ICAT) de-escalation training program used a stepped-wedge design to limit contamination issues and found the results demonstrated statistically significant reductions in use of force incidents, citizen injuries, and officer injuries in the post-training period (Engel et al., 2022).

An RCT studying a customized de-escalation program in Tempe, Arizona, found that according to officer surveys, post training, officers placed greater emphasis on compromise and self-reported greater use of compromise, knowing when to walk away, and maintaining officer safety (White and Orosco, 2021). Results from phone surveys of hundreds of citizens with recent encounters with Tempe officers reflected statistically significant differences between officers exposed to the training and officers in the control group, all favoring a positive training impact (White and Orosco, 2021). Findings regarding use of force were inconclusive given confounding issues of the pandemic and protests during the study; though the researchers did observe, through analysis of body-worn-camera footage, a significant reduction in citizens injured in use of force incidents for officers who received the training. Of note, the results of this RCT are based on surveys (captured attitudes) and not subsequent behaviors on the job.

An ethnographic case study of de-escalation training, presented at the workshop, suggested that training officers on threat assessment and slower thought processes in high-stress situations may reduce the rate at which they use indiscriminate force (Espinoza, 2021). Officers who do not know how they are expected to respond to perceived threats cannot be expected to use force in a way that is consistent with the ROL. Training which thoroughly explains to officers the boundaries of their agency’s use of force policy, and how far the organization can legally go to defend them, can

___________________

⁵ Forms of de-escalation training have varied across agencies and jurisdictions. One specific definition of police de-escalation exists within the 2017 National Consensus Policy and Discussion Paper on Use of Force (a collaborative effort among 11 law enforcement leadership and labor organizations in the United States). This definition states that police de-escalation is “taking action or communicating verbally or non-verbally during a potential force encounter in an attempt to stabilize the situation and reduce the immediacy of the threat so that more time, options, and resources can be called upon to resolve the situation without the use of force or with a reduction in the force necessary” (National Consensus Policy, 2020, p. 2).

provide critical context regarding the extent to which they may use force on the job. Espinoza (2021) observed that officers who know and take ownership of use-of-force regulations are more likely to make choices consistent with the ROL (Espinoza, 2021).

While promising, these studies have been limited to a few contexts, notably cities in the United States. It is therefore unclear whether similar effects would be observed when conducting such de-escalation trainings in different contexts.

The evidence base for effective police training to minimize use of force is therefore still quite limited. Most police training programs are not evaluated in ways that allow for rigorous assessment of the effectiveness of the training in controlling police use of force. This is frequently because of institutional restrictions within the police department that prevent researchers from linking police training to recorded police actions in the community. Instead, training is most commonly evaluated based on officer performance in simulation studies, or self-reported attitudes, which are subjected to various biases (e.g., social desirability bias and experimenter demand effects). In order to create an evidence base on effective training regarding police use of force, data on how much force police actually use must be collected in a way that can be linked with the training (or other such policy interventions). This includes implementing training programs in a way that allows for empirical data collection and evaluation. See Box 2-4 for a description of data linkages and levels of aggregation for evaluating training.

RCTs are generally considered to be the “gold standard” for producing causal evidence on the impact of a training, policy, or strategy. In a successful RCT, the researcher exposes people, or groups of people, to a treatment, randomly (e.g., via coin flip). When a random assignment process is the only thing that determines who is treated, any difference in average outcomes across control and treatment group can be causally attributed to the treatment.

RCTs are not always feasible or ethical. Sometimes, however, the implementation of a policy can create situations where treatment is arbitrary or “as good as random” and allow for natural experiments to be conducted in real time or after an implementation event. The most important characteristic of these natural experiments is that the treatment assignment rule is known to the researcher, and specifically known to be unrelated to the factors that can affect the outcomes of interest. Good examples of this include the adoption of a militarized policing after a close electoral win (used to study the impact of police militarization in Mexico (Dell, 2015) and the distribution of federal grants for hiring (used to study the impact of police manpower) (Evans and Owens, 2007). A bad example of this would be examining the effectiveness of a new code enforcement policy by comparing tax compliance across jurisdictions—some with codes in

place charging policing units with addressing revenue shortfalls caused by tax evasion and some that decided this wasn’t an issue. In this case, treatment assignment is determined by how much of a problem the outcome of interest was believed to be; places with varying levels of concern about tax evasion will generally have different rates of tax evasion, with or without a responsive policy. This is frequently referred to as “confounding” of treatment effects. Unfortunately, when confounding is present, the results of a study can be larger than or even the wrong direction of the true treatment effect, and there are limited diagnostic tools for figuring out how large a problem confounding is, when present (LaLonde, 1986).

