Title: Use of Force By Police: Overview of National and Local Data Series: Research Report Author: Kenneth Adams, Geoffrey P. Alpert, Roger G. Dunham, Joel H. Garner, Lawrence A. Greenfeld, Mark A. Henriquez, Patrick A. Langan, Christopher D. Maxwell, and Steven K. Smith Published: National Institute of Justice and Bureau of Justice Statistics, October 1999 Subject: Police discipline and misconduct; police recruitment, selection, and retention; police planning and management 110 pages 261,000 bytes ------------------------------- Figures, charts, forms, and tables are not included in this ASCII plain-text file. To view this document in its entirety, download the Adobe Acrobat graphic file available from this Web site or order a print copy from NCJRS at 800-851-3420 (877-712-9279 for TTY users). ------------------------------- U.S. Department of Justice Office of Justice Programs Use of Force By Police Overview of National and Local Data National Institute of Justice Research Report Jointly published with the Bureau of Justice Statistics U.S. Department of Justice Office of Justice Programs 810 Seventh Street N.W. Washington, DC 20531 Janet Reno Attorney General Raymond C. Fisher Associate Attorney General Laurie Robinson Assistant Attorney General Noel Brennan Deputy Assistant Attorney General Jeremy Travis Director, National Institute of Justice Jan M. Chaiken Director, Bureau of Justice Statistics ------------------------------- Office of Justice Programs World Wide Web Site http://www.ojp.usdoj.gov National Institute of Justice World Wide Web Site http://www.ojp.usdoj.gov/nij Bureau of Justice Statistics World Wide Web Site http://www.ojp.usdoj.gov/bjs ------------------------------- Use of Force by Police Overview of National and Local Data Contributors: Kenneth Adams Geoffrey P. Alpert Roger G. Dunham Joel H. Garner Lawrence A. Greenfeld Mark A. Henriquez Patrick A. Langan Christopher D. Maxwell Steven K. Smith October 1999 NCJ 176330 Jeremy Travis Director, National Institute of Justice Jan M. Chaiken Director, Bureau of Justice Statistics Robert J. Kaminski Program Manager, National Institute of Justice Acknowledgments: This report is indebted to many individuals and organizations for their valuable assistance and insights. Special thanks are extended to the law enforcement agencies that cooperated with the researchers whose findings appear in this report. In so doing, the following agencies demonstrated the type of leadership so critical to the advancement of policing practice and policy: Charlotte-Mecklenburg (North Carolina) Police Department, Colorado Springs (Colorado) Police Department, Dallas (Texas) Police Department, Eugene (Oregon) Police Department, Miami-Dade (Florida) Police Department, St. Petersburg (Florida) Police Department, San Diego (California) Police Department, San Diego County (California) Sheriff's Department, Springfield (Oregon) Police Department, and the many departments that have participated in the use-of-force database project of the International Association of Chiefs of Police. Points of view expressed by contributors to this report do not necessarily represent the official positions or policies of the U.S. Department of Justice. ------------------------------- The National Institute of Justice and the Bureau of Justice Statistics are components of the Office of Justice Programs, which also includes the Bureau of Justice Assistance, the Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime. ------------------------------- Contents Foreword Executive Summary 1. What We Know About Police Use of Force by Kenneth Adams 2. Revising and Fielding the Police-Public Contact Survey by Lawrence A. Greenfeld, Patrick A. Langan, and Steven K. Smith 3. IACP National Database Project on Police Use of Force by Mark A. Henriquez 4. Measuring the Amount of Force Used By and Against the Police in Six Jurisdictions by Joel H. Garner and Christopher D. Maxwell 5. The Force Factor: Measuring and Assessing Police Use of Force and Suspect Resistance by Geoffrey P. Alpert and Roger G. Dunham 6. A Research Agenda on Police Use of Force by Kenneth Adams Bibliography ------------------------------- Foreword Law enforcement officers are authorized to use force in specified circumstances, are trained in the use of force, and typically face numerous circumstances during their careers when use of force is appropriate--for example, in making some arrests, restraining unruly combatants, or controlling a disruptive demonstration. When the level of force exceeds the level considered justifiable under the circumstances, however, the activities of the police come under public scrutiny. Incidents involving the use of excessive force by the police frequently receive attention from the media, legislators, and, in some instances, civil and even criminal courts. Whether the excessive force is aberrant behavior of individual officers or is a pattern and practice of an entire law enforcement agency, both the law and public opinion condemn such incidents. This report is one in a series of publications by the National Institute of Justice (NIJ) and the Bureau of Justice Statistics (BJS) that seek to inform public discussion by examining police use of force from many perspectives. The report provides an overview of the state of research knowledge about police use of force, updates progress on the national BJS Police-Public Contact Survey (PPCS) and the database project of the International Association of Chiefs of Police, provides the latest findings from NIJ-supported use-of-force research projects in several local jurisdictions, and offers a researcher's suggestions for a future research agenda on police use of force, with special attention given to issues of excessive force. Research consistently demonstrates that a small percentage of police-public interactions involve use of force. Various data sources, including police use-of-force reports, civilian complaints, victim surveys, and observational methods, confirm this basic finding. For example, the 1996 pilot test of the PPCS found that about 1 percent of people reporting contacts with police said that officers used or threatened force. Beginning in July 1999, the PPCS is being fielded to a much larger sample than responded to the 1996 test, and the results will be presented in a report next year. In the years ahead, it is expected that the PPCS will provide the basis for a legislatively mandated annual report by the Attorney General documenting the prevalence of the use of excessive force. NIJ-sponsored research at the local level found that, in the context of the subset of police-public contacts involving adult custody arrests, police used physical force (handcuffing excluded) in less than 20 percent of 7,512 arrests studied (chapter 4). Even in those instances, police primarily used weaponless tactics, such as grabbing or holding, which is consistent with the view that relatively minor types of force dominate statistics on police use of force. That view is further supported by research indicating that in incidents involving resistance by suspects, their injuries resulting from police use of force were typically minor (chapter 5). Ongoing research by NIJ and BJS seeks to provide the perspective, insight, and factual data needed by police and others to address use-of-force issues constructively. Through this and other policing research, we seek to advance our goal of assisting law enforcement agencies in protecting the public, enhancing the safety of the community and officers, and building widespread support among those they serve. Jeremy Travis Director National Institute of Justice Jan M. Chaiken, Ph.D. Director Bureau of Justice Statistics ------------------------------- Executive Summary Recent developments have heightened concern about police use of force. They range from well-publicized incidents involving allegations of excessive force to the onset of "aggressive" policing, whose frequent emphasis on zero-tolerance enforcement is sometimes regarded as encouraging use-of-force abuses. No matter what specific event triggers concern about police use of force, how is the public to assess whether such force is, in the aggregate, a major problem? One way is to examine what research has unearthed. Overview: What Do We Know About Police Use of Force? As discussed in chapter 1, research-based knowledge about police use of force can be placed into three categories. The first pertains to knowledge that can be accepted with substantial confidence as "fact." The second relates to use-of-force knowledge that can be accepted only with modest confidence because, for example, additional research is warranted. The third category consists of knowledge yet to be developed through research--that is, what is not yet known. Known with substantial confidence Known with substantial confidence is that police use force infrequently. The data indicate that a small percentage of police-public encounters involve force. For example, about 1 percent of people who had face-to-face contacts with police said that officers used or threatened force, according to preliminary estimates based on the Bureau of Justice Statistics' 1996 pretest of its Police-Public Contact Survey (chapter 2). In 7,512 adult custody arrests, another study (chapter 4) notes that fewer than one out of five arrests involved police use of physical force (defined as use of any weapon, use of any weaponless tactic, or use of severe restraints). That can be considered a low rate in view of the study's broad definition of force. Also known with substantial confidence is that police use of force typically occurs at the lower end of the force spectrum, involving grabbing, pushing, or shoving. In the study focusing on 7,512 adult custody arrests, for instance, about 80 percent of arrests in which police used force involved use of weaponless tactics. Grabbing was the tactic used about half the time. About 2.1 percent of all arrests involved use of weapons by police. Chemical agents, such as pepper spray, were the weapons most frequently used (1.2 percent of all arrests), with firearms least often used (0.2 percent). From a police administrator's point of view, these findings are predictable. Officers are trained to use force progressively along a continuum, and policy requires that officers use the least amount of force necessary to accomplish their goals. The kinds of police actions that most arouse the public's concerns--such as fatal shootings, severe beatings with fists or batons that lead to hospitalization, and choke holds that cause unconsciousness or even death--are not typical of situations in which police use force. When injuries occur as a result of the use of force, they are likely to be minor. In one study (chapter 5), researchers found that the most common injury to a suspect was a bruise or abrasion (48 percent). Another research finding that can be accepted with substantial confidence is that use of force typically occurs when police are trying to make an arrest and the suspect is resisting. This conclusion is based on four types of data: arrest statistics, surveys of police officers, observations of police behavior, and reports by the public about their encounters with police. The foregoing findings leave open the issue of excessive force because issues of proportionality are not clearly addressed. Research findings suggest, however, that many debates over excessive force will fall into gray areas where it is difficult to decide whether an officer acted properly, given credible evidence that use of force was necessary. Known with modest confidence Regarding what is known with modest confidence about police use of force, chapter 1 identifies three conclusions suggested by research data: --Use of force appears to be unrelated to an officer's personal characteristics, such as age, gender, and ethnicity. This conclusion should be accepted with caution, however. Additional verification is needed. --Use of force is more likely to occur when police are dealing with persons under the influence of alcohol or drugs or with mentally ill individuals. Research findings in this area are inconsistent, however. Further investigation, with an emphasis on implications for training, could lead to a reduction in the risk of force and injury for both police officers and civilians. --A small proportion of officers are disproportionately involved in use-of-force incidents. More research is needed. Among what is not known As stated in chapter 1: "The incidence of wrongful use of force by police is unknown. Research is critically needed to determine reliably, validly, and precisely how often transgressions of use-of-force powers occur." Researchers and practitioners both tend to presuppose that the incidence of excessive force by police is very low. If use of force is uncommon, and civilian complaints are infrequent, and civilian injuries are few, then excessive force by police must be rare. That conclusion may indeed be correct, but to the extent that it hinges on official police statistics, it is open to serious challenge. Current indicators of excessive force, such as civilian complaints and civil lawsuits, are all critically flawed. The difficulties in measuring excessive force with complaint and lawsuit records have led academics and practitioners to redirect their attention to all use-of-force incidents. Theoretically, understanding all use-of-force incidents helps put wrongful use of force in perspective. As one example of how understanding all use-of-force incidents can help put excessive force in perspective, the study of 7,512 adult custody arrests (chapter 4) makes this observation: " . . . most arrests involve no force, excessive or otherwise. When force is used, it typically involves less severe forms of tactics and weapon use. These findings provide a context for understanding excessive force, which we know can involve low-level acts of force . . . as well as the acts of force that result in physical injury or death of civilians. Arrests that involve no force, however, cannot involve excessive force and arrests that involve low levels of force are less likely to involve excessive force." Additional gaps in use-of-force knowledge include the following: --The impact of differences in police organizations, including administrative policies, hiring, training, discipline, and use of technology, on excessive force is unknown. Although many conditions that arguably lead to excessive force by police seem obvious, or appear to be a matter of common sense, a great need for systematic research in this area exists. --Influences of situational characteristics on police use of force and the transactional nature of these events are largely unknown. For example, little is known beyond research indicating that situations most likely to involve police use of force are interpersonal disturbance and violent personal crime, and situations when suspects attempt to flee or physically resist arrest. Those findings, however, do not address the transactional, or step-by-step unfolding, of police-public encounters. Was suspect resistance the result of police use of force, or did police use force after experiencing suspect resistance? Updates on Two National Projects In 1996, the Bureau of Justice Statistics (BJS) and the International Association of Chiefs of Police (IACP) initiated projects involving collection of data encompassing police use of force. Both are currently ongoing. The BJS survey To learn more about police use of force requires an understanding of the reasons for and the results of police-public encounters. As a step toward developing that understanding, BJS supplemented the National Crime Victimization Survey with a pilot test of its Police-Public Contact Survey (PPCS) in 1996 (chapter 2). Among the findings was a preliminary estimate that about 1 percent of people reporting contacts with police indicated that officers used or threatened force. In the majority of those instances, respondents said that their own actions, such as threatening police, may have provoked officers. In July 1999, a second test of PPCS was fielded to a much larger sample than that used in the 1996 pilot test. In addition, BJS anticipates adding items to its periodic surveys conducted among nationally representative samples of those confined in local jails and prisoners held by State and Federal authorities. The new survey items would provide, for the first time, information about respondents' interactions, including use of force, with police during the arrest preceding incarceration. The IACP database project Initiated in 1996, the IACP database project is designed to collect use-of-force information from law enforcement agencies across the Nation (chapter 3). To promote accurate reporting and overcome potential reluctance of agencies to participate, IACP decided that provision of data would be both voluntary and anonymous. Collected data pertain to reported use of force stemming from police responses to calls for service, whether or not those responses resulted in arrests. About 150 agencies are expected to contribute data for the 1998-99 data year. Among preliminary findings: --Based on 1995 data reported by 110 agencies, the police use-of-force rate was 4.19 per 10,000 responded-to calls for service, or 0.0419 percent. --Based on data reported for 1996-97, 87 percent of 62,411 use-of-force incidents involved officers using physical force. Officers used chemical force in 7 percent of the incidents, firearms in about 5 percent. --Based on available data for 1996-97, about 10 percent of 2,479 officers using force sustained injuries, less than 1 percent serious. About 38 percent of subjects were injured due to police use of force, with 1.5 percent sustaining major injuries. Because the data are not yet nationally representative, conclusions about national use-of-force trends should not be attempted. Two Local-Level Studies One of the local-level studies measured the amount of force used by and against police in six jurisdictions. The other measured and assessed police use of force and suspect resistance in three jurisdictions and developed the force factor, a measure of the level of force used by officers relative to the level of resistance by suspects. Study on the amount of force used in six jurisdictions The six-jurisdiction study (chapter 4) gathered data about officers' and suspects' behaviors in connection with 7,512 adult custody arrests (arrests in which suspects are transported to a detention facility, in contrast to being issued a summons to appear before a judicial officer). The researchers focused on the amount of force used by and against police, with the expectation that this information would inform issues surrounding the use of excessive force. For instance, excessive force is typically but not necessarily associated with more severe forms of force that could or do result in injury or death. Emerging from the research is a more complete understanding of the frequency with which certain types of tactics are used and what types of weapons are displayed, threatened, or actually used. The consistent findings across all six jurisdictions are that most arrests (more than 80 percent) did not involve force by police (excluding handcuffing) or by suspects. In 98 percent of arrests where force was used, no weapon was used, threatened, or even displayed. When police used some form of weaponless tactic (hitting, kicking, wrestling, etc.), the most frequent tactic involved only grabbing (about half the time). In addition to providing data on the use of weapons and weaponless tactics, the study identified three other elements, sometimes included in the concept of use of force: --Restraints. In about 82 percent of all 7,512 arrests, officers reported use of handcuffs. Leg cuffs were used in 0.9 percent of arrests. Officers used more severe restraints in 0.4 percent of arrests. Restraints were not used in approximately 16 percent of arrests. --Motion. Pursuit on foot and by car occurred in 3 percent and 2.4 percent, respectively, of all arrests. Pursuit did not occur in 94.4 percent of arrests. Suspect flight most frequently occurred by foot (in 4.7 percent of arrests). In 93.5 percent of arrests, suspects did not flee. --Voice. In 61.2 percent of arrests, police reported they used a conversational tone with suspects. Among the measures of force used by police officers that were developed by the researchers are physical force and physical force plus threats. The study found that 17.1 percent of arrests involved physical force (use of a weapon, weaponless tactic, or severe restraint) and that 18.9 percent entailed physical force or the display or threatened use of any weapon. To better distinguish between different types of force--such as between grabbing and kicking--the researchers developed a maximum force measure, which involved officers' ranking 60 hypothetical types of force in terms of their severity on a scale from 1 (least forceful) to 100 (most forceful). When the measure was applied to the types of force officers reported using, the study found that the ranking score for commanding a suspect to do something (1.3 percent of all arrests) was 22; for using handcuffs (57.3 percent of arrests), 28.2; and for displaying a handgun (2.2 percent of arrests), 55.4. The average ranking score for the types of force used in all arrests was 30. The researchers state that their findings are beginning to provide a stable picture of police behavior and the amount of force that police use in arrest situations, but they note the findings remain tentative given the small number of jurisdictions involved in the research, among other reasons. Study on police use of force and suspect resistance The study collected use-of-force data from three law enforcement agencies--police departments in two Oregon cities and one county department in Florida (chapter 5). The areas served by the Oregon departments were considered one site. The two-city Oregon site of Eugene/Springfield. Researchers analyzed 562 police actions, 57 percent of which were taken by officers responding to calls for service and 33 percent by officers reacting to situations they had observed. The most common type of incident (25 percent) confronting officers was street violence. Most police action (76 percent) was taken to apprehend or control a person. Officers often used more than one verbal or physical control tactic per incident. For instance, 93 percent of 546 incidents involved at least two tactics; 87 percent, at least three; and 41 percent, at least four. The pattern of tactic use that emerged corresponds to the traditional use-of-force continuum. The first tactic used in an incident is nearly always the least severe use of force on the continuum; the second is almost always the second-most lenient; and so on, with very few exceptions. Of 504 reported incidents in which force was used, 1.8 percent resulted in injury to officers. They were most at risk for injury when wrestling, striking, or taking a suspect to the ground. The level of force used by the department's officers relative to the amount of the suspects' resistance--the force factor--averaged slightly higher than the amount of resistance encountered. On average, more force than resistance was used. This does not necessarily imply that the level of police force was excessive. For example, an officer may justifiably use more force than does a suspect to gain control of a situation. The Florida site. The study focused on data in 882 official Control-of-Persons Reports prepared by officers' supervisors in the Miami-Dade Police Department. Ninety-seven percent of suspects resisted. The type of resistance most often reported was actively resisting arrest (36 percent), followed by assaulting the officer (25 percent). Twenty one percent of suspects attempted to escape or flee the scene. The most common type of force used by suspects was striking or hitting the officer (44 percent). Initially calm suspects were least likely to resist officers but were the most likely to flee and the most likely to resist with a gun or assault officers with a vehicle. The most common type of suspect injury was a bruise or abrasion (48 percent of those injured), followed by lacerations (24 percent), and gunshot injuries (4 percent). The chance of suspect injury was significant no matter what type of force was used by police. For example, officer use of fists entailed an 81 percent chance of suspect injury; use of a PR-24 baton, a 67 percent chance; and use of a handgun, a 48 percent chance. Suspects who were reportedly impaired by alcohol or drugs were no more likely to resist officers than sober suspects. When they did resist, however, they were more likely than nonimpaired suspects to directly assault the officer and more than twice as likely to use a gun. The most common type of force used by officers was use of hands and arms (77 percent of use-of-force incidents). In 64 percent of incidents, officers grabbed or held suspects. There were no statistically significant differences in the level of force used by male and female officers. The ethnicity of an officer did not affect the general level of force used or whether force was used. Data suggest that officers are significantly at risk for injury when they use force, particularly when they strike a suspect with their fists (48 percent chance) or use their hands and arms to control a suspect (43 percent chance). Because most use-of-force incidents involved use of hands, arms, or fists by officers, they are most at risk for injury when using precisely the types of force that they report using most frequently. Police officers' use of force in relation to suspect resistance--the force factor--averaged slightly less force than the resistance encountered. Data indicate that officers are more likely to be injured when using less force than that used by resisting suspects. A Proposed Research Agenda The development of a research agenda on police use of force, with special attention given to issues of excessive force, should be guided by these general considerations. --Research should provide new knowledge that significantly increases our understanding of the problem. --Research should be policy relevant. --Research activities, taken as a whole, should be comprehensive and systematic. Within that general framework, more work is required on what various people--general public, minorities, police administrators, patrol officers, judges, offenders, etc.-- have in mind when they refer to excessive force and how they adjudge specific instances of police behavior when questions of excessive force arise. This research is important because social problems often require shared solutions, and shared solutions require a common basis of understanding and mutual respect for differences in views. Also needed is more and better data on police use of force. Most discussions occur in an empirical vacuum where arguments are made without the benefit of solid, useful information. Research is required on how use of force by police varies across time, cities, and individual police departments. Research also is needed on individual, situational, and organizational factors related to variations in use-of-force levels, along with excessive force levels and should focus on the relation between excessive use of force, meaning the frequency with which police use force, and excessive force, meaning instances in which police use more force than is necessary. Finally, interventions, changes, and reforms that may mitigate police use-of-force problems should be identified, documented, and evaluated. ------------------------------- Sidenotes The organization of the executive summary parallels that of the report as a whole; that is, the order of topics highlighted in this summary tracks the chapter sequence. Occasional cross-references to specific chapters are intended to assist readers in locating more detailed information. "NCVS is based on interviews conducted with a nationally representative sample of U.S. households and has become a highly useful platform for testing new questionnaires and periodically implementing supplements."-- Greenfield, et al., page 15 of this report. "...IACP designed the project from the outset to reflect operational realities of modern, street-level law enforcement, including the very meaning of 'police use of force,' defined as the amount of force required by police to compel compliance by an unwilling subject."--Henriquez, page 20 of this report. ------------------------------- About this report This report is one of a series of use-of-force publications (see Bibliography, page 75) generated by research supported by the National Institute of Justice or Bureau of Justice Statistics. The data and findings herein contribute to a better understanding of the extent and nature of police use of force and of the circumstances under which such force is applied. A major objective of chapter 1 is to provide an overview of what is known (and not known) about police use of force and thereby help readers put the issue in perspective. The next two chapters are updates of two national projects. One is designed to collect data on police-public contacts, including those involving police use of force, from a nationally representative sample of persons age 12 or older. The other seeks to encourage as many local law enforcement agencies as possible to submit voluntarily and anonymously use-of-force data to a central database for analysis. Chapters 4 and 5 focus on the local level. They present use-of-force findings based on data acquired from nine police agencies. The final chapter looks ahead by proposing a research agenda on police use of force, with special attention given to issues of excessive force. Thus, this report begins with an overview of what is known about police use of force, proceeds to outline what is being learned, and concludes with a proposed plan for future research. ------------------------------- Chapter 1 What We Know About Police Use of Force by Kenneth Adams Ambrose Bierce, a social critic known for his sarcasm and wit, once described the police as "an armed force for protection and participation."1 In this pithy statement, Bierce identifies three critical elements of the police role. First, by describing the police as "armed," their ability to coerce recalcitrant persons to comply with the law is emphasized. Because police carry weapons, it follows that the force they use may have lethal consequences. The capacity to use coercive, deadly force is so central to understanding police functions, one could say that it characterizes a key element of the police role. Second, the primary purpose of police is protection, and so force can be used only to promote the safety of the community. Police have a responsibility for safeguarding the domestic well-being of the public, and this obligation even extends in qualified ways to protecting those who violate the law, who are antagonistic or violent toward the police, or who are intent on hurting themselves. In dealing with such individuals, police may use force in reasonable and prudent ways to protect themselves and others. However, the amount of force used should be proportional to the threat and limited to the least amount required to accomplish legitimate police action. Third, the concept of participation emphasizes that police and community are closely interrelated. Police are drawn from the community, and as police they continue to operate as members of the community they serve. The community, in turn, enters into a solemn and consequential relationship with the police, ceding to them the power to deprive persons of "life, liberty, and the pursuit of happiness" at a moment's notice and depending on them for public safety. Without police, the safety of the community is jeopardized. Without community support, police are dispossessed of their legitimacy and robbed of their effectiveness. This three-element definition of police makes it easy to understand why abuse of force by police is of such great concern. First, there is the humanitarian concern that police are capable of inflicting serious, even lethal, harm on the public. Second, there is the philosophical dilemma that in "protecting" the whole of society, some of its constituent parts, meaning its citizens, may be injured. Third, there is the political irony that police, who stand apart from society in terms of authority, law, and responsibility, also are part of society and act on its behalf. Thus, rogue actions by a few police, if condoned by the public, may become perceived as actions of the citizenry. Recent developments in policing have elevated concerns about police use of force beyond ordinarily high levels. In particular, community policing, which is becoming widespread as a result of financial incentives by the Federal Government, and "aggressive" policing, which is becoming widely adopted as a solution to serious crime problems, have come to the fore as perspectives of choice by policing experts. Community policing emphasizes the role of the community as "coproducers" of law and order in conjunction with the police. Communities naturally vary in attributes, and they vary in how they are defined for the purposes of community policing. Consequently, some communities look to add restrictions on police use of force, while others are satisfied with the status quo, and still others seek to ease current restrictions. Regardless of the community's orientation on this issue, community policing means increased levels of accountability and responsiveness in key areas, such as use of force. Increased accountability hinges on new information, and new information stimulates debate. The other emerging perspective is "aggressive" policing, which often falls under the rubric of broken windows theory, and, as a strategic matter, is concerned with intensifying enforcement against quality-of-life and order maintenance offenses. The influence of aggressive policing can be seen in the proliferation of "zero tolerance" enforcement strategies across the Nation. The concern is that the threat posed by petty offenders may be exaggerated to the point that use of force becomes more commonplace and abuses of force more frequent. The Violent Crime Control and Law Enforcement Act of 1994 mirrored congressional concern about excessive force by authorizing the Civil Rights Division of the U.S. Department of Justice (DOJ) to initiate civil actions against police agencies when, among other conduct, their use of force reaches a level constituting a pattern or practice depriving individuals of their rights. DOJ exercised that authority when, for example, it determined that an urban police department engaged in such conduct and negotiated a consent decree that put in place a broad set of reforms, including an agreement by the department to document its use of force and to implement an early warning system to detect possible abuses.[2] Use-of-force concerns also are reflected in the attention the media give to possible instances of police abuse. An accumulation of alleged abuse-of-force incidents, widely reported in the media, encourages overgeneralization by giving the impression that police brutality is rampant and that police departments across the Nation are out of control. For example, Human Rights Watch states, "Allegations of police abuse are rife in cities throughout the country and take many forms."[3] Before considering the details of recent research efforts on police use of force, it is useful to summarize the state of our knowledge.[4] We know some details about police use of force with a high degree of certainty. These items represent "facts" that should frame our understanding of the issues. Other details about police use of force we know in sketchy ways, or the research is contradictory. These items should be subject to additional research using more refined methods of inquiry. Finally, there are some aspects of police use of force about which we know very little or next to nothing. These items represent critical directions for new inquiry. As is often the case with important policy questions, the information that we are most confident of is of limited value. In many cases, it does not tell us what we really need to know, because it does not focus squarely on the important issues or is subject to competing interpretations. Conversely, the information that is most critical for policy decisions often is not available or is very difficult to obtain. Such is the case with police use of force. The issues that most concern the public and policymakers lack the kinds of reliable and solid information that advance debate from the realm of ideological posturing to objective analysis. Nonetheless, it is important to take stock of our knowledge so that it is clear which issues can be set aside and which should be the target of efforts at obtaining new knowledge. What, then, is the state of knowledge regarding police use of force? We begin with issues about which we have considerable information and a high degree of confidence in our knowledge. Discussed next are issues where knowledge is modest and considerably more research is merited. Finally, we conclude with issues that are critical to debates over police use of force and about which little knowledge exists. What We Know With Substantial Confidence About Police Use of Force Police use force infrequently.Whether measured by use-of-force reports, citizen complaints, victim surveys, or observational methods, the data consistently indicate that only a small percentage of police-public interactions involve the use of force. As Bayley and Garofalo observed, police-citizen encounters that involve use of force and injury are "quite rare."[5] Because there is no standard methodology for measuring use of force, estimates can vary considerably on strictly computational grounds. Different definitions of force and different definitions of police-public interactions will yield different rates[6] (see sidebar "Working definitions"). In particular, broad definitions of use of force, such as those that include grabbing or handcuffing a suspect, will produce higher rates than more conservative definitions. The Bureau of Justice Statistics' (BJS) 1996 pretest of its Police-Public Contact Survey resulted in preliminary estimates that nearly 45 million people had face-to-face contact with police over a 12-month period and that approximately 1 percent, or about 500,000 of these persons, were subjected to use of force or threat of force[7] (see chapter 2). When handcuffing is included in the BJS definition of force, the number of persons increases to 1.2 million. Expanding and contracting definitions of "police-public" interactions also work to affect use-of-force rates but in an opposite way from definitions of force. Broad definitions of police-public "interactions," such as calls for service, which capture variegated requests for assistance, lead to low rates of use of force. Conversely, narrow definitions of police-public interactions, such as arrests, which concentrate squarely on suspects, lead to higher rates of use of force. The International Association of Chiefs of Police (IACP) is in the process of compiling statistics on use-of-force data being submitted by cooperating agencies (see chapter 3). These data indicate that force is used in less than one-half of 1 percent of dispatched calls for service. From this point of view, one might well consider police use of force a rare event. This figure is roughly consistent with the preliminary estimate reported by BJS, although the IACP figure is subject to the reporting biases that may exist in police agency data. Furthermore, IACP data are not yet representative of the national picture because of selection bias; the estimate is based on a small percentage of police departments that voluntarily report information on use of force. Garner and Maxwell found that physical force (excluding handcuffing) is used in fewer than one of five adult custody arrests (see chapter 4). While this figure hardly qualifies as a rare event, it can be considered low, especially in light of the broad definition of force that was used. In characterizing police use of force as infrequent or rare, the intention is neither to minimize the problem nor to suggest that the issue can be dismissed as unworthy of serious attention. Society's ends are best achieved peaceably, and we should strive to minimize the use of force by police as much as possible. However, it is important to put police use of force in context in order to understand the potential magnitude of use-of-force problems. Although estimates may not completely reassure everyone that police are doing everything they can to minimize the use of force, the data do not support the notion that we have a national epidemic of police violence. Another purpose for emphasizing the infrequent nature of police use of force is to highlight the methodological challenges of trying to count or study infrequent events. In this regard, methodological approaches can vary considerably in terms of cost efficiency, reliability, and precision of information obtained. In BJS's 1996 pilot household survey of 6,421 persons, 14 respondents, or roughly 1 in 450, said that they were subjected to use of force or threat of force by police over a year's time. The household survey approach has the benefit of providing national-level estimates based on data that are free of police agency reporting biases. However, as noted by BJS, the preliminary estimates derived from such a small number of respondents are subject to a wide margin of error. This issue is particularly important if one is interested in tracking changes over time, because a very small change in reporting can have a very large impact on estimates. In the survey's continuing development, the next pilot test will use a sample about 10 times the size of the 1996 pilot test as well as involve a redesigned questionnaire. Police use of force typically occurs at the lower end of the force spectrum, involving grabbing, pushing, or shoving. Relatively minor types of force dominate statistics on police use of force. Garner and Maxwell (see chapter 4) observed that police use weaponless tactics in roughly 80 percent of use-of-force incidents and that half the time the tactic involved grabbing the suspect. Alpert and Dunham (see chapter 5) found that in Miami 64 percent of use-of-force incidents involved grabbing or holding the suspect. In the BJS pilot national survey, it was estimated, preliminarily, that about 500,000 people were "hit, held, pushed, choked, threatened with a flashlight, restrained by a police dog, threatened with or actually sprayed with chemical or pepper spray, threatened with a gun, or experienced some other form of force."[8] Three-fifths of these situations, however, involved only holding. Finally, Pate and Fridell's survey of law enforcement agencies regarding use of force and civilian complaints also confirms that minor types of force occur more frequently than serious types.[9] As a corollary finding, when injuries occur as a result of use of force, they are likely to be relatively minor. Alpert and Dunham (see chapter 5) observed that the most common injury to a suspect was a bruise or abrasion (48 percent), followed by laceration (24 percent). The kinds of police actions that most captivate the public's concerns, such as fatal shootings, severe beatings with fists or batons that lead to hospitalization, and choke holds that cause unconsciousness or even death, are not typical of situations in which police use force. These findings reassure us that most police exercise restraint in the use of force, even if one has concerns over the number of times that police resort to serious violence. From a police administrator's point of view, these findings are predictable. Officers are trained to use force progressively along a continuum, and policy requires that officers use the least amount of force necessary to accomplish their goals. Another affiliated finding is that police rarely use weapons. According to Garner and Maxwell (see chapter 4), 2.1 percent of adult custody arrests involved use of weapons by police. Chemical agents were the weapons most frequently used (1.2 percent of arrests), while firearms were the weapons least often used (0.2 percent of arrests). Most police departments collect statistics on all firearm discharges by officers. These data consistently show that the majority of discharges are accidental or are directed at animals. Only on infrequent occasions do police use their firearms against the public. One implication of these findings is that increased training in how to use standard police weapons will be of little value in dealing with day-to-day situations that involve use of force. Training, if it is to be effective in reducing the use of force, needs to focus on how to gain compliance without resorting to physical coercion. Use of force typically occurs when police are trying to make an arrest and the suspect is resisting. Research indicates that police are most likely to use force when pursuing a suspect and attempting to exercise their arrest powers. Furthermore, resistance by the public increases the likelihood that police will use force. These findings appear intuitively sound given the mandate that police have regarding use of force. Police may use force when it is necessary to enforce the law or to protect themselves or others from harm. The findings also seem logical in view of police training curriculums and departmental regulations. Alpert and Dunham (see chapter 5) find that police almost always follow the prescribed sequence of control procedures they are taught, except when suspect resistance is high, in which case they tend to skip the intermediate procedure. The conclusion that police are most likely to use force when dealing with criminal suspects, especially those who are resisting arrest, is based on four types of data: arrest statistics, surveys of police officers, observations of police behavior, and reports by the public about their encounters with police. Arrest statistics show that resisting-arrest charges often are involved in situations in which officers use force. The interpretation of this finding is ambiguous, however, because officers may bring such charges in an attempt to justify their actions against a suspect. Some commentators even would argue that resisting-arrest charges are a good indication that police officers acted inappropriately or illegally. Because we are relying on official reports by officers who are involved in use-of-force incidents, and because they have self-interest in presenting the situation in the most favorable light possible, we cannot rely on arrest records alone in determining what happened. Fortunately, other research is available to help clarify the situation. The pilot national household survey by BJS included a series of questions about the respondent's behavior during contact with police.[10] The preliminary analysis revealed that of the 14 respondents in the sample who reported that police used or threatened force against them, 10 suggested that they might have provoked the officer to use force. The provocative behaviors reported by suspects include threatening the officer, assaulting the officer, arguing with the officer, interfering with the arrest of someone else, blocking or interfering with an officer's movement, trying to escape, resisting being handcuffed, and resisting being placed in a police vehicle. Research by Alpert and Dunham (see chapter 5) confirms that criminal suspects are not always cooperative when it comes to arrest. In almost all (97 percent) cases in which police officers used force in a Florida jurisdiction, the suspect offered some degree of resistance. In 36 percent of use-of-force incidents, the suspect actively resisted arrest, and in one-quarter of the incidents the suspect assaulted the officer. The researchers observed that the most common type of suspect force was hitting or striking a police officer (44 percent). Garner and colleagues, after using statistical controls for more than 50 characteristics of the arrest situation, the suspect, and the police officer, found that forceful action by suspects was the strongest and most consistent predictor of use of force by police.[11] Furthermore, they found that while 22 percent of arrests involved use of force by police, 14 percent of arrests involved use of force by suspects. Police officers in Phoenix completed a use-of-force survey after each arrest to generate these data. Finally, Bayley and Garofalo tallied 36 instances of force used by police or suspects out of 467 police-public encounters observed firsthand by researchers.[12] They found that in 31 incidents police used force against suspects and in 11 incidents suspects used force against police. One implication of the research is that the decision to use some level of force probably has legal justification in most cases. Force is likely to be used when suspects resist arrest and attempt to flee. Also, in a significant number of instances, suspects use force against the police. These findings leave open the issue of excessive force, since issues of proportionality are not clearly addressed. However, the findings do suggest that many debates over excessive force will fall into gray areas where it is difficult to decide whether an officer acted properly, because there is credible evidence that the use of force was necessary. What We Know With Modest Confidence About Police Use of Force Use of force appears to be unrelated to an officer's personal characteristics, such as age, gender, and ethnicity.A small number of studies suggest that use of force by police is not associated with personal characteristics, such as age, gender, and ethnicity Bayley and Garofalo concluded that use of force is not related to age, although it may be related to experience.[13] Worden, in an analysis of observational data on 24 police departments in 3 metropolitan areas, concluded that the personal characteristics of police officers do not have a substantively significant effect on use of force.[14] Likewise, Garner and colleagues reported that the race of suspect and officer is not predictive of use of force.[15] However, they found that incidents involving male police officers and male suspects are more likely to involve force. Alpert and Dunham (see chapter 5) found that officer characteristics are of little utility in distinguishing between force and nonforce incidents. Hence, gender and ethnicity appear unrelated to use of force. Given the limited research in this area, these conclusions should be accepted with caution and additional verification of these findings is needed. It is widely accepted in criminology that violence, along with a wide variety of other risk-taking and norm-violating behaviors, is a young man's game. Thus, we should expect that young, male police officers should use force more than their female colleagues or older officers. The fact that this is not clearly the case seems surprising. A lack of relationship between age and gender, on the one hand, and use of force, on the other, may be a function of police hiring and deployment practices. Retirement plans keep the age of police officers lower than that of most other occupations, and seniority, which is derivative of work experience, often brings more choice in work assignments, including duties that limit one's contact with criminal suspects on the street. Both these tendencies serve to constrain variation in the age of police officers who are exposed to potentially violent situations. This may attenuate the relationship between age and use of force. However, it is equally plausible that young male officers are assigned to high-crime areas where frequent use of force is necessary to gain compliance. Finally, it is possible that exposure to the police culture works to encourage the use of force, thus counterbalancing the decline in aggressivity that comes with age as demonstrated in criminological studies. More research is needed to disentangle these relationships. The finding that an officer's race is unrelated to the propensity to use force runs counter to the argument that racial animosity lies at the heart of police abuse. Indeed, Alpert and Dunham's research (see chapter 5) indicates that officers are more likely to use force against suspects of their own race. The lack of relationship between race and use of force, as well as between gender and use of force, is probably disheartening to those who argue that integration of police agencies along racial and gender lines will do much to reduce the incidence of police violence. Again, more research is needed to understand the situation of minority and female police officers with regard to their use of force. Use of force is more likely to occur when police are dealing with persons under the influence of alcohol or drugs or with mentally ill individuals. More research is needed. Police come across a wide variety of situations in their work. They encounter problems that range from relatively minor to serious to potentially deadly. They also interact with people exhibiting various mental states, including persons who are hysterical, highly agitated, angry, disoriented, upset, worried, irritated, or calm. Two situations that often give police officers cause for concern are when suspects appear to be under the influence of alcohol or drugs and when civilians appear to suffer from serious mental or emotional impairments. The concern stems from the fact that in such situations a person's rational faculties appear impaired. In dealing with problem situations, officers most often talk their way, rather than force their way, into solutions. For this reason, when a civilian is in a highly irrational state of mind, the chances of the police officer having to use force presumably increase and the possibility of injury to both officer and civilian increases as well. Research carried out for the President's Commission on Law Enforcement and Administration of Justice observed that alcohol use by either a suspect or an officer increased the chances that force will be used.