At the same time, just because random assignment occurred does not necessarily mean that observed difference in outcomes between treated and control groups is the effect of the policy; it is possible that two groups of people will have different outcomes due to chance (which can be thought of as “random” confounding). This is particularly a concern when the number of units in each group is small. In addition, the context in which an RCT is implemented could influence measured effects. For example, in some police departments beat officers may have better lines of communication with their supervisors, which can affect how a community policing policy is implemented in practice. This would mean that a perfectly implemented RCT of community policing across divisions of one well-integrated department would have a different effect than the same policy studied in a department with less communication (Blair et al., 2021). Attempting to replicate RCTs in different contexts can address both issues of “random” confounding and context-specific effects. It is important to note that replicating confounded non-experimental studies is not guaranteed to produce unbiased causal effects, unless there is good reason to believe that the confounding issue is, itself, randomly distributed across studies—and this is rarely the case.

BARRIERS TO INTERNAL CONTROL MECHANISMS

Many police agencies have internal affairs units or other internal control mechanisms that monitor officer behavior and formally handle accusations that arise against officer use of force. The effectiveness of these internal control and accountability mechanisms is far from guaranteed. For example, effectiveness can be limited by cultural norms within the organization, by a weak system of complaints processing and tracking, or by a lack of the supervision necessary to activate such controls when a violation has occurred.

A significant hindrance to promoting ethical and accountable policing is the existence of a “code of silence” within police organizations, which prevents officers from reporting the problematic behavior of other officers

to internal control mechanisms or their supervisors. The phrase “code of silence” refers to both the reluctance of police officers to report corrupt activities carried out by their fellow officers and the reluctance of police administrators to acknowledge the existence of corruption within their agencies. In its investigations into corruption in the New South Wales police service, the Wood Commission noted that the code of silence was “essentially a shield against complaint, for example, in circumstances where an officer has stepped out of line in the heat of the moment, but it is equally powerful in those cases where deliberately corrupt practices have been embraced” (Wood, 1997, p. 108). Notably, a 2009 report building on a previous analysis of survey data on police agency integrity found that under certain conditions “police culture fosters corruption” (see Marche, 2009, p. 463; Klockars et al., 2000).

A code of silence can be found in every country, but the conduct officers are willing to protect varies greatly by country (Ivković, 2021; Loyens, 2009; Miller, 2006). Historical, political, social, and economic conditions in a country likely influence the level of police integrity and norms of police behavior.

Correlational evidence presented to the committee by Sanja Kutnjak Ivković (2021) showed that an officer’s likelihood of reporting misconduct depended on the perceived severity of the case, assumptions about the likelihood of other officers reporting (i.e., perceived norm), and perceptions that the expected discipline will be fair or too harsh. Wolfe and Piquero (2011) also report that officers who experience fair treatment by their supervisors show less commitment to the code of silence. Further, there is some evidence that other officers reporting misconduct can promote a new organizational culture that is more supportive of reporting (Ivković et al., 2018).

There are also whistleblower mechanisms that allow for individuals to break the code of silence. Whistleblowers play an important role in exposing wrongdoing that threatens public interests, and by protecting them from unfair treatment such as retaliation, discrimination, or disadvantage, individuals may be emboldened to report wrongdoing (Transparency International, 2018). Emerging international standard and practice is to allow whistleblowers to directly report the wrongdoing to competent external authorities, such as independent oversight bodies, anti-corruption commissions, ombudsman institutions (Transparency International, 2018). This practice does remove or replace internal reporting mechanisms, and it is still recommended that employees report internally first. However, legally obliging them to do so may result in complete silence for those who do not feel comfortable reporting internally for various reasons.

The role and influence of supervisors have been framed both positively and negatively in research. In theory, if supervisors take their responsibility seriously and have the resources to monitor, track, mentor, and guide

their officers according to the ROL and with the intention of protecting the population, then police use of force may be better controlled. However, there is another view: that in practice, supervisors do not have the resources and cannot devote sufficient attention to effectively monitor and guide their officers. Aggressive supervisors, moreover, may not serve as good role models for controlling uses of force by officers.