[16] Garner and colleagues found that alcohol impairment by suspects was a consistent predictor of police use of force, while drug impairment predicted increased use of force for some but not all measures of use of force.[17] In contrast, Alpert and Dunham (see chapter 5) observed that alcohol or drug impairment of suspects was unrelated to police use of force or subsequent injury. That finding is interesting because, although impaired civilians did not demonstrate an increased propensity to resist an officer's actions, when they did resist they were more inclined to do so by actively resisting or assaulting the officer. Part of the disparity in findings between the President's Commission's research and more recent studies may be attributed to the fact that police officers today are better trained in how to deal with impaired civilians. Most police officers now receive training in a variety of violence reduction techniques, and this development is partly attributable to concerns over the President's Commission's findings and over the frequency with which police now are called to respond to large-scale violence, such as riots. Questions about how police deal with civilians who appear to have impaired mental states are important from administrative and practical points of view. Police officers are expected to exercise restraint in dealing with impaired civilians, while at the same time they need to be cautious about protecting their safety as well as the safety of other civilians. This puts them in a precarious situation, one in which mistakes of judgment or tactics can have grave consequences. From a practical standpoint, police regularly encounter civilians with impaired mental states, which makes the problem more than academic. Alpert and Dunham (see chapter 5) found that in 42 percent of use-of-force situations, suspects appeared to be under the influence of alcohol or drugs. Overall, the research on whether police use force more frequently in relation to civilians with impaired mental states is inconsistent. Further investigation, with an emphasis on implications for training, could reduce the risk of force and injury for both police officers and civilians. A small proportion of officers are disproportionately involved in use-of-force incidents. More research is needed. We often are told that a small number of people are responsible for most of the productive or counterproductive work in an organization. For example, we hear about the 80/20 rule in organizational management. That is, 20 percent of the workers account for 80 percent of the work. Policing has its counterpart explanation for deviant or illegal behavior. It is called the rotten apple or rogue officer theory, and it is often used to explain police corruption. Recently, a variation of this theory has become the principal explanation for use-of-force problems in police departments. In this context, we speak of "violence prone" police officers and we point to these individuals as the reason why a department has problems with the use of force.[18] People with extraordinary work performance, either good or bad, are noticeable when compared with their colleagues, and their salience leads us to think that their work is highly consequential to the good fortunes or misfortunes of an organization. The utility of this perspective for police managers attempting to deal with illegitimate use of force lies in the presumed concentration of problem behaviors in the work force. If only a handful of police officers accounts for most of the abuses, then effective solutions targeted at those individuals should deal with the problem. The nature of the solution, be it employee selection, training, oversight, or discipline, is less important than its degree of effectiveness and its ability to be directed at the problem group of employees. The Christopher Commission, which investigated the Los Angeles Police Department subsequent to the Rodney King incident, highlighted the "violence prone" officer theory.[19] The Commission, using the department's database, identified 44 officers with 6 or more civilian allegations of excessive force or improper tactics in the period 1986 through 1990. For the 44, the per-officer average for force-related complaints was 7.6 compared with 0.6 for all officers identified as having been involved in a use-of-force incident for the period January 1987 through March 1991. The 44 officers were involved in an average of 13 use-of-force incidents compared with 4.2 for all officers reported to be using force. Put another way, less than one-half of 1 percent of the department's sworn officers accounted for more than 15 percent of allegations of excessive force or improper tactics. The degree of disproportion (30:1) is striking and suggests that focusing efforts on a handful of officers can eliminate roughly 1 out of 7 excessive force incidents. This finding has led many police departments to implement early warning systems designed to identify high-risk officers before they become major problems. Most of these systems use administrative records, such as disciplinary records and citizen complaints, to monitor officer performance for possible problems. The concept of an early warning system for risk management of problem police officers is not new. In the early 1980s, a report on police practices by the United States Commission on Civil Rights found that "'(e)arly warning' information systems may assist the department in identifying violence-prone officers."[20] Consequently, it was recommended that "(a) system should be devised in each department to assist officials in early identification of violence-prone officers."[21] Until recently, these systems received limited acceptance, owing in part to concerns over possible abuses. The abuses include use of inaccurate information, improper labeling of officers, misuse of confidential records regarding discipline and other personnel matters, and social ostracism by peers and community for officers identified as problematic. There also were concerns about limited resources and about increased legal liability for the organization and individual officers. As Toch observes, the violence-prone officer paradigm often is based on a variety of loosely articulated theories of violent behavior.[22] The theories include concepts such as racial prejudice, poor self-control, and ego involvement. Furthermore, these theories often overlook the possibility that greater-than-average use of force may be a product of situational or organizational characteristics. For example, an officer's work assignment may involve a high-crime area that contains a high proportion of rebellious offenders. Also, divisive, dehumanizing views of the world, such as "us-them" and "good guy-bad guy," that facilitate violent behavior may be supported by the organizational culture. Further, administrative views of work roles and products, communicated formally or informally, that emphasize crime control through aggressive police behavior may encourage confrontational tactics that increase the chances of violent behavior by either civilian or police officer. Unless the reasons for violence propensity are accurately identified, the effectiveness of interventions targeted at violent police officers is a hit-or-miss proposition. Of the 44 officers identified by the Christopher Commission in 1991, 14 subsequently left the department as of October 1997. Of the 30 remaining officers, two had a use-of-force complaint that was sustained after review between 1991 and 1997.[23] This low number may be due to a variety of reasons, such as difficulties in sustaining citizen complaints, reassignment of work duties, negative publicity leading to a change in behavior, or greater circumspection when engaging in misconduct. However, the finding also may reflect regression to the mean. This is a statistical phenomenon postulating that extreme scores gravitate toward the mean or average score, thereby becoming less extreme over time. For example, groups of police officers who receive many citizen complaints, or who are disproportionately involved in the use of force, or who frequently are given poor performance ratings, will tend to become "better" over time, in the sense of statistically looking more like the "average" officers, even if nothing is done about these problems. Statistical regression represents a serious threat to the validity of early warning systems based on the assumption that extreme patterns of behavior persist over extended periods of time. What We Do Not Know About Police Use of Force The incidence of wrongful use of force by police is unknown. Research is critically needed to determine reliably, validly, and precisely how often transgressions of use-of-force powers occur. We do not know how often police use force in ways that can be adjudged as wrongful. For example, we do not know the incidence of excessive force, even though this is a very serious violation of public trust. We could pull together data on excessive force using police disciplinary records and court documents, for example, but the picture would be sketchy, piecemeal, and potentially deceiving. When it comes to less grave or less precise transgressions, such as "improper," "abusive," "illegitimate," and "unnecessary" use of force, the state of knowledge is even more precarious. In discussing this issue, we will concentrate on excessive force, because these transgressions are of utmost concern to the public and because well-established professional and legal criteria are available to help us evaluate police behavior. Notwithstanding a generally agreed-upon terminology, we should recognize that developing a count of excessive force that is beyond all dispute is an unworkable task. This is so because difficult judgments are involved in deciding whether use of force fits the criteria for these categories in a given situation, and reasonable people will disagree in such judgments. We clearly need more accurate, reliable, and valid measures of excessive force if we are to advance our understanding of these problems. Academics and practitioners both tend to presuppose that the incidence of excessive force by police is very low. They argue that, despite their shortcomings, agency statistics provide a useful picture of the use-of-force problem. These statistics show that most officers do not engage in force on a regular basis, that few people are injured by police use of force, that only a small number of people complain about police misconduct involving use of force, and that only a handful of these complaints are sustained. The argument has appeal. We believe that the vast majority of police officers are professionals who respect the law and the public. If use of force is uncommon, civilian complaints are infrequent, and civilian injuries are few, then excessive force by police must be rare. That conclusion may indeed be correct, but to the extent that it hinges on official police statistics, it is open to serious challenge. Current indicators of excessive force are all critically flawed. The most widely available indicators are civilian complaints of excessive force and civil lawsuits alleging illegal use of force. Civilian complaints of excessive force are infrequent, and the number of substantiated complaints is very low. These figures are consistent with the argument that excessive force is sporadic. However, complaint mechanisms are subject to selection and reporting biases, and the operation of complaint systems, which typically is managed by police, wields considerable influence on whether people will come forward to complain. Civil lawsuits against police are exceedingly rare relative to the number of times that police use force. Because the legal process is highly selective in terms of which claims get litigated, lawsuits are a very unreliable measure of illegal use of force. With both civilian complaints and lawsuits, small changes in administrative practices can have a large impact on the magnitude of the problem measured in these ways. The difficulties in measuring excessive and illegal force with complaint and lawsuit records have led academics and practitioners to redirect their attention to all use-of-force incidents. The focus then becomes one of minimizing all instances of police use of force, without undue concern as to whether force was excessive. From this perspective, other records, such as use-of-force reports, arrest records, injury reports, and medical records, become relevant to measuring the incidence of the problem. From a theoretical perspective, understanding all use-of-force incidents helps us to put wrongful use of force in perspective. However, because political, legal, and ethical issues are very serious when we are dealing with excessive force, pressures to know the incidence and prevalence of these events with precision will always be present. As a corollary of our current inability to measure excessive force, we cannot discern with precision changes in the incidence of these events over time and across places. This means that we can neither determine whether excessive force problems are getting better or worse nor determine the circumstances under which those problems are more or less severe. The impact of differences in police organizations, including administrative policies, hiring, training, discipline, and use of technology, on excessive and illegal force is unknown. Research is critically needed in this area. A major gap in our knowledge about excessive force by police concerns characteristics of police agencies that facilitate or impede this conduct. Although many of the conditions that arguably lead to excessive or illegal force by police seem obvious, or appear to be a matter of common sense, we still greatly need systematic research in this area. We need to know, for example, which organizational characteristics are most consequential, which characteristics take on added significance in various environments, and which characteristics are redundant or derivative of other characteristics. Many formal aspects of the organization--such as hiring criteria, recruit training, in-service programs, supervision of field officers, disciplinary mechanisms, operations of internal affairs, specialized units dealing with ethics and integrity, labor unions, and civilian oversight mechanisms-- plausibly are related to levels of officer misconduct. It makes sense that poorly educated, badly trained, loosely supervised, and inadequately disciplined officers are likely to be problematic, and that when such officers are in the majority, the organization is on the road toward disaster. Yet, we lack research that systematically addresses these questions. Less formal aspects of police organizations--officer morale, administrative leadership, peer culture and influence, police-community relations, relations with other government agencies, and neighborhood environments--also plausibly have a part in levels of officer misconduct. Alienated officers who do not have a clear vision of their role and responsibilities and who are working in disorganized agencies and interacting with the public under stressful circumstances probably are more likely to abuse their authority, including their authority to use force. Research that systematically addresses these questions is lacking. Methodological investigation of relations between organizational elements and use-of-force transgressions will help explain police misconduct at a theoretical level. More importantly, research on these questions will allow us to deal effectively with police misbehavior. Faced with serious misconduct problems in a police agency, we need to focus scarce resources on those aspects of police organizations that are most clearly related to ensuring proper conduct of officers with regard to use of force. Generalized efforts to reform police organizations that are expected to reduce misconduct problems tend to be inefficiently focused and thus appear clumsy, inadequate, and misinformed. Research must focus on establishing the relative cost-effectiveness of various strategies to reduce or eliminate police misconduct. Furthermore, only strategies that are solidly grounded in theory, practice, and empirical research will provide reliable solutions with predictable costs and benefits. Influences of situational characteristics on police use of force and the transactional nature of these events are largely unknown. More research is necessary. Research on police-citizen encounters reveals that use of force by police is situational and transactional. That is, police respond to circumstances as they first encounter them and as they unfold over time. For example, Bayley and Garofalo observed that the situations most likely to involve police use of force are interpersonal disturbance and violent personal crime.[24] Beyond this, however, we do not know much about the types of events that enhance the likelihood that police will use force. Similarly, we have noted that when suspects attempt to flee or physically resist arrest police are more likely to use force. We also noted that in many cases both police and suspects use force against each other. However, these findings do not address the transactional nature of police-public encounters in that they do not describe the step-by-step unfolding of events and interactions. Knowing that police use force if suspects physically resist arrest, it matters if police use force without provocation and the suspect responds by resisting or vice versa. A variety of situational elements plausibly are related to police use of force. If police are called to a scene where there is fighting, they may have to or believe they have to use force to subdue the suspects. If they are called to a domestic dispute where emotions are running high, they may have to or believe they have to use force to gain control of the situation. If they are called to intercede with a civilian who is recklessly brandishing a weapon, they may have to or believe they have to use force to protect themselves and others. Use of force in such circumstances may be justifiable, but to the extent that it is predictable, we can prepare officers for these encounters and devise alternative strategies that minimize or eliminate the use of force. Some situational factors may increase the chances that force of questionable legitimacy will be used. For example, officers sometimes use force on the slightest provocation following a high-speed car chase, when adrenaline levels are high. They may use force more frequently when they are alone, because they feel more vulnerable or believe that they can get away with it. They may use force more frequently as a way of emphasizing their authority when suspects are disrespectful or when there is a hostile audience to the encounter. At this point, however, knowledge about the types of police-citizen encounters in which police are likely to use force is rudimentary. Police-public encounters are transactional in the sense that all the actors in a situation contribute in some way to its development and outcome. Understanding the transactional nature of police use of force is important because it emphasizes the role of police actions in increasing the chances that force will be used. From this perspective, it is possible to minimize the use of force by modifying the behavior and tactics of police officers. By understanding the sequences of events that lead police to use force, we can gain a greater degree of control over those situations and possibly redirect the outcome. But we have only a basic understanding of the transactional nature of use-of-force situations, despite the fact that sequences of actions and interactions are highly germane to determining whether use of force was excessive or illegal. Organization of the Report The next four chapters of this report focus on major research studies dealing with police use of force. They represent significant projects currently under way to understand police use of force. Two of the projects are attempts at measuring the incidence of police use of force nationwide. BJS has developed a national-level data collection effort using a household survey methodology to investigate police-public interactions, with a component on use-of-force issues (see chapter 2). IACP is collecting data on police use of force through a voluntary reporting system (see chapter 3). Two other projects on police use of force involve citywide investigations across several locations. Chapter 4 reports on research in six jurisdictions; the research is important because it identifies factors associated with use of force and because it addresses difficult measurement issues. Focusing on three police agencies, chapter 5 discusses research that centers on the use of force by both police and suspects; the research is important because it contributes significantly to understanding the transactional nature of police-citizen encounters. The final chapter outlines suggested directions for future research. A selected bibliography concludes this report. Notes 1. Bierce, Ambrose, The Devil's Dictionary, New York: Dover, 1958: 101. 2. "Justice Department Consent Decree Pushes Police to Overhaul Operations," Pittsburgh Post-Gazette, March 1, 1998, C-1. 3. Based on an investigation in 14 cities, Human Rights Watch described the brutality situation as follows: "(p)olice officers engage in unjustified shootings, severe beatings, fatal chokings, and unnecessarily rough physical treatment in cities throughout the United States, while their police superiors, city officials and the Justice Department fail to act decisively to restrain or penalize such acts or even to record the full magnitude of the problem." Human Rights Watch, Shielded from Justice: Police Brutality and Accountability in the United States, New York: Human Rights Watch, 1998: 1, 27. 4. A previous summary of research on police use of force can be found in McEwen, Tom, National Data Collection on Police Use of Force, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, April 1996, NCJ 160113. 5. Bayley, David H., and James Garofalo, "The Management of Violence by Police Patrol Officers," Criminology, 27(1)(February 1989): 1-27; and Bayley, David H., and James Garofalo, "Patrol Officer Effectiveness in Managing Conflict During Police-Citizen Encounters," in Report to the Governor, Vol. III, Albany: New York State Commission on Criminal Justice and the Use of Force, 1987: B1-88. 6. Adams, Kenneth, "Measuring the Prevalence of Police Abuse of Force," in And Justice For All: A National Agenda for Understanding and Controlling Police Abuse of Force, ed. William A. Geller and Hans Toch, Washington, DC: Police Executive Research Forum, 1995: 61-97. 7. Greenfeld, Lawrence A., Patrick A. Langan, and Steven K. Smith, Police Use of Force: Collection of National Data, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, November 1997, NCJ 165040. 8. Ibid. 9. Pate, Anthony M., and Lorie A. Fridell, with Edwin E. Hamilton, Police Use of Force: Official Reports, Citizen Complaints, and Legal Consequences, Vols. I and II, Washington, DC: The Police Foundation, 1993. 10. Greenfeld, Lawrence A., Patrick A. Langan, and Steven K. Smith, Police Use of Force: Collection of National Data. 11. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding Use of Force By and Against the Police, Research in Brief, Washington, DC: U.S. Department of Justice, National Institute of Justice, November 1996, NCJ 158614. 12. Bayley, David H., and James Garofalo, "The Management of Violence by Police Patrol Officers"; and Bayley, David H., and James Garofalo, "Patrol Officer Effectiveness in Managing Conflict During Police-Citizen Encounters." 13. Ibid. 14. Worden, Robert, "The 'Causes' of Police Brutality," in And Justice For All: A National Agenda for Understanding and Controlling Police Abuse of Force, 31-60. 15. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding Use of Force By and Against the Police. 16. Reiss, Albert J., Jr., Studies on Crime and Law Enforcement in a Major Metropolitan Area, President's Commission on Law Enforcement and Administration of Justice, Field Survey No. 3, Washington, DC: U.S. Government Printing Office, 1967. 17. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding Use of Force By and Against the Police. 18. Toch, Hans, "The 'Violence-Prone' Police Officer," in And Justice For All: A National Agenda for Understanding and Controlling Police Abuse of Force, 99-112. 19. Independent Commission on the Los Angeles Police Department, Report of the Independent Commission on the Los Angeles Police Department, Los Angeles, CA: Independent Commission on the Los Angeles Police Department, 1991. 20. United States Commission on Civil Rights, Who's Guarding the Guardians? A Report on Police Practices, Washington, DC: United States Commission on Civil Rights, 1981: 159. 21. Ibid. 22. Toch, Hans, "The 'Violence-Prone' Police Officer," 112. 23. Office of the Inspector General, Los Angeles Police Commission, "Status Update: Management of LAPD High-Risk Officers," Los Angeles: Los Angeles Police Commission, 1997. 24. Bayley, David, H., and James Garofalo, "Patrol Officer Effectiveness in Managing Conflict During Police-Citizen Encounters." ------------------------------- Sidenotes Kenneth Adams, Ph.D., is Associate Professor and Chair of the Criminal Justice Faculty, School of Public and Environmental Affairs, Indiana University-Indianapolis. "Thus, the Commission concludes that factors substantially contributing to misperceptions about use of physical and deadly force by law enforcement officers include...[f]ailure to appreciate the relative infrequent use of physical and deadly force by law enforcement personnel...." --New York State Commission on Criminal Justice and the Use of Force, Report to the Governor, Vol. 1, New York: New York State Commission on Criminal Justice and the Use of Force, May 1987: 6. "The first tactic used in an incident is nearly always the least severe use of force on the continuum, and the secondÉis nearly always the second-most lenient." --Alpert and Dunham, page 48 of this report. Regarding suspect force as a consistent predictor of police use of force: "This remained true when controlling for the possibility that some suspect use of force could be a reaction to police use of force." --Garner, et al. (see note 11). "Suspects reported as impaired were more than twice as likely than sober suspects to use a gun to resist the police." --Alpert and Dunham, page 51 of this report. ". . . a significant number of officers...repetitively misuse force and persistently ignore the written policies and guidelines of the Department regarding force. By their misconduct, this group of officers tarnishes the reputations of the vast majority of LAPD officers who do their increasingly difficult job of policing the City with courage, skill, and judgment." --Independent Commission on the Los Angeles Police Department, Report of the Independent Commission on the Los Angeles Police Department, Los Angeles, CA: Independent Commission on the Los Angeles Police Department, 1991: 31. ------------------------------- Working definitions Police use of force is characterized in a variety of ways. Sometimes, these characterizations are functionally interchangeable so that one can be substituted for another without doing injustice to the factual interpretation of a statement. At other times, however, differences in terminology can be very consequential to a statement's meaning. For example, "deadly force" refers to situations in which force is likely to have lethal consequences for the victim. This type of force is clearly defined and should not be confused with other types of force that police use. In contrast, "police brutality" is a phrase used to describe instances of serious physical or psychological harm to civilians, with an emphasis on cruelty or savageness. The term does not have a standardized meaning; some commentators prefer to use a less emotionally charged term. In this report, the term "excessive force" is used to describe situations in which more force is used than is allowable when judged in terms of administrative or professional guidelines or legal standards. Criteria for judging excessive force are fairly well established. The term may also include within its meaning the concept of illegal force. Reference also is made to "excessive use of force," a similar, but distinctly different, term. Excessive use of force refers to high rates of force, which suggest that police are using force too freely when viewed in the aggregate. The term deals with relative comparisons among police agencies, and there are no established criteria for judgment. "Illegal" use of force refers to situations in which use of force by police violated a law or statute, generally as determined by a judge or magistrate. The criteria for judging illegal use of force are fairly well established. "Improper," "abusive," "illegitimate," and "unnecessary" use of force are terms that describe situations in which an officer's authority to use force has been mishandled in some general way, the suggestion being that administrative procedure, societal expectations, ordinary concepts of lawfulness, and the principle of last resort have been violated, respectively. Criteria for judging these violations are not well established. To varying degrees, all of the above terms can be described as transgressions of police authority to use force. ------------------------------- Chapter 2 Revising and Fielding the Police-Public Contact Survey by Lawrence A. Greenfeld, Patrick A. Langan, and Steven K. Smith To learn more about police use of force requires an understanding of the reasons for and the results of police-public encounters. As a step toward developing that understanding, the Bureau of Justice Statistics (BJS) fielded a pilot test in 1996 of the Police-Public Contact Survey (PPCS). Its primary purpose was to obtain information to help guide future development of a final questionnaire on the topic. This chapter briefly reviews the 1996 survey, discusses improvements incorporated into a second pilot survey, and describes other planned BJS efforts to learn more about police-public encounters. First Pilot Test of the Survey Questionnaire BJS supplemented the National Crime Victimization Survey (NCVS) with a pilot test of PPCS during May, June, and July 1996. The objective was to collect answers from respondents to a series of questions about the nature and consequences of their face-to-face interactions with police. (NCVS is based on interviews conducted with a nationally representative sample of U.S. households and has become a highly useful platform for testing new questionnaires and periodically implementing special supplements. The NCVS sample consists of all household members age 12 or older residing in more than 40,000 U.S. households; each household member is interviewed twice during a calendar year, resulting in about 200,000 interviews annually.) The PPCS pilot test involved interviews with 6,421 persons during the 1996 trial period. The respondents were asked about their contacts with police during the 12 months prior to the interviews. Respondents interviewed in May 1996 were asked about contacts that occurred anytime during the period June 1995 to May 1996; those interviewed in June 1996 were asked about contacts between July 1995 and June 1996; and interviews in July 1996 covered the period from August 1995 to July 1996. On average, the 12-month reference period included 6 months in 1995 and 6 in 1996. Administration of PPCS went smoothly. Among persons who had no contact with police, the interview took 1 minute, on average, to complete. Among those who had police contact, the interview averaged 10 minutes. Findings from the first PPCS were reported in 1997 in the BJS-National Institute of Justice publication Police Use of Force: Collection of National Data.[1] (See sidebar "Selected findings from the Police-Public Contact Survey.") The 1996 PPCS was intended as a pretest of the questionnaire; the survey and its findings were not meant to be viewed as a source of indepth or precise statistics on police use of force. Rather, survey findings provided empirical information to help guide future development of an improved questionnaire. Second Pilot Test of the Survey Questionnaire During the first half of 1999, BJS focused on improving the content and administration of the PPCS questionnaire. Three areas of the instrument were improved: items relating to screen questions used to identify whether a contact occurred, the type of contact, and circumstances surrounding the contact; the determination of whether force was used or threatened, the type of force used or threatened, the circumstances surrounding its use, and provocative actions by the respondent; and the respondent's perceptions of the appropriateness or inappropriateness of police conduct during the contact. Contact screen questions The pretest revealed a number of broad categories of types of face-to-face contacts that needed encoding on the questionnaire. A number of respondents reporting contacts indicated that their employment brought them into regular contact with police. This was so for hospital workers, tow-truck drivers, and lawyers, among others. In addition, respondents indicated that they had contacts with police as the result of appearing in court as jurors or witnesses. A major area for further development of the screening elements of PPCS relates to traffic stops. One of the major findings from the initial PPCS field test was that many of the face-to-face contacts the public has with police are in connection with traffic stops. They may represent an area of potentially argumentative interactions between police and the public and result in use-of-force incidents. This is likely to be true especially if such stops are thought to be motivated by factors not strictly related to law enforcement purposes. Recent studies in New Jersey and Maryland suggest that black motorists may be more likely than others to be stopped by police.[2] The PPCS questionnaire has been revised to capture more detailed information about traffic stops, including more about the nature and frequency of traffic stops involving persons of different races. To gather more information on the nature and consequences of traffic stops, BJS has added a group of questions to the PPCS. Of particular interest will be an effort to determine from respondents what they believe to be the basis for stops, the reason given by the officers for stops, and the content of the interaction during stops, such as tickets issued, warnings, verbal questioning, searches, and arrests. Information gathered from the survey can be used to build on prior research regarding the treatment of different categories of the populace by the criminal justice system. Those traffic stops resulting in handcuffing, threats, or use of force will also be identified. For such incidents, new items have been added to the survey to ascertain potentially provocative behaviors or drug or alcohol use by respondents. Several other modifications to the questionnaire were needed. For example, existing questions about respondents' traffic-related contacts with police were modified to determine whether such respondents had been passengers or drivers. In addition, screen questions were added to determine explicitly who initiated contact between respondents and police officers. The 1996 questionnaire provided information only on the type of contact; who initiated it had to be inferred (e.g., "Received a traffic or parking violation" implies the officer initiated the contact). Determining whether force was used The 1996 PPCS asked respondents to indicate whether force was threatened, used, or not used during contacts. BJS replaced this and substituted a list of police behaviors that might have occurred, including threatening or restraining respondents during encounters. Handcuffing, considered a standard police practice, is listed within the enumerated group of police behaviors and will not be treated in the questionnaire separately from other behaviors that might have occurred during contacts. Paralleling these queries about police behavior during contacts are questions about the respondents' behavior during encounters to learn about provocative actions that may have occurred. Such questions will apply to all respondents, not merely to those indicating force had been used. Experience with a pat-down or frisking was asked only of those respondents who indicated that in contacts with the police, they believed that they were suspects in a crime. This was too limiting, and BJS has expanded the use of this item to all respondents. Similarly, questions relating to subsequent criminal charges resulting from contacts were limited to respondents who believed that the contacts occurred because police considered them suspects in a crime. BJS will ask such questions of all respondents. Respondents' characterizations of incidents The 1996 PPCS asked respondents to characterize whether they believed police behavior was proper or improper during use-of-force incidents. Such a question will be asked of all respondents who have police contact. In addition, the 1996 PPCS did not ask those who experienced use of force whether they perceived police behavior to have been excessive for the circumstances; BJS has added such an item to the next version of the survey. Survey Administration for Second Pilot Test BJS and the U.S. Bureau of the Census began the 1999 pilot test using the revised PPCS in July. The fieldwork will continue through December 1999. This will result in a national sample of respondents about 10 times the size of the first pilot test. Prior to the upcoming test, BJS redesigned the questions as noted above, circulated them for comment, and secured Office of Management and Budget clearance for the use of the supplement for the full NCVS national sample. In addition, BJS and the Bureau of the Census conducted extensive testing of the revised instrument with volunteer respondents in the Census Bureau's Cognitive Research Laboratory. Additional Efforts to Learn About Police-Public Encounters During the next 3 to 5 years, BJS will be undertaking the regular periodic surveys conducted among nationally representative samples of those confined in local jails and prisoners held by State and Federal authorities. BJS anticipates introducing new survey items that will gather information about respondents' interactions with police during the arrest preceding their incarceration. The new items will provide, for the first time, information on the content of those contacts between police and criminals, including the methods by which they were apprehended, use of force during such events, and provocative behaviors by known offenders. Notes 1. Greenfeld, Lawrence A., Patrick A. Langan, and Steven K. Smith, Police Use of Force: Collection of National Data, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, November 1997, NCJ 165040. 2. See Lamberth, John, "Driving While Black: A Statistician Proves That Prejudice Still Rules the Road," Washington Post, August 16, 1998, C1. ------------------------------- Sidenotes Lawrence A. Greenfeld, Patrick A. Langan, Ph.D., and Steven K. Smith, Ph.D., are statisticians with the Bureau of Justice Statistics, a component of the Office of Justice Programs within the U.S. Department of Justice. Screen questions are designed to determine quickly at the beginning of an interview whether a respondent needs to be asked subsequent questions. For example, if a respondent had no police contact during the specified period, subsequent questions about the nature, content, and consequences of police-public encounters would be omitted. ------------------------------- Selected findings from the Police-Public Contact Survey In 1996 the Bureau of Justice Statistics conducted a pilot test of the Police-Public Contact Survey. Although not intended as a source of detailed or precise statistics on police use of force, the pilot survey did provide preliminary estimates of the prevalence of the public's contact with police, including contacts during which police used force. Among the survey's findings are the following: --An estimated 44.6 million persons (one in five U.S. residents age 12 or older) had face-to-face contacts with police officers during the prior 12 months. Men, whites, and people in their twenties were the most likely to have those contacts. --An estimated 33 percent of residents who had contact with police had either asked for assistance from officers or provided it to them. About 32 percent of those who had contact with police had reported a crime, either as a victim or witness. --For nearly half of those with contacts, the encounters were initiated by the public. For just under one-third of persons with contacts, police initiated them. --Age is an important factor in both the frequency and type of police contacts experienced. Young people were the least likely to initiate contact with police (their contacts most often were police initiated), while persons age 60 or older were the most likely to initiate contacts with officers. --About 1 percent of people reporting contacts with police indicated that officers used force or threatened force. In the majority of those instances, respondents said that their own actions, such as threatening police or resisting arrest, may have provoked officers. ------------------------------- Chapter 3 IACP National Database Project on Police Use of Force by Mark A. Henriquez Individual police agencies, law enforcement ----organizations, and some States have long considered collection of use-of-force data as a means to better serve their communities and to better distinguish between misperception and reality of police use-of-force issues. Reflecting this, Congress enacted legislation in 1994 that, among other things, directed the U.S. Attorney General to collect data on police use of excessive force. In 1995 the Bureau of Justice Statistics (BJS) and the National Institute of Justice (NIJ)--components of the Office of Justice Programs (OJP) within the U.S. Department of Justice (DOJ)--proposed to cofund development of a national use-of-force database by the International Association of Chiefs of Police (IACP). One purpose of such a database would be quantification of the extent and types of force used by police. Following approval of first-year funding, initial project activities began in September 1996. Early in 1997, NIJ and BJS cofunded the project for another year. Since early 1998, IACP has been responsible for all project funding. The balance of this chapter discusses basic concepts underlying IACP's use-of-force database project, highlights selected preliminary findings derived from the database, and discusses the future of the project. The chapter's addendum provides details about key elements involved in database development. Basic Concepts Underlying the Project Essential to creating IACP's police use-of-force database was the belief that data contributions should be voluntary and anonymous. Also of basic importance was that the definition of "police use of force" should reflect operational realities of modern, street-level law enforcement. Voluntary and anonymous reports Police use of force is an extremely sensitive issue, in part because agencies and the public alike harbor preconceptions and because data and resulting reports could be used or interpreted either accurately or inaccurately. For example, the legislation directing the U.S. Attorney General to "acquire data about the use of excessive force by law enforcement officers" also makes deprivation of civil rights unlawful as evidenced by "pattern or practice" and allows the Attorney General, through civil action, to "obtain appropriate equitable and declaratory relief to eliminate pattern or practice." Thus, if use-of-force data provided to IACP from the field-- whether standing alone or accompanied by analysis--were associated with individual departments, litigation could result. Under those circumstances, the law enforcement community would have faced the dilemma of being very reluctant to provide use-of-force data, yet realizing that fundamental elements of modern police service are pursuit of truth and subordination to the will of the communities that agencies are sworn to serve. Therefore, IACP decided that provision of such data would be both voluntary and anonymous to promote accurate reporting and overcome potential reluctance of agencies to participate. Reflecting operational realities When examined from the perspective of day-to-day law enforcement activities, many previous use-of-force definitions were considered not sufficiently workable or functional to be applied nationwide to all jurisdictions and department types for the purposes of the IACP project. Thus, IACP designed the project from the outset to reflect operational realities of modern, street-level law enforcement, including the very meaning of "police use of force," defined as the amount of force required by police to compel compliance by an unwilling subject. The project team developed a unique software package to track the basic types of force used by officers and suspects in typical encounters. On the basis of data submitted through the software, the team identified a "street continuum" of force, which consists of the types of force used by officers on a day-to-day basis to bring subjects under control. The IACP database consists of data on reported use of force stemming from police responses to calls for service, whether or not those responses resulted in arrests.[1] Categories of force include the following: --Physical force (use of fists, hands, feet, etc.). --Chemical force (the discharge of Mace, pepper spray, and similar agents). --Electronic force (the discharge of Tasers, stun guns, or other electronic weapons). --Impact force (use of batons and the like). --Lethal force (firearm discharge of any kind). Additionally, IACP software enables departments to track various subcategories of force, such as dog bites, edged weapons, vehicles, and nail guns. However, the project team excluded certain measures of force because they were considered to fly in the face of reality and practicality. (See sidebar "Measures of force excluded from the database.") Selected Preliminary Use-of-Force Findings Although the database is in the early stages of development, the level of interest and support from local police agencies is encouraging. (See sidebar "Agencies contributing to the use-of-force database.") Nonetheless, the data are not yet nationally representative. Conclusions about national use-of-force trends, therefore, should not be attempted at this time. But preliminary findings based on the data may provide useful insights, however tentative, into current police and subject use-of-force issues. The IACP calculation of the use-of-force rate is based on dispatched calls for service. For example, based on 1995 data reported by 110 agencies, the police use-of-force rate was 4.19 per 10,000 responded-to calls for service, or 0.0419 percent. Jurisdiction size for 78 of those agencies was 35,000 population or less; for 4 departments, 500,000 population or more. Data reported for 1995-97 indicate that of 62,411 use-of-force incidents during the period, about 87 percent involved officers using physical force.[2] Officers used chemical force in 7 percent of the incidents, firearms in about 5 percent. Data received by IACP through September 1998 from 26 agencies reporting for 1996 and 27 for 1997 on 2,310 police use-of-force incidents indicate that the vast majority of such incidents occurred in arrest-related situations. For the 1996-97 period, those agencies also reported the following: --Of 2,264 use-of-force confrontations for which the race of officers and subjects was known, 909 were intraracial (officer and subject of the same race) and 1,335 were interracial. --About 10 percent of 2,479 officers using force sustained injuries. Less than 1 percent of the injuries were major; none resulted in death. About 38 percent of the subjects were injured as the result of police use of force, including approximately 1.5 percent with major injuries. (Data spanning the 1995-97 period indicate that of 75,082 use-of-force incidents, 3,274, or about 4 percent, resulted in officer injuries, all but 39 minor.) --Of 3,972 reported incidents involving use of force, 20 resulted in complaints by subjects. Future of the Database From the inception of the database project in 1996, it has had--and is expected to continue to have--a three-tiered impact on police use-of-force policies and practices in the United States. At the national level, the IACP annual report Police Use of Force in America provides summary and incident information to police leaders, the public, and the media. At the State level, State associations of chiefs of police increasingly use the project to provide leadership to their constituent law enforcement agencies. Finally, local police agencies use IACP-provided software to capture use-of-force information pertaining to their own departments and use it for such purposes as the following: --To promote improved policies, training, and procedures governing departmentwide use of force. --To reduce potential department liability arising from frivolous or unfounded legal actions related to use of force. --To contribute data to statewide use-of-force data repositories, where applicable, and to the IACP database. The IACP database is a dynamic one. The number of departments providing new or updated data increases almost on a daily basis. For example, from a modest beginning of 7 pilot States, database software is now in place in more than 1,000 agencies nationwide and in the last quarter of 1998, Illinois and Rhode Island joined the project, followed by Maryland and Missouri in early 1999. Newly recruited States may contribute data from as far back as 1991. IACP believes that the benefits of its use-of-force database project will continue to increase as the number of contributing departments approaches national representation. Notes 1. Data submitted to IACP by any given agency may also reflect use-of-force incidents resulting from responses to other than calls for service--for example, from officer-initiated contacts with the public. 