Nonetheless, supervisors are generally seen as having significant influence over their officers, whether positively or negatively, by their actions or inactions. Early studies on the role of supervision in police use of force used surveys to correlate officer attitudes and behaviors to supervisors’ attitudes and behaviors (Engel, 2000, 2001, 2002; Engel and Worden, 2006). Using these techniques, Engel (2000) identified that officers who were more likely to use force also had supervisors who were more aggressive in the field in that they tended to take over situations or played more of a dual role as both a street officer and a supervisor. We note that these results should be considered correlational, as the assignment process that connected officers to supervisors was unreported.

A more recent study used a RCT research design to examine the impact of procedurally fair supervision on officers’ uses of force (Owens et al., 2018). In this study, officers identified as working in high-risk-circumstances were randomly assigned to either a treatment or control condition. In the treatment condition, officers had a non-disciplinary meeting with their supervisor in which the supervisor modeled procedurally fair behavior, reviewed a recent incident the officer engaged in, and pointed out areas of success and potential areas for improvement. This cognitive debrief was juxtaposed to the control group who were subject to only “status quo” supervision. As a result of the intervention, treated officers were less likely than control officers to engage in uses of force in the six weeks following their supervisory meeting. The study by Owens and colleagues (2018) is particularly promising as it not only finds support for the idea that supervisors have a meaningful impact on officers’ uses of force but also identifies a specific intervention that can promote this type of supervision in a U.S. city context. The extent to which the findings could generalize to other contexts is unknown, but it points to a promising avenue for future research.

Preparation for supervisors is a needed complement to training efforts. While leading by example and good mentorship are two important components of supervision, providing organizational accountability mechanisms that can support supervision is also key. For example, supervisors need to be empowered to be able to effectively identify, monitor, correct, discipline, track, and report problematic behavior. In its second report on policing training (NASEM, 2022, pp. 29–30), the committee discussed these concepts at length:

In sum, supervisors are influencers within their departments, with results from early studies suggesting that officers’ use of force is correlated with the leadership style and activities engaged in by their supervisors (Engel, 2000). More recent results demonstrate short-term reductions in use of force among officers that experienced meetings with supervisors’ modeling of procedurally fair behavior and providing reflection and feedback on their interactions with citizens compared to officers without such meetings (i.e., only subjected to the “status quo” supervision) (Owens et al., 2018). These findings reinforce the potential importance of reform efforts that target first-line supervisors, specifically providing them with the resources necessary to carry out effective supervision.

TECHNOLOGICAL INNOVATIONS

Technologies could be used to support effective supervision and also to more directly manage officers. Using technology to monitor officer behavior has gained increasing interest in recent years, especially given that policing

is a profession in which supervision is often lacking or not present at all. In its first report (NASEM, 2021), the committee identified several emerging technologies capable of capturing information on police and monitoring officers on the job, including automatic vehicle locators, body-worn cameras, information technologies, and systems for early identification of officers at risk of serious misconduct or suicide. This discussion focuses primarily on body-worn cameras given the larger base of research on the utilization of this technology compared to others and appreciation that findings regarding reductions in use of force may apply to other technologies designed to monitor police behaviors. Of note, the usage of any technology for monitoring police behavior and police–citizen interactions comes with well-founded concerns regarding privacy and civil rights. When considering integrating any form of technology into policing and the ability of officers to leverage technologies such as connected devices and private surveillance systems, agencies should consider expectations of privacy, the impact of surveillance on marginalized communities, and the need for transparency and accountabilty, particularly as technology evolves faster than regulations.

Vehicle cameras and officer-worn/body-worn cameras (BWCs) are an accountability technology that has been rapidly diffusing across police agencies globally. BWCs are small video and audio devices worn by officers on their clothing or glasses, which can be turned on manually or automatically, depending on the policies and rules of agencies and sometimes at the officer’s discretion. When turned on, a BWC records interactions and events from the officer’s point of view, as cameras are pointed outward from the officer. The reasons and expectations for BWC adoption and use can differ between police and citizens (Lum et al., 2019). In North America, the rapid adoption of BWCs was fueled by citizens’ concerns about police use of excessive force and lack of accountability in officer-involved shootings, particularly shootings of Black Americans. Citizens and advocacy groups expect that BWCs will increase the transparency and accountability of the police, especially in their use of force, and deter officer misbehaviors (Ariel et al., 2015; Crow et al., 2017; Culhane et al., 2016; Ellis et al., 2015; President’s Task Force on 21st Century Policing, 2015; Sousa et al., 2018). However, several surveys of officers using BWCs indicate that officers like using them because they believe cameras can hold citizens accountable for their criminal activity (Lum et al., 2019).