2. The data for 1996 and 1997 reflect agency use-of-force reports received through September 1998 and, therefore, are not yet complete. Although the IACP database project started in 1996, some agencies supplied data for 1995. ------------------------------- Sidenotes Mark A. Henriquez is Project Coordinator for the IACP National Police Use-of-Force Database Project. For a more detailed account of use-of-force project findings, see International Association of Chiefs of Police, Police Use of Force in America: Research in Progress Report on the IACP National Database Project, Alexandria, Virginia: International Association of Chiefs of Police, April 1998. ------------------------------- Measures of force excluded from the database IACP's project team decided to exclude from the database certain measures of force that the team considered were too broad to allow agency reporting in an accurate and timely fashion or beyond what police typically perceive or record as applications of force: --Presence of a police officer at the scene. --Presence of a K-9 at the scene. --Presence of chemical or electronic less-than-lethal devices at the scene. --Verbal commands by an officer. --Routine or voluntary handcuffing of prisoners for transport or during field questioning or investigation. --Display or presentation of an officer's weapon. Although any of the above measures could be and are included in some academic studies of police use of force, IACP excluded them from its database to allow creation of a concise, universally accepted, and practically achievable information base on police use of force in the United States. Inclusion of the elements listed above would have overly complicated the project and substantially reduced local agency participation. ------------------------------- Agencies contributing to the use-of-force database IACP has obtained a substantial level of use-of-force data from participating agencies. Hundreds of others are establishing deadlines for submitting data either to their State chiefs' organizations or to IACP directly: --About 4,000 agencies have requested the requisite software. --An estimated 1,000 agencies are using the software to capture use-of-force data. --Some 150 agencies are expected to contribute data for the 1998-99 data year. Although not nationally representative at this early stage of project development, the received data provide indicators regarding police use of force against subjects and subject use of force against officers. As of this writing, agencies representing populations ranging from 1,000 to more than 1 million have contributed data on thousands of incidents. Reported use-of-force incidents applicable to 1996 and 1997 total 24,383 and 24,033, respectively (based on data submitted through September 1998). ------------------------------- Addendum: Key Elements in Building the IACP Use-of-Force Database To ensure the success of the database project, IACP created two advisory panels to support initial project activities. The first was an ad hoc committee of police and justice leaders brought together at IACP headquarters to react to the project's scope and give general policy input and advice. This group consisted of DOJ representatives, State police, county sheriffs, local law enforcement officers and other criminal justice professionals. Consensus on key issues, such as definitions and data elements, was achieved during the project's start-up phase. A second and continuing advisory committee consists of representatives from each of seven pilot State associations, discussed below, and a representative from the U.S. Border Patrol (Immigration and Naturalization Service). This group, from the outset, has provided advice on local concerns, State association perspectives, and logistical issues. This group also provides midyear and end-year input on project direction and selected milestones. The SACOP role The State Associations of Chiefs of Police (SACOP) Division of IACP laid the foundation for this program from its inception. In particular, the Virginia Association of Chiefs of Police had already instituted a statewide use-of-force reporting program that served as the model for IACP's national effort. The strength and reach of SACOP organizations within most of the States provided a flexible framework for developing and expanding the project. In collaboration with SACOP leadership, IACP initially identified seven pilot State associations that would help coordinate the contribution of information for the national database effort: Arkansas, New Jersey, New York, Vermont, Virginia, Washington, and West Virginia. DOJ also arranged early Federal support by designating the U.S. Border Patrol to contribute data. Each pilot State association identified five local police departments to contribute data to the project. To accomplish the data collection and transfer, IACP developed a new software application specifically for this purpose. A three-tiered approach to data collection Data flow was designed from the outset to emanate from local departments, through the respective SACOP offices, and then to IACP. The identities of all participating departments are masked by use of a self- installing agency reporting code, which is automatically appended to all local agency data records. This code is the only means by which individual data records can be distinguished from one contributing source to another. The SACOP organizations have the capability to generate regional use-of-force data based on the anonymous records reported to them by departments in their States. Larger departments, which maintain electronic repositories of their own use-of-force data, have the option of reporting their data (in any data format) directly to IACP in cases where the appropriate SACOP office is not participating in the project. Even in these instances, the identity of the contributing agency is protected by an agency reporting code in the same manner that the identity of a department is protected when contributing data through its State organization. Software design and training Based on the large number of data elements relevant to this study (see sidebar "Database content"), it soon became clear that an automated data collection system was required. The ideal system would need to: --Be compatible across the widest possible spectrum of computers used in departments nationwide. --Employ a graphical user interface that would facilitate self-instruction by the user. --Handle the relationship between incidents involving a single officer and multiple subjects, multiple officers and a single subject, and any other possible combination. Staff decided on a system that could meet these needs and was scalable in that it could support the inevitable evolution of revisions, customizations, additions, and expansions inherent in software projects of this type. Because data flowed from local agencies to SACOP regional data repositories and from there to IACP, two versions of the software were produced. The first, the local agency version, was intended to automate data capture at the local department level. This software was equipped with a simplified data export function by which the user could write the captured use-of-force data to a floppy disk by simply clicking an on-screen button. The second version of the software was intended for use by the SACOP data repositories. The SACOP version of the software remains identical to the local agency version in terms of the number and type of data elements captured. However, the SACOP version is provided with an import capability by which data on disks originating from local agencies can easily be incorporated into the regional database. Data from each of the regional databases are periodically exported to floppy disk and sent to IACP for inclusion in its database. Several software improvements have been made. Training for IACP database administrators was provided in a formal class setting by IACP in August 1997. The 8-hour class provided hands-on technical training in setting up, operating, and troubleshooting the use-of-force software and in the philosophy underlying the project. At the conclusion of the training, students, all of whom were sworn law enforcement personnel from participating local police departments, were presented with certificates identifying them as IACP use-of-force database administrators. ------------------------------- Database content Guided by the broad framework of use-of-force issues and by IACP's experience with other sensitive police policy issues, the project team decided upon the following elements for inclusion in the use-of-force database: Department characteristics: --Report year. --Jurisdiction size. --Department type. --Calls for service. --Ethnicity demographics. --Number of use-of-force incidents by type. --Number of force-related complaints. --Complaint resolution. --Types of less-than-lethal weapons authorized. --Use-of-force training and policies in place. --Administrative policies for use-of-force complaint. Incident characteristics: --Incident time. --Incident date. --Incident year. --Number of officers involved. --Number of subjects involved. --Number of third parties involved. --Age of officer(s)/subject(s). --Type of assignment. --Duty status. --Education of officer(s)/subject(s). --Race/ethnicity of officer(s)/subject(s). --Type of force used by officer(s)/subject(s). --Officer/subject injury. Complaint characteristics: --Complaint time. --Complaint date. --Complaint year. --Number of officers involved. --Number of subjects involved. --Number of third parties involved. --Age of officer(s)/subject(s). --Type of assignment, duty status. --Education of officer(s)/subject(s). --Race/ethnicity of officer(s)/subject(s). --Type of force used by officer(s)/subject(s). --Officer/subject injury. --Previous complaints against officer. --Complaint disposition. ------------------------------- Chapter 4 Measuring the Amount of Force Used By and Against the Police in Six Jurisdictions by Joel H. Garner and Christopher D. Maxwell This study examines the amount of force used by and against law enforcement officers and more than 50 characteristics of officers, civilians, and arrest situations associated with the use of more or less force. Data were gathered about suspects' and police officers' behaviors from more than 7,500 adult custody arrests[1] in 6 urban law enforcement agencies. The participating agencies were the Charlotte-Mecklenburg (North Carolina) Police Department, Colorado Springs (Colorado) Police Department, Dallas (Texas) Police Department, St. Petersburg (Florida) Police Department, San Diego (California) Police Department, and San Diego County (California) Sheriff's Department. To organize, present, and understand the nature and characteristics of the force used in representative samples of arrests, this study developed four measures of the amount of force used by police officers and four parallel measures of the force used by arrested suspects. The participating agencies also played a major role in designing and implementing this research. Senior police managers in each department actively contributed to delineating research goals, designing data collection forms, and constructing the measures of force reported here. The departments encouraged officers to complete police survey forms and assisted the research team in gaining access to suspects in local detention centers. Throughout the project, these law enforcement managers worked together as a group to promote the use of consistent measures across the six jurisdictions and to enhance the use of the findings by the participating departments. This collaborative design was intended; in fact, it is not clear whether there was any other way to produce the information included in this chapter. Emerging from this research is a more complete understanding of the frequency with which certain types of tactics are used and what types of weapons are displayed, threatened, or actually used. The consistent findings across all six jurisdictions are that most arrests do not involve any force by the police or by suspects. In those situations where some type of force is used, typically no weapon is used, threatened, or even displayed. When police use some form of weaponless tactic (hitting, kicking, wrestling, etc.), the most frequent tactic involves grabbing only. The Importance of Systematic Samples Prior research on the use of force has, in many instances, been limited to instances where some type of force, usually deadly force, was used.[2] This use of samples that do not represent all police behavior limits our ability to describe when force is used and when it is not used. One senior researcher, Albert J. Reiss, Jr., has argued that the absence of systematic samples of police behavior means that the entire approach to the study of deadly force was flawed "because analysts of police use of deadly force focus on situations in which the decision was made to use it, such as firing a weapon . . . they ignore all decisions where force gave way to alternative ways of coping with situations."[3] The flaws that Dr. Reiss identified can have real-world, and sometimes disastrous, consequences. For instance, the congressional investigation into the Challenger shuttle disaster identified the failure of the engineers at NASA to examine evidence from a systematic sample of all of the previous 21 shuttle flights.[4] The engineers examined only the eight Shuttle flights with identified 0-ring problems, and among those flights, there was no correlation between 0-ring failures and temperature at the launch pad. However, in the complete set of all 21 flights, the ultimately deadly correlation was clear. This research employed systematic samples of adult custody arrests in order to provide a comparison of circumstances when force is used with the circumstances when force is not used. This is an important characteristic. By using a complete set of police behavior, even when no force is involved, we have sought to avoid types of errors that occurred in the Challenger study as well as most of the other police studies. Our approach to data collection compromised ideal research procedures with the real-life necessities of working within an operational agency. We wanted to obtain a sample of adult custody arrests representative of each department's annual arrests. We estimated that the number of arrests needed to obtain reliable estimates of the amount of force varied between 900 and 1,200 in the six jurisdictions. However, we did not draw a random sample of arrests throughout the year because that would have entailed complicated procedures for starting and stopping data collection by police officers. We chose to sample arrests continuously over a 2- to 7-week period, depending on the size of the department and the rate at which officers made arrests. Data collection began at different times in different departments, so the total sample included arrests during the summer, fall, and the winter of 1996-97. We began data collection in the Colorado Springs Police Department in mid-August 1996 and completed data collection in Charlotte-Mecklenburg in the second week of February 1997. It took 50 days to obtain data on 1,249 arrests in Colorado Springs but only 2 weeks to obtain data on 1,192 arrests in Dallas. In all 6 jurisdictions, we obtained 7,512 usable surveys. The proportion of adult custody arrests for which we obtained completed, usable surveys varied from 85 percent to 93 percent, and we determined that there were no systematic differences between arrests with or without a completed survey. The large size and representative nature of our sample provide a solid basis for describing the nature of use of force in the six participating departments. This research also sought to avoid other methodological flaws. We collected information both from police officers and from arrested suspects and did so in ways that protected the confidentiality of officers and suspects (see sidebar "Measuring force"). In assessing the factors associated with the use of force, this study used multivariate statistical tests appropriate for the measurements available and the hypotheses being tested. Simple descriptive information alone can be misleading, especially when research seeks to associate the use of force with characteristics of officers, suspects, and arrest situations. Our design has some limitations. There are some circumstances in which the police use force but do not make an arrest, and our approach will miss those incidents. In addition, the primary source of our data is self-reported marks by police officers on a two-page form. These types of measures provide some uniformity but often miss important distinctions that can be obtained by more indepth interviews or firsthand observations by independent observers. For instance, we asked officers to record whether they used a weapon, displayed a weapon, or threatened to use a weapon. However, it is not always clear when a weapon is actually used. Does a firearm have to be discharged to be "used"? In addition, is a firearm or a baton displayed just by being carried by an officer, or does display mean only removed from a holster? If a weapon is displayed, is an officer or a suspect threatening to use it? We think the distinctions between use, threatened use, and display are important but acknowledge that the survey form might be an imprecise method to measure these differences. The design of this research--systematic samples, multiple sources of information, and multivariate analysis--was guided by an assessment that much of the prior research had confounded the measurement of force with definitions of what is and is not excessive force. In this project, we deferred the difficult task of defining and measuring excessive force. We focused on the measurement of the amount of force, with the expectation that this information would inform issues surrounding the use of excessive force. For instance, excessive force is typically but not necessarily associated with more severe forms of force that could or do result in injury or death. Our findings are that most arrests involve no force, excessive or otherwise. When force is used, it typically involves less severe forms of tactics and weapon use. These findings provide a context for understanding excessive force, which we know can involve low-level acts of force (such as verbal threats or cursing against compliant suspects) as well as the acts of force that result in physical injury or death of civilians. Arrests that involve no force, however, cannot involve excessive force and arrests that involve low levels of force are less likely to involve excessive force. Although the exact relationship between the amount of force and excessive force remains to be clarified, this research seeks to inform future law enforcement policies, practices, and training by identifying what kinds of force are and are not currently being used by and against law enforcement officers. The Elements of Force This chapter emphasizes measuring the amount of force used by law enforcement officers and by suspects. The task of measuring the amount of force required establishing an understanding of the specific behavioral acts that constitute "force" and how much force is involved in each of those behaviors. Prior research had traditionally employed simple dichotomies between the presence or absence of physical force or abuse of force or excessive force without much attention to what elements of the arrest or the police-public encounter constituted force, abuse of force, or excessive force. Our approach has been to explicitly define and measure force and, building on prior research in Phoenix, Arizona,[5] we identified five elements of force: weapons, weaponless tactics, restraints, motion, and voice. Weapons There is a strong consensus that the use of a weapon constitutes force and that the use of certain types of weapons--e.g., handguns and rifles--involves more force than the use of such other weapons as batons and oleoresin capsicum (pepper spray). What is less clear is the meaning of "use." For instance, does a firearm have to be discharged to be "used"? Also unclear is whether the possession, threatened use, or display of a weapon constitutes force by law enforcement officers or by suspects. Our approach to this uncertainty was to have officers mark on the data collection form whether they or the suspects possessed, displayed, threatened to use, or used any of seven different types of weapons. Combining data from all sites, exhibits 4-1 and 4-2 show the frequency with which officers report that they or arrested suspects used, threatened to use, or displayed certain weapons. Use of weapons is infrequent; in 97.9 percent of all adult custody arrests, police did not use a weapon (exhibit 4-1). In 99.3 percent of all such arrests, suspects did not use a weapon (exhibit 4-2). As noted in exhibit 4-1, the most frequent weapon used by the police was some form of a chemical agent, mostly oleoresin capsicum;[6] it was used in 88 or 1.2 percent of the arrests in this study. The second most frequent weapon was the flashlight, used in 41 (0.5 percent) arrests. Handguns were used by the police in 11 (0.1 percent) arrests; rifles or shotguns were used by the police in 7 (0.1 percent) of the arrests (no officer reported that such use involved discharge of a firearm). The most frequent weapon used by suspects was a knife; it was used in 18 (0.2 percent) of all arrests (exhibit 4-2). Suspects used handguns in 12 (0.2 percent) arrests and rifles or shotguns in 5 (0.1 percent) arrests.[7] A somewhat different pattern emerges in exhibit 4-1 where we examine instances in which weapons were either displayed or used. Handguns were displayed or used by police in 202 (2.7 percent) arrests and rifles or shotguns were used or displayed 31 (0.4 percent) times. Thus, firearms are infrequently used but are the most frequent weapon displayed. On the other hand, chemical agents were the most frequently used weapon but ranked second (118 or 1.6 percent) when use and display are counted. Exhibit 4-1 indicates that in 215 (2.8 percent) of 7,512 arrests, officers went so far as to display a firearm but did not, ultimately, use a firearm. These findings suggest important differences in incidents in which weapons are used and those in which they are displayed but not used. Exhibit 4-1 also reveals a finding similar to one reported in the Phoenix study: police officers report that they use, display, and threaten to use a flashlight more often than they use, display, or threaten to use a baton. Batons were used in 15 (0.2 percent) arrests; flashlights in 41 (0.5 percent). In addition, officers report that they used motor vehicles as weapons in 15 arrests and either used or threatened to use them in 21 arrests. These findings do not conform to conventional thinking about the relative frequency of weapon use or even on the types of equipment used as weapons. Officers report that suspects use, display, and threaten to use weapons less frequently than officers do. Exhibit 4-2 notes that knives were the weapons most frequently used (18 or 0.2 percent) by suspects, followed closely by sticks (17 or 0.2 percent) and motor vehicles (14 or 0.2 percent). When handguns, rifles, and shotguns are combined, the weapon most frequently displayed or used (43 or 0.6 percent) by suspects is a firearm. Suspects threatened, displayed, or used a firearm in 63 arrests. These findings about suspects confirm our earlier findings about the police: understanding the use of force is advanced by considering not only the use but also the display and the threatened use of a weapon. Weaponless tactics Police officers use and are trained to use a variety of weaponless tactics, from carotid control holds to simply grabbing a suspect by the arm. Each of these tactics involves direct physical contact between the officer and the suspect and does not involve use of specific objects in applying force. The police survey form listed 12 tactics and exhibit 4-3 displays the frequency with which officers reported the use of these tactics. In 6,328 or 84.2 percent of the arrests in this study, the police reported that they used no weaponless tactics. Among arrests involving a weaponless tactic, the most frequent "tactic" was grabbing, used 954 times (12.7 percent of all arrests). Other common tactics involved using a control hold (164 times or 2.2 percent of arrests), arm twisting (281 or 3.7 percent), wrestling (233 or 3.1 percent) and pushing or shoving (145 or 1.9 percent). Police reported that suspects used weaponless tactics in 412 or 5.5 percent of all arrests (exhibit 4-4). Wrestling was the tactic most often used by suspects (262 times or 3.5 percent of all arrests), followed closely by pushing or shoving by the suspect (166 times or 2.2 percent). As with weapons, the use of weaponless tactics by officers and by suspects was infrequent. Although weapons and weaponless tactics are typically included in most understanding of what constitutes the use of force, the three other elements discussed next are sometimes, but not always, considered part of the use of force. Restraints One element of force that officers alone employ is use of restraints.[8] The police survey form lists three possible types of restraints--handcuffs, leg cuffs, and more severe restraints, such as hobbles or body cuffs. As with most other items, the survey provided for a specific "yes" or "no" response for each item. In the 7,512 arrests in this study, officers reported that they used handcuffs in 6,182 (82.3 percent) (exhibit 4-5). In 67 (0.9 percent) arrests, the police used leg cuffs and in 29 (0.4 percent) they used more severe restraints.[9] The use of restraints appears to be frequent but not universal; handcuffs predominate, but in a small proportion of arrests (1.3 percent) restraints more severe than handcuffs were used. Handcuffing alone is not typically perceived as involving force, but our understanding of force might include the use of more severe restraints, some of which have been associated with injury to suspects and, in some instances, even death.[10] Motion One aspect of police-public encounters in arrest situations is suspect flight and officer pursuit. Although most research and policy discussions on the use of force do not address issues around either flight or pursuit, we include both as potential elements of force. Our police survey form recorded whether suspects attempted to flee and, if so, whether they fled on foot, in a motor vehicle, or by other means. As displayed in exhibits 4-6 and 4-7, in most arrests, there was no flight (7,027 or 93.5 percent) and no pursuit (7,089 or 94.4 percent). In a small number of arrests, suspects fled on foot (354 or 4.7 percent) or in a motor vehicle (128 or 1.7 percent). When the police did pursue a suspect, it was most often (224 or 3.0 percent) on foot; motorized pursuits, including helicopter pursuits, occurred in 199 (2.7 percent) arrests in our sample. Flight and pursuit do occur, but it is not clear the extent to which these actions involve the application of what we typically mean when we speak of the use of force by police or by suspects. Still, flight and pursuit can result in serious injury to officers, suspects, or bystanders, especially if conducted in a motorized vehicle, and such actions are included in some definitions of what constitutes use of force. Voice We include as a potential element of force what police said to suspects and what suspects said to police. Our police survey form listed four categories of speech--conversational, commands, shouting or cursing, and verbal threats. In more than half of all arrests (4,599 or 61.2 percent), police reported that they used a conversational tone with the suspect; in 2,297 or 30.6 percent of arrests, they reported that they commanded the suspect to do something (exhibit 4-8). Police reported shouting or cursing at suspects in 73 (1.0 percent) of the arrests and threatening them in another 58 (0.8 percent) arrests. Finally, the police reported that they said nothing to the suspects in 485 or 6.5 percent of all arrests in this study. Suspects, according to the police survey, spoke to the police in a conversational tone in 4,970 (66.2 percent) arrests (exhibit 4-9). The police reported that suspects used a command voice in 240 (3.2 percent) arrests, shouted or cursed at officers in 638 (8.5 percent) arrests, and made verbal threats in 473 (6.3 percent) arrests. The survey form could not capture the details of what was said by officers or suspects, but officers reported the use of shouting or profane language as well as the use of threats in a small percentage of arrests; in a larger percentage, but still a distinct minority, of arrests, suspects made threats or conversed using a raised voice or obscene language. Although the core understanding of the use of force typically does not involve what is said but what is done, the nature of verbal communication, especially if it involves threats, shouting, or cursing, can be an element of force and needs to be incorporated into how we think about and measure the use of force. Summary: Elements of force The five elements of force--weapons, weaponless tactics, restraints, motion, and voice--identify different dimensions of the use of force and provide a framework to measure the existence of force and the amount of force in any given situation. We have identified and elaborated on these elements in order to record a broad range of activities that, under different definitions, could be considered use of force. This project purposefully attempted to measure aspects of police-public encounters, such as weapon use, that clearly involved force and other aspects, such as speaking in a conversational tone, that did not involve physical force. The design of this research was to measure many specific and concrete behaviors against which different definitions of force could be applied. To not record all aspects of each element of force would preclude the possibility of using those aspects to determine if force was used and, if it was used, how much force was used. The behaviors recorded capture the details of specific behaviors by officers and suspects, but they do not constitute complete measures of force. Taken singly, none of the five elements of force captures fully everything that is typically meant by the use of force. A fully developed measure of force requires the use of definitions that determine precisely how combinations of these elements constitute the presence of force or increases in the amount of force used by and against the police. The next section provides more detail on how we translated our five abstract sets of behaviors into measurements of force. Four Measures of Force Using the items included on the police survey, the research team constructed four measures of force used by police officers--Physical Force, Physical Force Plus Threats, Continuum of Force, and Maximum Force. We also developed four comparable measures of force by suspects. Each of these measures is a summary of behaviors derived by combining specific actions by law enforcement officers and by suspects in different ways. We recognized that no single measure is likely to capture well all the different understandings of the use of force. Thus, this research used multiple measures of force in order to incorporate more precisely the various ways in which force is conceptualized by the police, the public, and researchers. Physical force The first measure is a traditional conceptual dichotomy of those arrests where physical force was or was not used. We defined the use of physical force for officers and for suspects in a parallel but slightly different manner (see first definition in sidebar "Definitions of measures of force"). For both the police and for suspects, the definition of physical force includes any arrest in which any weapon or weaponless tactic was used. In addition, we include as examples of physical force by the police arrests in which officers used a more severe restraint--prone cuffing, hobble, body cuff, or leg cuff. Physical force plus threats Our second measure, Physical Force Plus Threats, includes all the elements of Physical Force but adds use of threats and displays of weapons. This measure combines actual physical force with threatened force. Although this combination may be inappropriate for some purposes, threats of violence are typically reported as violence in other measures, such as in the FBI's Uniform Crime Reports. Our second measure incorporates the threat component of the use of force. The use of dichotomies between physical force and no physical force is a traditional approach to understanding and measuring the use of force. The strengths of these dichotomous measures include their ability to be applied consistently across all jurisdictions and types of law enforcement agencies and to capture those elements of force that are frequently salient to the police and to the public. Their weakness is that they group together all uses of force, from a push or a shove to the discharge of a firearm, and make no distinctions among activities, such as the use of handcuffs or pursuits, that are not typically included in definitions of physical force. To address these potential limitations of these two dichotomous measures, we developed two other measures with other strengths and other weaknesses. Continuum of force The third measure of force, Continuum of Force, developed in this project captures the rankings of force commonly used by law enforcement agencies to indicate distinct levels of suspect resistance and levels of police response (see third definition in sidebar "Definitions of measures of force.").[11] The gradients of force used by the participating departments are similar to those used by many police departments in their arrest tactics training and in their policies on the appropriate use of force. Our measurement of this "continuum of force" is intended not only to reflect the official policies of the participating law enforcement agencies but also to incorporate into our research the widely held notion that the force/no force dichotomy is sometimes inadequate to capture all the important variations in the ways police handle encounters with the public and the nature of suspect resistance to the police. Unlike the Physical Force dichotomies, the Continuum of Force measures are purposefully responsive to the specific use of force policy and training in each department. Because these measures are not consistent, it is not possible to combine cases from the six jurisdictions into one measure of police use of force and one measure of suspect use of force. The Continuum of Force measures capture distinctions among types of force (like various weaponless tactics and the use of weapons) that are not possible in the two Physical Force dichotomous measures. The Continuum of Force measures have a natural ranking of categories from less forceful to more forceful. This research quantifies that natural ranking and creates a scale in which each category is considered more forceful than the previous category. One weakness of using these policy categories as a numerical scale is that this formulation, in Charlotte-Mecklenburg for instance, considers the difference between officer presence and verbal directions to be the same as the difference between intermediate weapons and lethal force. Few observers would accept that these differences are equivalent. Maximum force This multisite research developed a fourth measure of force, which we call Maximum Force. This measure ranges from 1 to 100, with a ranking score of 1 being the least forceful and 100 being the most forceful. We created this measure in a two-step process. In five of the six participating law enforcement agencies, we asked 503 experienced officers to rank more than 60 hypothetical types of force on a scale from 1 to 100. For instance, one item was "An officer uses a baton" and another was "An officer threatens to use a handgun." Officers were asked to rank these items based not on departmental policy but on their personal experience. We asked all officers to indicate, in their opinion, how much force was involved in each type of force. This exercise resulted in a measure that makes reasonable (but not necessarily perfect) distinctions between different types of force. Officer presence, conversation, and commands are ranked near the bottom and the use of weapons, especially firearms, are ranked near the top. These rankings include some elements of force that are included neither in most discussions of force nor in our Physical Force or our Continuum of Force measures. For instance, officers rank the use of handcuffs at 28.2 and chasing a suspect in a car at 41.4. Experienced officers in our survey ranked these behaviors as involving substantial amounts of force, but our measures of Physical Force and Continuum of Force would count arrests that involved just handcuffing or just a pursuit as involving no physical force or as mere officer presence. The second step in developing the Maximum Force measure is to determine if such behaviors occurred in our sample of 7,512 arrests and, if so, to weigh them according to the rankings made by police officers. When police officers reported that they twisted a suspect's arm, the amount of force for that arrest was measured as 35.1; when they used a carotid hold, the amount of force was measured as 56. When officers reported that they engaged in two or more forceful acts, we recorded the one with the highest ranking--hence the name Maximum Force. These rankings, depicted later in exhibit 4-13, are presented not as a perfect or universal scale but as an example of how the amount of force can be quantified in a way that approximates our understanding of variation in the use of force. For purposes of this research, the important issue is that this type of measure captures important aspects of the use of force that would be missed if research were limited to simple dichotomies or the Continuum of Force measures. There are differences between grabbing and kicking and between threatening to shoot someone and actually shooting them; the Maximum Force measure is an attempt to measure those real but imprecisely known differences. The development and use of a variety of detailed measures of force is intended to encourage researchers and policymakers to explicitly include or exclude specific behaviors and to explicitly consider the severity of different types of force. It may be too early to establish uniform measures, but it is not too early to start proposing uniform measures. Summary: Measures of force The use of force is not a simple concept that is easily measured. This research has taken the issue of measurement seriously and developed a variety of measures that, as a group, capture many if not all the crucial distinctions that are commonly made about the amount of force used by and against police officers. Our efforts at measurement are not definitive. The definitions of Physical Force and Physical Force Plus Threats are, we believe, reasonable but not the only reasonable definitions that could be used. The Continuum of Force measures are derived from departmental policies, but these policies vary from department to department and within departments over time. The Maximum Force measure is the most innovative effort and perhaps the least well developed, but unlike the other measures, it reflects the relative ranking of experienced police officers. We are not yet prepared to assert that one form of measurement is to be preferred over other forms. Certainly, improvements can be made in the measures we have developed, but future research needs to be explicit about how force is measured and to justify why the particular measures of force used are appropriate. Until such measures are developed and justifications provided, we recommend the four measures reported here. The Distribution of Force by Police and Against the Police Exhibit 4-10 displays the number and percent of the adult custody arrests in this study that met our definition of Physical Force. Law enforcement officers reported that they used physical force in 1,283 or 17.1 percent of our sample of adult custody arrests; they used physical force or threats of force in 1,418 or 18.9 percent of the 7,512 arrests in this study. Thus, whichever of these two definitions is used, our findings are that law enforcement officers used or threatened to use physical force in fewer than one of every five adult custody arrests. However, as we argued earlier, this simple dichotomous measure may not provide a sufficiently clear understanding on the nature of force used by the police. To help better understand the nature of force, exhibit 4-11 displays the frequency with which different elements of force (threats, restraints, weaponless tactics, or the use of weapons) were the most severe form of force used among all instances in which law enforcement officers used some force. The exhibit shows that the predominant types of force used by police officers do not involve firearms or other weapons but some form of direct physical contact, which we categorized as "weaponless tactics." In almost 80 percent of all incidents involving physical force or threats of force, the most severe form of force used by law enforcement was a weaponless tactic; less than 12 percent of the arrests which met our definition of Physical Force Plus Threats involved the use of a weapon. Focusing more closely on weaponless tactics, exhibit 4-12 indicates those instances in which officers use some form of weaponless tactics. This exhibit reveals that the most frequent type of weaponless tactic was grabbing the suspect. In 7.8 percent of all 7,512 arrests and 49.7 percent of all 1,184 arrests in which the police used at least one tactic, the most severe tactic used was categorized as a "grab." Other less frequent types of tactics involved the use of control holds, arm twisting, pushing or shoving, wrestling, or hitting or kicking. These findings confirm the results of prior research that established that most adult custody arrests do not involve force or threats of force and those arrests that do involve force are typically at the low end of severity. When the Maximum Force measure is applied, with its 1-to-100 ranking scores, most arrests still involve little or no force but a proportion of arrests involve substantial amounts of force. Exhibit 4-13 notes the large number of arrests (4,305 or 57.3 percent of all 7,512 arrests) at a ranking of 28.2. In these arrests, the most forceful behavior by the police was handcuffing. The exhibit also reveals that there was variation among arrests in which some form of forceful action was taken. Some actions, such as the display or the use of a handgun, were very forceful and generate a Maximum Force ranking of 55.4 and 81.7. The Maximum Force measure captures a number of activities, such as police chasing suspects in a car, that are not counted as force in our definitions of Physical Force, Physical Force Plus Threats, and in departmental policies on the Continuum of Force. In addition, this measure takes into account differences, sometimes severe, between types of force. Grabbing a suspect and using a firearm are both examples of Physical Force and are counted equally in that measure; in the Maximum Force measure, grabbing ranks at 33.0 and using a handgun at 81.7. Thus, this measure captures items that officers think involve force and weigh the amount of force based on a scale that can range from 1 to 100. Summary and Discussion This research collected information from a systematic sample of adult custody arrests and used that information to construct a variety of measures of force. We have used these data to describe the amount of force used by the police in six urban jurisdictions. We have emphasized various definitions of force and demonstrated a variety of methods that explicitly and quantitatively describe force. Our research suggests that no one measure captures well all the elements that go into our understanding of what comprises force by police officers against civilians during an arrest. At the present time, there is no single conception of what constitutes the use of force, and this constrains our ability to implement precise measures of the presence of force or the amount of force. The multiple measures we have developed and implemented here illustrate how different elements of force can be combined into meaningfully distinct measures. In this research we found that the use of force is relatively infrequent, regardless of the measure used. When the use of force does occur, the amount of force is usually at the low end of our measures of force. Combined with the similar findings from our initial study in Phoenix, these substantive findings are beginning to provide a stable picture of police behavior and the amount of force the police use in arrest situations. These findings, however, remain tentative given the small number of jurisdictions involved in this research and the room for improvements in methods of data collection and in precision in measuring the amount of force. Although this research demonstrates that police agencies and researchers can work collaboratively to describe the amount of force used by police officers, much remains to be done to improve our measurements and to use those measures to determine the types of circumstances in which more force is used. This multisite research project collected information about more than 50 potential predictors of force, and additional analysis of those data will be forthcoming in the near future. This line of research holds great promise for identifying the actual nature of police use of force as well as identifying those characteristics of police recruitment, training, tactics, and philosophy that can assist police departments in moving away from a reliance on the use of force and toward an increased reliance on the use of information and cooperation with the communities they serve. Future research also needs to focus on the relationship between the overall amount of force used in a law enforcement agency and the nature, scope, and extent to which the force that is used meets various social understandings and legal definitions of what constitutes excessive force. Our understanding of these relationships might benefit from more precise understandings of how force is measured and a comparison of the relationship between the amount of force used by the police and the amount of force used against the police. Notes 1. Custody arrests involve transporting suspects to a detention facility; instances where suspects are arrested and issued a summons to appear before a judicial officer are not included in this research. 2. Geller, William A., and Michael Scott, Deadly Force: What We Know, Washington, DC: Police Executive Research Forum, 1992. 3. Reiss, Albert J., Jr., "Police Brutality--Answers to Key Questions," in Law and Order Police Encounters, ed. Michael Lipsky, New Brunswick, NJ: Aldine, 1980: 127. 4. Vaughan, Diane, The Challenger Launch Decision: Risky Technology, Culture, and Deviance at NASA, Chicago: University of Chicago Press, 1996. 5. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding the Use of Force By and Against the Police, Research in Brief, Washington, DC: U.S. Department of Justice, National Institute of Justice, November 1996, NCJ 158614. 6. Some arrests involved the use of more than one weapon. 7. After some discussion, the research team (researchers and agency personnel) decided to include on our officer survey form the possibility that a motor vehicle might be used as a weapon. 8. This research assumed that in successfully completed arrests, suspects would not be using restraints on officers. 9. In 666 (8.8 percent) arrests, the officers explicitly marked on our forms that they did not put handcuffs (or other restraints) on the adults that they arrested and took into custody. In another 568 (7.5 percent) of the arrests, the officers did not indicate whether they did or did not use any type of restraint. 10. Granfield, John, and Jami Onnen, Executive Brief: Pepper Spray and In-Custody Deaths, Alexandria, VA: International Association of Chiefs of Police, 1994; and National Institute of Justice, Positional Asphyxia--Sudden Death, National Law Enforcement Technology Center, Washington, DC: U.S. Department of Justice, National Institute of Justice, June 1995. 11. For purposes of this research, some of the participating law enforcement agencies collapsed several formal departmental rankings into a smaller number of measurable categories that reflected distinct and measurable differences in officer behavior. ------------------------------- Sidenotes Joel H. Garner, Ph.D., is Research Director, Joint Centers for Justice Studies, Inc. Christopher D. Maxwell, Ph.D., is Assistant Professor of Criminology, Michigan State University. The authors' research was supported under grant number 95-IJ-CX-0066 by the National Institute of Justice. ------------------------------- Measuring force The primary means by which the research project collected information on the use of force was a one-page, front and back, form completed by law enforcement officers on a systematic sample of adult custody arrests. This form, illustrated on the two following pages, was used to record characteristics of the arrest situation, the suspect, the officer, and the specific behavioral acts of officers, suspects, and bystanders in a particular arrest. The form was derived from a similar study* conducted in Phoenix, Arizona, during 1994 but was modified to conform to the local characteristics, police terminology, and departmental policies of the participating agencies. Similar items were asked of a smaller sample of suspects interviewed in local jails shortly after arrest. The forms were completed by arresting officers but were not reviewed or controlled by police managers; thus, the forms were not departmental records but research data. Both the police officers completing the forms and the suspects who were interviewed were research subjects, and the confidentiality of their responses was protected by the data management and storage procedures of the Joint Centers for Justice Studies. Under the legislation authorizing the research program at the National Institute of Justice, Congress made confidential research data "immune from legal process" and specified that data identifiable to an individual shall not be "used for any purpose in any judicial, legislative, or administrative proceeding."** This confidentiality protection was communicated to officers by their departments and included on the survey form; interviewed suspects were told of this protection prior to their agreeing to participate in the research. These procedures increased the likelihood that officers and suspects would provide more accurate information because officers would be less constrained by fears that their individual answers might be communicated to others within the department and might even possibly be used against them. The confidentiality provided to research subjects by the Congress makes the findings of this research more reliable and, therefore, more useful to congressional and other policymakers. * Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding the Use of Force By and ----Against the Police, Research in Brief, Washington DC: U.S. Department of Justice, National Institute of Justice, November 1996, NCJ 158614. ** 42 U.S.C. 3789(g). ------------------------------- Definitions of measures of force Definition 1: Measure of Physical Force Police Use of Severe Restraints Use of Any Weaponless Tactic Use of Any Weapon Suspect Use of Any Weaponless Tactic Use of Any Weapon Definition 2: Measure of Physical Force Plus Threats Police Use of Severe Restraints Use of Any Weaponless Tactic Use, Display, or Threatened Use of Any Weapon Suspect Use of Any Weaponless Tactic Use, Display, or Threatened Use of Any Weapon Definition 3: Categories of Police Continuum of Force Charlotte-Mecklenburg Officer Presence Verbal Direction Soft Empty Hand Oleoresin Capsicum Hard Empty Hand Intermediate Weapons Lethal Force Colorado Springs Officer Presence Verbal Control Soft Control Techniques Control and Compliance Hard Control Techniques Impact Weapons Lethal Force Dallas Officer Presence Verbal Control Empty Hand Control Intermediate Weapons Lethal Force St. Petersburg Officer Presence Verbal Direction Restraint Devices Transporter Takedown Pain Compliance Countermoves Intermediate Weapons Lethal Force San Diego Police Officer Presence Verbal Commands Control/Compliance Soft Impact Lethal Force San Diego Sheriff Deputy Presence Verbal Direction Soft Hand Control Chemical Agents Hard Hand Control Intermediate Weapons Lethal Force ------------------------------- Chapter 5 The Force Factor: Measuring and Assessing Police Use of Force and Suspect Resistance by Geoffrey P. Alpert and Roger G. Dunham Collecting and interpreting information on police use of force is a persistent problem for police managers and researchers. Although such data are critical to both the police and the public, they remain difficult to collect, measure, and interpret objectively. There has been an energetic effort to collect data on all police use-of-force incidents, including excessive force, by various groups and by assorted methods.[1] The problems with data collection on such organizationally sensitive and controversial acts suggest the need for standardizing measurement and providing reliable and valid measures.