Although this technology has only become more prevalent in use during the past 10 years, researchers have developed a substantial body of research on BWCs. The two most comprehensive reviews of the literature in this area have been conducted by Lum and colleagues (2019), who reviewed all empirical research on BWCs related to a wide variety of outcomes, and Lum and colleagues (2020), who conducted a Campbell meta-analysis of the most rigorous research on the impacts of BWCs on officer and citizen

behavior. These studies have examined evidence from various settings, including small and large agencies within the United States, the United Kingdom, and other countries. These reviews report on BWCs’ impacts on several outcomes, but the measured outcomes most relevant to the current discussion are complaints against the police and use of force by officers.

The 2020 meta-analysis conducted by Lum and colleagues suggests that current evidence is insufficient for concluding that BWCs consistently reduce officer use-of-force. On average, officers with cameras, compared to officers without them, experienced a 6.8 percent relative reduction in the use of force incidents across the studies examined. This finding was not statistically significant, and there is substantial uncertainty in this effect. Employing a moderator analysis, Lum and colleagues (2020) suggest that BWCs could be more likely to reduce police use of force if officers’ discretion in turning cameras on and off is highly restricted by their agencies (although this idea is still very tentative from the totality of research and does not often reflect reality in BWC use). In other words, a decline in the use of force through BWCs likelyrequires strict enforcement and supervision to ensure that cameras are worn and regularly turned on to discern where use of force would not be clearly proportionate to the event, or only for newer officers.

This meta-analysis also found that officers wearing cameras tend to have lower levels of citizen complaints than officers who are not wearing cameras. However, the research does not indicate whether this decline in complaints reflects a decline in community members’ reporting behaviors (they might be less inclined to file minor or unfounded complaints), changes in officer or citizen behaviors toward one another, or improvements in police–community relationships more generally. In other words, it is not clear from the evidence on citizen complaints that BWCs have improved community members’ satisfaction with or perceptions of the police, which would be expected if BWCs had a substantial effect on police behavior. Tellingly, surveys of police officers indicate that they do not believe that BWCs will substantially change their behavior (Headley et al., 2017; Pelfrey and Keener, 2016).

Further research on BWCs continues to accumulate, and now includes two evaluations in South America. A recent RCT of officers in Santa Catarina, Brazil, conducted by Barbosa and colleagues (2021), found that BWCs can lead to less use of force and improved interactions between people and police, particularly for dispatched events that are lower in seriousness and for those events involving more junior-level officers. Using a “negative interaction index” that combined charges of contempt, disobedience, resistance, and use of handcuffs or arrests (developed by Anderson, 2008), Barbosa and colleagues also found that police dispatches that involved BWCs had significant declines in these negative interactions.

Again, these findings were most likely for less-serious events and with more junior officers. In Uruguay, Mitchell and colleagues (2018) examined the impact of BWCs using a combination of BWCs and procedural justice scripts that officers used when stopping individuals for traffic offenses. This study indicated that complaints declined for officers with BWCs, although the data were unable to disentangle the impact of BWCs from the procedural justice scripts.

Further concerns have been raised regarding the implementation of BWCs. In many locations across the United States, the cost of BWCs remains the primary concern, with agencies having to account not only for the cost of the cameras themselves, but also video storage and disposal, and ongoing maintenance (Van Ness, 2020). There are also no established, uniform international standards for the implementation and use of BWCs, leaving a patchwork of policies that may result in a loss of the potential positive benefits of using BWCs. While the International Association of Chiefs of Police released model policy in 2014 on the purpose (IACP, 2014), policy, and procedures around BWCs, the lack of uniform standards and regulations raises questions regarding the discretionary power given to officers in operating the camera, and the potential for invasion of privacy.