[2] This chapter presents information collected from the police departments of the sister cities of Eugene and Springfield, Oregon, and the Miami-Dade Police Department. After a brief description of the sites and a presentation of their use-of-force information, the concept of the "force factor," which is a measure of police force incorporating the officers' actions relative to the suspects' physical resistance, is developed. Description of the Sites Eugene and Springfield, Oregon, are located in the Willamette valley, cover 52 square miles, and had a combined population of 178,000 in April 1995, when the data were collected. The two police departments had 204 officers, 110 of whom were assigned to patrol duties. The departments had 150,841 contacts with the public, 7 complaints of excessive force, 31 complaints of discourtesy, and 2 lawsuits during 1995. The Miami-Dade Police Department (MDPD), located in Dade County, Florida, is responsible for all law enforcement activities in the unincorporated areas of the county. (It was formerly known as the Dade County Sheriff's Department and Metro-Dade Police Department.) In addition, MDPD contracts with many municipal agencies in Dade County to perform specialized services within those agencies' jurisdictions. In 1995, the unincorporated areas of Dade County had a population of approximately 2 million, and the county included 1,840 square miles. The Department had 2,725 sworn officers, 845 of whom were assigned to patrol. There were more than 1 million reported contacts (arrests and nonarrests) during the 3-year study period (fourth quarter 1993 through 1995), of which 133 resulted in complaints of excessive force, 243 in complaints of discourtesy, and 18 in lawsuits. These police departments differ on important characteristics. There are obvious geographic and size differences that make the findings relevant to a wide audience. Further, differences in the social and ethnic environments of the cities studied provide diverse contexts in which to analyze police use of force. Unfortunately, different data elements were collected from the sites, which make some comparisons problematic. Eugene and Springfield Police Departments The Eugene and Springfield dataset was created from items in the Police Officers' Essential Physical Work Report Form, which was completed by department members during April 1995. These data are unique because they include a broad range of police work but were not collected to evaluate force used by the police. This dataset was part of a larger effort to identify physical abilities necessary for police work. Data included all police-public contacts, including those related to forceful encounters. As a result, an unobtrusive measure of police use of force was available. The findings from the Eugene and Springfield Police Departments are reported in the following order: --Circumstances surrounding the incidents. --Suspects' characteristics and actions. --Officers' characteristics and actions. Most of the 562 police actions analyzed were initiated by dispatched calls (57 percent), although 33 percent of the incidents were initiated by the officer who observed a situation and reacted to it. Circumstances surrounding the contacts The most common type of incident was street violence (25 percent). However, 14 percent of all incidents involved domestic violence, and another 14 percent pertained to resisting an investigation. Thirty-two percent of incidents did not fall into one of the predefined categories. Most of the police action was taken to apprehend or control a person (76 percent). Suspect characteristics and behavior Suspects ranged in age from 12 years to 86 years. The average age of suspects was between 28 and 29. Eighty-four percent of the suspects were males. Although no ethnic information about suspects was collected on the agency form, there were very few minorities. Fifty-two percent of suspects were calm, reasonable, and cooperative. However, 19 percent were reported as under the influence of drugs or alcohol and 17 percent as emotionally upset or abusive. Eleven percent of suspects appeared mentally unstable and unpredictable, and 2 percent were violent. Most suspects were perceived by officers to be average (55 percent) or below average (20 percent) in physical abilities. The majority of suspects did not resist the officer (61 percent); 18 percent put up only slight resistance. Four percent were characterized as having a high level of resistance, 2 percent as violent, and 1 percent as explosive. The most common type of resistance was to push or pull the officer to resist an arrest or to escape. Officer characteristics and behavior The ages of the officers ranged from 25 to 60, with a mean of 37 years. Length of service as an officer ranged from 9 months to nearly 34 years. Average length of service was 12 years. Eighty-six percent of the officers were males. Most officers were assigned to patrol (91 percent). Control tactics ranked by severity. Data presented in exhibit 5-1 show that at least two verbal or physical control tactics were used per attempt to apprehend a suspect. Of the 546 incidents covered by the exhibit, all involved at least one tactic (96 percent being verbal); 93 percent, two tactics; and 87 percent, three. There is a fairly large percentage drop for incidents involving use of a fourth tactic (41 percent). Use of a fifth (8 percent of incidents) or sixth (5 percent) tactic was relatively uncommon. Exhibit 5-1 shows use of multiple tactics and how they fall on a use-of-force continuum. The types of force listed down the lefthand side of exhibit 5-1 are in a sequence that reflects a typical use-of-force continuum, from the lowest level of force (verbal commands) to the highest (use of a firearm). Across the top of the table is the number of tactic types the officer used in a given incident. Although tactics are not necessarily listed in the temporal sequence that they occurred, it can be argued that officers do follow a use-of-force continuum. For example, when an incident involves at least one tactic type, the tactic listed as the first used is almost always the lowest level of force: a verbal command (used in 96 percent of incidents). Three percent of such incidents involved handcuffing as the first tactic used, also at the low end of the continuum of force. Very few of such incidents involved higher levels of force as the first tactic used. When incidents involved at least two tactic types, the second did not include verbal commands, in contrast to 94 percent of the incidents involving handcuffing the suspect. Searching the suspect occurred in 4 percent of the incidents; few involved force more severe than a search as a second tactic. For incidents entailing three tactic types, no officers listed verbal commands or handcuffing as the third used. Most listed searching the suspect (95 percent of the incidents); about 4 percent of the incidents involved a wrist or arm lock as the third tactic, which is the next level of force on the continuum; and very few incidents were recorded as involving force levels further down the continuum. The pattern that emerges follows the traditional use-of-force continuum. The first tactic used in an incident is nearly always the least severe use of force on the continuum, and the second tactic used in an incident is nearly always the second-most lenient. Officers apparently follow the continuum with very few exceptions, and those exceptions seem to deviate by only a small degree in relation to the whole range. This same pattern seems to pertain to those incidents involving use of a fourth tactic by officers. None listed the first three levels of force. In 82 percent of four-tactic incidents, officers listed use of wrist or arm lock techniques (fourth level on the continuum) as the fourth tactic used. In 11 percent of incidents with four tactics, officers listed takedowns, which constitute the next degree of force on the continuum. The exception at the fourth-tactic stage is firearm use (4 percent of such incidents). Apparently, continuum adherence breaks down more often in incidents when four types of force are necessary. This becomes even more apparent in situations when five or six types of force are used by the officer. In these situations, the use of force is more scattered and distributed along the more severe end of the continuum. Findings indicate that officers used multiple force tactics most of the time, usually two tactics, three tactics in 87 percent of the incidents, and four in about 40 percent. In these instances, officers also seem to follow a typical continuum of force, with minor deviations that do not vary by more than one or two levels on the continuum. However, in the few instances when more than four force tactics per incident are used, they are scattered over the more extreme end of the force continuum. Although we can infer a probable order of tactics from the data, a temporal sequence is not clear in all cases. Future research should focus on this sequential ordering of officers' use of multiple force tactics and the concurrent level of suspect resistance. A well-conducted interaction model could contribute to understanding police/suspect interactions in these dangerous situations. Control tactics used with varying amounts of suspect resistance. The purpose of this analysis was to determine how many officers followed the typical continuum of force for a given level of suspect resistance. The continuum of force reflects an escalation from verbal commands to deadly force. Nearly 97 percent of the incidents involved initial use of a verbal command when the suspect offered no resistance. Deviating from this typical process of verbally directing the suspect increased when suspect resistance was moderate or high (13 percent of all suspects), and violent or explosive (12 percent). Three percent of officers deviated from first handcuffing the suspect during the encounter when there was no or slight resistance by the suspect. Deviating from the typical control process increased to almost 16 percent of officers when suspect resistance was moderate to high. Less than 1 percent (0.4 percent) deviated from searching the suspect when there was no resistance. Slightly more than 4 percent of the officers deviated from this typical process when resistance was slight, 19 percent when resistance was moderate or high, and 13 percent when resistance was violent or explosive. Highest level of force used. The next analysis involved determining the highest level of officer force used in each incident and comparing that with the level of resistance by the suspect. This analysis provides another way to determine if the level of the force used by an officer was consistent with the level of the suspect's resistance. According to the data presented in exhibit 5-2, when there was no resistance by the suspect (327 incidents), most incidents involved officer verbal commands (8 percent), handcuffing (65 percent), or wrist/arm locks (27 percent). However, three incidents (1 percent) did involve use of a firearm. The force used by some officers when facing slight resistance was more than the force used by officers who faced no resistance. When the suspects used slight resistance, most incidents involved officer use of verbal commands, handcuffing, or wrist/arm locks (altogether 90 percent). There were a few cases of takedowns (3 percent), one incident in which an officer struck a suspect, and six situations in which an officer used a firearm (6 percent) as the most severe tactic used. When suspects resisted at a moderate or high level, officers used verbal commands, handcuffing, or wrist/arm locks as their highest level of force in 48 percent of incidents. Finally, when suspect resistance was violent or explosive, all incidents involved force beyond verbal commands and handcuffing. When suspects acted violently, officers reported four incidents in which a wrist or arm lock was the highest level of force used against the suspect (24 percent). The most frequently used type of force was a takedown (29 percent), followed by striking the suspect (18 percent), wrestling the suspect (12 percent), and using a baton (12 percent). One officer reported an incident involving use of a firearm (6 percent). The data seem to suggest that, in terms of the incidents as a whole, officers' use of force reflected a continuum ranging from lower to higher levels. Miami-Dade Police Department The Miami-Dade dataset included 882 official Control-of-Persons Reports from the last quarter of 1993 and all of 1994 and 1995. These data were reported by the officer's supervisor after interviewing the officer, suspect, and available witnesses. The department's computerized information was used to create the dataset. Findings from the Miami-Dade Police Department are reported in the following order: --Suspects' characteristics and actions. --Arresting officers' characteristics and actions. --Analyses of interaction patterns between officers and suspects. --Analysis of officer and suspect ethnicity. Suspect characteristics and behavior Suspects ranged in age from 12 to 90. The mean age was 28.6 years. Of the 882 suspects, 46 percent were black and 54 percent white. Thirty-five percent of the total number of suspects were Hispanic, most of whom were white. Eighty-nine percent of the suspects were male, and 11 percent were female. Of the 42 percent of suspects who appeared impaired by alcohol or drugs at the time of the incident, 24 percent were reported affected by alcohol and 18 percent were affected by a variety of illegal drugs. A number of suspects were highly agitated or erratic in their behavior during the encounter with the police officer. Officers reported erratic behavior 24 percent of the time. However, 23 percent of the time, suspects were calm when interacting with the officer. Suspect resistance and injury. Because these cases involved some degree of use of force by the officer, it is not surprising that almost all cases involved suspects who showed some degree of resistance (97 percent). The category of resistance most often reported was actively resisting arrest (36 percent), followed by assaulting the officer (25 percent). Twenty-one percent of suspects attempted to escape or flee the scene. The most common type of suspect injury was a bruise or abrasion (48 percent of those injured). The next most common injuries were lacerations (24 percent) and injuries from gunshots (4 percent). Most suspects resisted by using their hands and arms only (65 percent). An additional 14 percent used their fists against the officer, and 12 percent used their feet or legs. Less than 5 percent used a gun (handgun, rifle, or shotgun). One percent used a vehicle to assault the officer, and another 1 percent used a cutting instrument. The most common type of force used by the suspect was striking or hitting the police officer (44 percent). In 27 percent of incidents, the suspect pushed or pulled the officer, and in another 20 percent, the suspect grabbed or held the officer. Eight percent of incidents involved verbal threats or threatening movements or behaviors by the suspect. Role of alcohol or drug impairment on the suspects' behavior. Another important question addressed was whether the suspect appeared intoxicated by alcohol or impaired by drugs and how that affected the confrontation. The 370 suspects who were reported intoxicated by alcohol or impaired by drugs were less likely to be calm and more likely to appear visibly upset (23 percent and 18 percent, respectively), and more likely to be erratic in their behavior (24 percent) or highly agitated (33 percent). Suspects who were reported impaired were no more or less likely to resist the officer than sober suspects, but when they did resist, they resisted in different ways. Impaired suspects were more likely to resist actively or to directly assault the officer than nonimpaired suspects. In spite of this, suspects who were reported impaired were no more likely to receive force by the officer or to be injured during the arrest than were sober suspects. Similarly, suspect impairment by drugs or alcohol was not related to whether the officer was injured during the incident. Although the overall significance of the relationship between impairment and type of suspect resistance was not statistically significant, there was a fairly large difference in resistance with a gun. Suspects reported as impaired were more than twice as likely than sober suspects to use a gun to resist the police. Officer characteristics and behavior The officers ranged from 21 to 66 years of age, with a mean age of 34. Most officers were Anglos (54 percent), Hispanics (31 percent), and blacks (14 percent). Eighty-nine percent of the officers were male, and 11 percent were female. Most officers were assigned to patrol (92 percent), and 5 percent were sergeants. Officer force and injury. The most common type of force used by officers was use of hands and arms (77 percent of use-of-force incidents). In 8 percent of use-of-force incidents, officers used (discharged) their weapons, and in another 7 percent they used dogs (K-9s). In a majority of the incidents (64 percent), officers grabbed or held the suspects. The next most common use of force was to strike or hit the suspect (10 percent of the incidents). The most common injury to officers was bruises or abrasions (64 percent of those injured), followed by sprains or strains (15 percent), and lacerations (15 percent). Of injured officers, 2 percent were bitten by the suspect, 2 percent suffered broken or fractured bones, and 1 percent were injured by gunshots. The vast majority of injured officers received no treatment (76 percent). However, 12 percent were given first aid, and 6 percent were treated by emergency rescue personnel at the scene. Less than 1 percent of officers were treated at a hospital or by their personal physician. Role of officer characteristics. In no department do all officers respond precisely the same to situations, although rules, regulations, and policies of the department should narrow the range of officers' responses to within acceptable and appropriate limits. In the cases examined here, officer characteristics did not make much of a difference in whether force was used or in the level of force used. There were no statistically significant differences in the level of force used by male and female officers. Further, the ethnicity of the officer did not affect the general level of force used or whether force was used. Officer age differences were statistically significant, but the differences may reflect the differences in assignments of younger versus older officers, which was not studied. As the average age of the officers increased, the level of force they used decreased. Interaction patterns between officer and suspect This section focuses on the interaction patterns between officer and suspect. In other words, is there a relationship between the suspect's initial behavior and the officer's response? Ninety-two percent of suspects offered some resistance. The categories included "attempted to flee" (31 percent), "actively resisted the officer" (23 percent), "passively resisted" (20 percent), "assaulted the officer" (17 percent), and "resisting to incite others" (1 percent). Although calm suspects were the least likely to actively resist or assault the officer, they were the most likely to attempt to flee, even more so than suspects perceived to have mental deficiencies or problems (as defined by Florida's Baker Act). Further, it was the suspects who initially acted in a calm manner who were the most likely to resist an officer with a gun or to assault the officer with a vehicle. The initial behavior of the suspect did not influence whether the suspect was injured during the arrest, but it did influence the level of force used by the officer. Suspects who were initially calm were the least likely to have force used against them. They were no more likely to have slight force used against them than other suspects. However, they were among the top two groups to be forcibly subdued by the officer using some method other than hands. An analysis of suspects' initial behavior and officers' injuries resulted in an interesting finding relating to the dangerousness of Baker Act suspects. Suspects who were described as visibly upset or highly agitated inflicted more officer injuries than other suspects (40 percent and 39 percent respectively), and Baker Act suspects inflicted fewer injuries than other suspects (20 percent). There was a strong relationship between the level of officer force and the chance of officer injury. Increasing levels of officer force, regardless of the level of suspect resistance, corresponded with higher probabilities of officer injury. When no force was used, 2 percent of officers were injured. Minimal force situations resulted in 15 percent of officers being injured, and situations involving officers forcibly subduing suspects with their hands resulted in 69 percent of officers being injured. However, when officers used force other than their hands, injuries were reduced to 15 percent. Clearly, increasing levels of suspect resistance increase the chance of an injury to the attending officer. No resistance or passive resistance seldom resulted in an officer injury. However, when the suspect attempted to flee or actively resisted arrest, the chance of an officer injury is increased dramatically. The chance of an officer injury increased even further when the suspect incited others or directly assaulted the officer. Ethnicity of officers and suspects The relationship between the ethnicity of the officer and that of the suspect in force situations is important. Officers used higher levels of force against suspects of their own ethnic group than against suspects of other ethnic groups. For example, Anglo officers used higher levels of force against Anglo suspects than black or Hispanic officers used against Anglo suspects. Black officers used higher levels of force against black suspects than did Anglo or Hispanic officers, and Hispanic officers used more force against Hispanic suspects than did Anglo or black officers. The differences were the least pronounced for Anglo officers and the most pronounced for black officers. Among other explanations, this could be due to a tendency to deploy officers in areas with a preponderance of citizens of their own ethnicity. However, with the greater diversity of neighborhood ethnicity in recent years, this finding may reflect a proclivity on the part of officers to respond differently to members of various ethnic groups. If this were true, each ethnic group might feel more comfortable using force on suspects from its own group. Another interpretation is an officer's possible concern that race and politics might be dragged into the situation when an officer uses force against a suspect of another ethnic group. As a result, officers may try to avoid such situations. Data in exhibit 5-3 compare officer/offender ethnic matches with the degree of resistance by suspects. Although there does not seem to be a relationship between ethnic matches and whether a suspect offers resistance, there are differences in the levels of resistance. Although based on a small number of cases, the ethnic match resulting in the greatest likelihood of a suspect assaulting the officer occurs when a black officer is arresting an Anglo suspect (46 percent). Contrast this to the likelihood of assault when an Anglo officer is arresting an Anglo suspect (14 percent), or when a black officer is arresting a black suspect (17 percent). In exhibit 5-4, officer ethnic matches are compared with the level of force used by the officer. Force was used most often when the officer was black and the suspect was Anglo (100 percent) or Hispanic (100 percent). Force was used least often when the officer was Hispanic and the suspect black (93 percent). Force with hands was used most often when the officer was black and the suspect was either Anglo (73 percent) or Hispanic (77 percent). Force, other than hands, was used most often when the officer was Anglo and the suspect black (32 percent). The Force Factor Prior research on use of force by police has focused on the highest level of force used or the highest level reached in an encounter. This analysis differs from previous ones because its focus is on the level of force used by the police relative to the suspect's amount of resistance, which we call the force factor.[3] This section describes the force factor and concludes with comments on its implications as it applies to policy and training. To calculate the force factor, both the suspects' level of resistance and the officers' level of force must be measured and scaled in the same manner. Even though the force factor is a relative measure, in situations where the level of police force is greater than the level of resistance, there is no necessary implication that the level of police force was excessive or improper. For example, an officer may justifiably use more force than does a suspect to gain control of a situation. Similarly, it is possible that a suspect's resistance may exceed the level of force used by the officer. A force factor representing such a disparity does not necessarily mean that the officer's level of force was too weak or improper. A weaker police use of force, relative to the suspect's level of force, could represent an incident in which a suspect shoots an officer who was unable to respond. Similarly, it could represent a suspect who attacked an officer but who was controlled with a minimum of police force. In any case, the most interesting cases are those that reflect the greatest differences between force and resistance. An important application of the force factor is the analysis of police use of force within a police department. Comparisons can be made between units to understand the use of force and the reasons for differences. Other comparisons can be made for various officer characteristics, such as tenure with the department, training, and assignment, to gain insight into variations of use of force found within the department. Findings can help guide training and supervision. Oregon use-of-force data Using the Oregon data, we measured the level of suspect resistance in four ordinal categories: (1) no resistance, (2) slight resistance, (3) moderate or high resistance, and (4) violent or explosive resistance. The corresponding categories for officer levels of force are (1) no force, (2) slight force, (3) forcibly subdued suspect with hands, and (4) forcibly subdued suspect using methods other than hands. (See sidebar "Force-related terminology.") To calculate the force factor, we subtracted the level of resistance (1-4) from the level of police force (1-4), Force minus Resistance=Force Factor. The range of the force factor is from -3 to +3. A zero is interpreted as force commensurate with the level of resistance. For example, no resistance and no force would be 1 minus 1=0, or passive resistance and minimal police force would be 2 minus 2=0. If the level of force is higher than the level of resistance, the force factor is positive, with one point for each level of discongruence up to a maximum of +3. If the level of force is lower than the level of resistance, then the force factor is negative, one point for each level of discongruence up to a maximum of -3. Exhibit 5-5 depicts the Oregon police officers' use of force in relation to suspects' resistance. The distribution of scores resembles a normal (bell-shaped) curve. This distribution of cases indicates that most incidents fall in the middle, with fewer cases at the extremes. In the Oregon data, the distribution is slightly skewed to the positive side, meaning that, on average, more force than resistance was used. Miami-Dade use-of-force data In this dataset, the level of civilian resistance from the Control-of-Persons Reports was recoded into four ordinal categories similar to those used to analyze the Oregon data: (1) no resistance, (2) passive resistance, (3) active resistance, and (4) assaulted officer. The corresponding categories for levels of police force are (1) no force, (2) minimal force, (3) forcibly subdued suspect with hands, and (4) forcibly subdued suspect using methods other than hands. The force factor was calculated using the same method explained above. The distribution of scores for the Miami-Dade data is close to a normal curve, but slightly skewed to the negative side, indicating, on average, the use of less force than resistance (exhibit 5-6). Although a comparison between the two sites of Miami-Dade and Eugene/Springfield is compelling, interpreting any differences could be problematic because each dataset represents a different selection of incidents as discussed above. Conclusions and Implications for Policy and Training Police use-of-force policies set the tone for how legitimate force can be used against civilians in a particular jurisdiction. Whether departmental policies have an impact in the area of nonlethal force is an empirical question that has yet to be answered. However, research on policies regarding the discharge of firearms and pursuit driving indicates that policies, training, and accountability systems make a significant difference in the number of firearm discharges and pursuits in which officers and agencies are involved.