Overall, BWCs have real potential to curb the most egregious behaviors of police violence, and as the most recent study in Brazil indicates, they may curb excessive force in lower-seriousness events that may not warrant the use of force. However, given the totality of the research evidence, the anticipated effects of BWCs as an accountability technology for policing may have been overestimated. This is consistent with the mixed, modest, and sometimes unintended effects that technologies have more generally in policing and with the fact that the impact of technology depends on agency infrastructure and functioning. Like other technologies, the effectiveness of BWCs depends on how they are used, who uses them, and how their use is monitored and supervised. These mediating factors depend on the will of the agency to implement several accountability mechanisms that could affect these factors. Mechanisms such as effective training in the use of these technologies from the outset, external monitoring bodies, and administrative policies and interventions to disincentivize the use of force are all possible tools that may be coupled with the implementation of technology to create a more effective plan for meeting the goal of reducing police use of force.

ADMINISTRATIVE INCENTIVES

This category includes policies and interventions administered internally to discourage use of force across the entire police agency. Though there are many types of administrative measures, this section focuses on promising approaches carried out in Brazil as reported in Cano (2021). The unifying

element of these initiatives is that they change the benefits, or costs, that officers may associate with using force.

Initiatives to Monitor Police Use of Force by Institution and Unit

These initiatives were generally designed to bring down lethality and improve internal control but also serve to potentially raise awareness that an officer’s actions are being recorded and observed by their supervisors and the public. The simple fact that force is being monitored can send a signal that deaths in the course of police work should be limited. For example, in the State of São Paulo, Brazil, a Special Commission for the Reduction of Lethality in Actions involving Police Officers (Comissão Especial Para Redução da Letalidade em Ações Envolvendo Policiais) was created in 2000 (Cano, 2021). This commission was tasked with reviewing incident reports from police officers involved in shootouts to identify risk factors and with proposing preventive measures to reduce deaths in police encounters. There are no known evaluations of the impact of this commission. The State of Minas Gerais, also in Brazil, published an annual report on the use of lethal force for a number of years starting in 2005. These reports ranked police units according to the number of people killed by their officers in an effort to increase transparency and create pressure on commanders to reduce their officers’ use of force or otherwise justify the fatal interactions (Cano, 2021).

Initiatives to Monitor Police Use of Force by Individual Officer

Initiatives to monitor individual officers’ use of force may be helpful for establishing patterns among units and discovering which officers may need to be retrained, or even disciplined. Where data exist, trends tend to show that incidents of lethal force are concentrated among specific units and particular officers (Cano, 2021). In 2014, the Military Police of Rio de Janeiro asked the State University of Rio to create an index to help identify officers at risk of using force excessively (Osse and Cano, 2017). The index measured individual officers’ consumption of ammunition during the previous six months, weighting it by the prevailing level of violence in the area of his or her beat and by the type of function the officer performed (on-foot patrolling, car patrolling, tactical response, etc.). Those who exceeded the expected average by more than two standard deviations were chosen for retraining. The program was introduced experimentally but later discontinued. It was not submitted to evaluation.

Economic Incentives

Several Brazilian states have resorted to economic incentives to try to bring crime rates down. In 2009, the state of Rio de Janeiro created the Integrated System of Targets and Results Monitoring (Sistema Integrado de Metas e Acompanhamentos de Resultados—SIM) (Estado do Rio de Janeiro, Brasil, 2009). Under this system, police units receive pay bonuses according to homicide and robbery rates in their area, with the reduction of the former having more weight. An indicator called “violent lethality” was added in 2011 and included the number of civilians killed in police interventions together with recorded homicides (Cano, 2021). Police units who used lethal force more often reduced their likelihood of receiving a bonus. This policy change was just one of several taking place in Rio de Janeiro at the time. It is unclear whether this policy, another policy, or the whole set of policy changes had the greatest influence on the decline in the number of persons killed by police. Nonetheless, in 2011, 2012, and 2013, this metric reached its lowest points in many years: 523, 414, and 413 deaths, respectively (a roughly 50 percent drop from 2009, when police killed 1,048 civilians) (Cano, 2021).

CONCLUSION

As this chapter makes clear, there are sizeable gaps in our knowledge of the efficacy of use of force interventions, but several promising approaches warrant further implementation, testing, and evaluation. As required by an evidence-based approach to policing, the committee emphasizes that the promising approaches outlined above, when implemented, must be evaluated with attention to systematically measuring field outcomes across time (short, medium, and long term), and when possible—across space. These approaches need to be continuously assessed to determine what conditions are optimal, and the context in which certain measures work well, do not work, or have unintended consequences.

This page intentionally left blank.