[4] Assuming that use-of-force incidents follow the same trend, a relationship should exist between the use of force by police and the policies that govern such behavior. Policies that govern use of force should focus on four main objectives: maximizing the safety of officers, minimizing injuries to civilians, protecting the rights of those against whom force is used, and providing officers with the tools needed to make arrests effectively and restore order. The major objective of the policies and training is to reduce or minimize injuries. The Oregon and Miami-Dade data paint somewhat different pictures of the injuries suffered by officers during use-of-force incidents. Of 803 incidents analyzed for injuries from Miami-Dade, 308 (38 percent) resulted in a reported officer injury. The vast majority of reported injuries (79 percent) were minor and consisted of bruises, strains, or soreness. Nevertheless, 45 officers were lacerated, 6 were bitten, 5 suffered a broken bone or fracture, 1 received a puncture wound, 1 received internal injuries, and 3 were shot. The series of blue bars in exhibit 5-7 depict the chances of officer injury (not including simple soreness) according to some of the more common ways in which Miami-Dade officers reported using force. These figures suggest that Miami-Dade officers are significantly at risk for injury any time they use force, particularly when they strike a suspect with their fists or use their hands and arms to control a suspect. Because most use-of-force incidents (80 percent) involve the use of hands, arms, or fists, Miami-Dade officers are most at risk for injury when using precisely the type of force that they report using most frequently. Overall, Oregon data show far fewer injuries to officers during incidents involving the use of force. Of 504 reported incidents where force was used, 9 (1.8 percent) resulted in an injury to an officer. Officers in Springfield and Eugene are most at risk for injury when wrestling (21.1 percent), striking (12.5 percent), or taking a suspect to the ground (3.8 percent). In none of the eight incidents when a police baton or pepper spray was used did an officer suffer an injury. The chances of suspect injury are significant no matter what type of force is used by the police. The series of brown bars in exhibit 5-7 summarizes the chances of suspect injury when various types of force are used by Miami-Dade officers. A suspect is more likely to suffer injury if struck with a fist than with a PR-24 police baton. This may be due to the training that police receive in how to use the baton in a manner that minimizes the risk of injury. In any event, the chances of a suspect being injured are greatest when the officer uses his fists, hands, arms, feet, or legs during the encounter. The force factor analysis of the Miami-Dade data yields two important findings with respect to injuries. First, the data indicate that officers are more likely to be injured when using less force relative to the resistance of the suspect. In other words, if an officer does not escalate the amount of force used in response to an increasingly violent suspect, the officer is more likely to be injured. Second, the data show that injuries to suspects increase only minimally as the amount of force used by the police increases relative to the amount of resistance. Although suspects are more likely to receive injuries when police use more force relative to resistance, this increased likelihood of injury is small. Furthermore, even in cases where a suspect was injured, the force factor mean was still negative (-0.114), indicating that, overall, officers use force that is less than the resistance offered by suspects. These findings can be an important source of information for formulating policies and training that help reduce the possibility of injuries. In addition to the policy implications above, the findings from this research point to several training issues that need to be addressed by police agencies. These issues include: --Better training is needed in the use of weaponless (empty hand) control tactics. Because the vast majority of use-of-force incidents are low level in nature, police officers will continue to rely on their hands, arms, and feet to control most resistive suspects. Currently, these common types of encounters result in a disproportionate number of injuries to officers and suspects. If officers were better trained and prepared to deal with these types of encounters, it seems likely that the number and severity of injuries arising from them would decrease. --A use-of-force continuum that matches suspect resistance with officer response levels, combined with a robust training program that reinforces what level of force is appropriate in a given encounter, should help reduce officer and suspect injuries. --Significantly more training is needed in the proper use of chemical agents. The Oregon data indicate that pepper spray was used in 2 of 547 use-of-force encounters. Similarly, pepper spray was used 4 times in 803 encounters by the Miami-Dade police. The Miami-Dade figures are undoubtedly low because the police department does not issue chemical agents to patrol officers but does permit use of chemical agents in specialized tactical operations. The Oregon officers appear to be using pepper spray infrequently. More training on the use and potential abuse of chemical agents may help reduce the number of officer and suspect injuries. --If the PR-24 baton is to be retained, officers need regular retraining and practice in how to use it effectively. In Dade County, every reported instance when the PR-24 side-handled baton was used involved a strike. To those who advocate its use, the advantage of the PR-24 is its ability to be used as a defensive and control-type weapon. When employed properly, the PR-24 can be used to trap and hold the hands and arms of suspects to bring them under control. Apparently, the PR-24 is not being used to its full capacity. This is not surprising because the use of a PR-24 is a diminishing skill that takes a great deal of practice to retain one's ability to use it to full advantage. If officers cannot remain proficient in its proper use, then police agencies should reevaluate whether to continue to issue the PR-24 or whether another impact weapon may be more appropriate. --Officers need more and better training in how to avoid or defuse violent encounters before they arise. If future policies require officers to take reasonable measures to avoid the use of force, then officers must be properly trained in conflict avoidance and crisis management techniques. How successful an officer is at avoiding violence is a function, at least in part, of how well trained the officer is in defusing emotionally charged situations. Directions for Future Research In examining the use-of-force landscape and in discussing the findings of this research, at least four important areas remain unexplored. First, we know very little about the effectiveness of various types of nonlethal force used by police. What is needed is a comprehensive evaluation of the effectiveness of all types of police force commonly used in street-level encounters. Second, research is needed that identifies in detail the sequential order of how violent encounters unfold. As noted above, there is a great need to develop an interactive model that can better explain the active and reactive aspects of these encounters. Although anecdotal evidence is abundant, there is little empirical research on what factors immediately trigger the use of force by and against police, how force is actually used by suspects against police, and how officers respond. Third, little reliable research exists that identifies the extent of police use of excessive force. Although we can say with relative conviction that police use of force occurs on an infrequent basis, we cannot conclude with nearly the same certainty how many of those incidents involve excessive force. Finally, there is a need to explore measurement issues and uses of the force factor. Studying police use of force without taking into account levels of suspect resistance should be avoided. Research results that do not include the relative measure of force fail to impart a thorough understanding of the police-public encounter. Creating force factor scores for individual officers, assignments, units, and departments can be an important step in understanding and controlling police use of force. Of course, measuring excessive force is highly problematic; indeed, even defining excessive force is difficult and definitions may vary considerably depending on the situation.[5] In spite of this difficulty, if we consider the importance to the Nation of knowing how often its police officers abuse their authority, comprehensive research on excessive force must continue to receive a high priority. Notes 1. Greenfeld, Lawrence, A., Patrick A. Langan, and Steven K. Smith, Police Use of Force: Collection of National Data, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, November 1997, NCJ 165040; and McEwen, Tom, National Data Collection on Police Use of Force, Bureau of Justice Statistics, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, April 1996, NCJ 160113. 2. Alpert, Geoffrey, and William Smith, "How Reasonable Is the Reasonable Man?: Police and Excessive Force," Journal of Criminal Law and Criminology, 85(1994): 481-501. 3. Alpert, Geoffrey, and Roger Dunham, The Force Factor: Measuring Police Use of Force Relative to Suspect Resistance, Washington, DC: Police Executive Research Forum, 1997. 4. Alpert, Geoffrey, Police Pursuit: Policies and Training, Research in Brief, Washington, DC: U.S. Department of Justice, National Institute of Justice, May 1997, NCJ 164831; and Alpert, Geoffrey, and Lorie Fridell, Police Vehicles and Firearms, Prospect Heights, IL: Waveland Press, 1992. 5. Alpert, Geoffrey, and William Smith, "How Reasonable Is the Reasonable Man?: Police and Excessive Force." ------------------------------- Sidenotes Geoffrey P. Alpert, Ph.D., is Professor of Criminology, College of Criminal Justice, University of South Carolina. Roger G. Dunham, Ph.D., is Professor, Department of Sociology and Criminology, University of Miami. "There do not appear to be any empirically validated research studies which support the assertion that race, ethnicity, gender, or age of police officers are related to misuse of physical or deadly force."--New York State Commission on Criminal Justice and the Use of Force, Report to the Governor, Vol. I, New York: New York State Commission on Criminal Justice and the Use of Force, May 1987: 301. ------------------------------- Force-related terminology Suspect resistance: No resistance. Suspect was cooperative and followed all verbal instructions given by the officer. Slight resistance. Suspect resisted the officer's actions and the officer had to use strong directive language and/or minimal force (skills) to encourage suspect to cooperate and follow directions. Moderate or high resistance. Suspect impeded officer's movement or resisted cuffing or placement in a car. This level of resistance required the officer to use arm/wrist locks and/or distraction techniques or fighting skills to gain compliance and control. Violent or explosive resistance. In this, the most extreme, level of resistance, the suspect struggled or fought violently and required the officer to (1) use fighting skills to disengage, (2) use a chemical agent, baton, or firearm, or (3) continue fighting to gain control. In some cases in this resistance category, the officer decided that he or she needed to use weapons or other special tactics to gain control instead of engaging the suspect directly. Officer force: No force. Officer used typical verbal commands. Slight force. Officer had to use strong directive language and/or minimal physical force to encourage the suspect to cooperate and follow directions. Forcibly subdued suspect with hands. Officer used an arm/wrist lock, takedown, block, punch, or kick, and/or struck or wrestled the suspect. Forcibly subdued suspect using methods other than hands. Officer used chemical agent, baton, gun, or other special tactics or weapons. ------------------------------- Chapter 6 A Research Agenda on Police Use of Force by Kenneth Adams This chapter puts forth a research agenda on police use of force, with special attention given to issues of excessive force. In some respects the task is easy. We have a limited empirical understanding of the problem; hence, any new research will be a welcome addition to the current state of knowledge. In other respects, however, the task is difficult. If left to unbounded development of inquiry, a research agenda quickly loses focus and takes on a scattershot quality that diminishes the return on one's scientific investments. Research budgets are not unlimited, so it is important to establish strategic priorities. Also, elements of uncertainty and unpredictability in the scientific enterprise make it difficult to say which avenues of investigation will be most fruitful. Three general considerations guided the development of this research agenda: --Research should provide new knowledge that significantly increases our understanding of the problem while operating within real-world constraints. --Research should be policy relevant. It should lead to improvements in our efforts to deal with problems. --Research activities, taken as a whole, should be comprehensive and systematic. Development of a research agenda on use of force by police entails several considerations. Use of force is a relatively infrequent event. Consequently, sample sizes will tend to be small and measurement error will tend to be large. This situation places the researcher on the horns of a dilemma. If one decides to live with the problem of small sample size, there will be serious limitations on one's ability to conduct statistical analyses. As a result, conclusions will be subject to heavy qualification, and the value of new information will be diminished. If one opts for larger samples, the cost of research increases dramatically. This means that fewer investigations will be carried out for a given sum of money, bringing the cost-benefit issue to the foreground as a major consideration. There are several strategies to be discussed later, in particular triangulation and use of targeted samples, that help the researcher walk a middle road through this dilemma. From a pragmatic point of view, use of force by police is difficult to study because it is a sensitive and politically charged topic, especially when issues of excessive force are involved. In many instances, research will depend on the cooperation of police agencies, and for police administrators cooperation with researchers can be a mixed bag, carrying the possibility of harm as well as benefit. Notions of professionalism and commitments to excellence push police executives to deal with use-of-force problems. Yet, they also are keenly aware of political realities. Pressures from within the organization, such as that from unions, for example, can offer resistance to reform, and groups outside the organization, such as elected government officials and community organizations, often stand poised to gripe at the first hint of failing. The project of the International Association of Chiefs of Police (IACP) is viable because it collects data under two important conditions: voluntariness and strict confidentiality (see chapter 3). These conditions serve to insulate police departments from potentially negative repercussions. They also work to distort the scientific picture by presenting data only from departments willing to cooperate and by limiting access to important information regarding the characteristics of those police departments. What in the political arena is considered prudent is in the scientific arena a handicap, operating as a source of error and bias. Within the context of these challenges and limitations, a multifaceted research agenda that moves forward on several fronts is needed. Four priority areas for future research on police use of force are proposed: --Establishing the conceptual boundaries of "excessive force." --Improving measurement. --Identifying important variation and correlates. --Evaluating efforts to manage use-of-force problems. Establishing the Conceptual Boundaries of Excessive Force In studying police use-of-force problems, we eventually come to ask, "What is excessive force?" and "How does excessive force relate to other forms of violence and misconduct by police?" At an abstract level, the concept of excessive force is not hard to delineate. A variety of definitions already exist, some of which are widely used and accepted. However, when it comes to applying a definition of excessive force to individual situations, thereby rendering a judgment that a particular police officer acted wrongly in a specific situation, linguistic neatness breaks down into untidy legal complexity. Judgments will differ, sometimes widely, about whether force used by police in a specific situation was excessive. If we are to advance our understanding of excessive force, we need not only a common definition but also common perceptions of which instances fit the definition. Likewise, the relation of use of force to other issues of police behavior and misconduct is easy to delineate conceptually. As a practical matter, however, we do not know the size and strength of these hypothesized relations. This information will allow us to determine when interrelated phenomena should be treated separately or as one, or when one phenomenon can be substituted for the other, as convenient or necessary. Judging what force is excessive Research on excessive force by police must grapple with the problem of determining what is excessive. The complication is that labeling force as excessive involves a judgment, the outcome of which depends on the information that is available, the criteria that are used, and the manner in which the criteria are applied. An obvious problem is that the criteria used in making judgments about excessive force are not always the same, leading to differences of opinion. Judges apply legal standards; police administrators apply professional standards; and citizens apply "common sense" standards. The information available on which to make these judgments also varies considerably. Judges typically render judgments in the context of an adversarial trial process that, in principle, is designed to lay bare all relevant information. Administrative review of police conduct generally takes place behind closed doors by trained professionals who have to temper objective reasoning (e.g., going by the book) with subjective understanding (e.g., "putting themselves in another officer's shoes"). Regarding the thoroughness of internal police investigations, there is a professional "push" to be exhaustive and demanding and a collegial "pull" to be politic and practical. The public tends to make judgments about police use of force with more emotion and less information than do judges or police administrators. They react, somewhat reflexively, to newspaper and television reports of excessive force or stories from friends and neighbors claiming that the police victimized them or someone they know. Finally, there are differences in how often decisions are made and in the consequences of decisions. Lawsuits alleging excessive force by police are relatively uncommon, although the consequences of a single decision can immediately and acutely transform entire organizations and communities. Police administrative decisions regarding excessive force occur more frequently than judicial decisions, usually in the context of disciplinary proceedings or policy reviews. The short-term consequences of these decisions focus on the individuals involved in the situation. The public's judgments on excessive force occur habitually, being made by innumerable people on a continual basis. The immediate consequences of these decisions are seemingly trivial, but the cumulative long-term effect can be devastating in terms of police-community relations. We know something about judicial and administrative judgments of excessive force because decision criteria are spelled out, decisionmaking processes are in place, and decision outcomes are visible. However, much more can be known about these issues, particularly with regard to variations across localities and organizations. Perhaps because the public's judgments are so silently ubiquitous, we know almost nothing about how these decisions are made. There is a need for research on judgments of excessive force. How do police differ in their judgments from judges and lawyers? The answer holds implications for the outcome of lawsuits. How does the public differ from police administrators in their judgments? In an era of community policing, the answer holds serious implications for police-community relations. How do people vary in their judgments with regard to personal characteristics, such as race, age, gender, education, and experience with the criminal justice system? Answers regarding racial differences, for example, are relevant to perceptions of fairness and building of community solidarity. Judgments of excessive force are complex because they involve many considerations. Which factors carry the most weight in making these judgments? What role does the actions and characteristics of the suspect play? How are the actions and characteristics of the officer weighed? How do situational factors, such as time of day, location, and number of bystanders, influence judgments? What role does the police organization's rules and procedures play in judgments? What does the public, as well as lawyers and police, think about legal and professional criteria for determining excessive force? These issues are important if we are to know what different kinds of people think about the excessive force problem. Likewise, these issues are important if we are to understand and interpret data on excessive force generated by courts and police organizations as well as by survey research. The problem of excessive force relates to several other areas of police behavior that raise similar issues. In particular, excessive force overlaps with use of force generally (excessive and otherwise), civilian injuries by police, police misconduct and illegal behavior, and work performance. Focus on the broader picture For several reasons, it is useful for research to focus on the broader picture of all use-of-force events. First, the process of identifying use-of-force incidents is more factual and less judgmental than that of identifying excessive force incidents. Hence, there are fewer problems, both normative and scientific, in counting and studying events. Also, recordkeeping systems already exist in many police departments that document use-of-force events in considerable detail. This means that data are readily available for analyses within and across departments. Second, a focus on all use-of-force events leads us to consider the problem of excessive use of force, as distinguished from the problem of excessive force. This turn of phrase describes situations in which use of force, though arguably legitimate and justifiable on a case-by-case basis, can be seen in the aggregate as disquietingly frequent and potentially problematic. A focus on excessive use of force shifts the question from "Was force used legitimately in this situation?" to "Why is force being used so often?" By concentrating on use-of-force levels that are potentially counterproductive and hence inadvisable, we indirectly address the excessive force problem because in scrutinizing use-of-force incidents for necessity, excessive force events are likely to stand out in stark relief. Third, a broader perspective, one that emphasizes questions of how often force is used and whether force is used too often, leads us to consider a wider array of options for dealing with problems of police-public violence. By not sorting use-of-force events into categories of "good" and "bad" and then focusing on the "bad," discussions of solutions can more easily move away from issues of punishment and discipline. By downplaying the moral element, we more quickly come to consider the variety of remedial mechanisms that exist in police organizations, such as rules and procedures, training, and other programs. Thus, our attention is drawn to aspects of the formal organization that can be manipulated in ways that address use-of-force problems. Another approach to the excessive force problem is to concentrate on instances in which police injure civilians. Not every excessive force incident involves civilian injuries, and not all injuries to civilians involve excessive force. A focus on injuries is nonetheless useful because it emphasizes outcomes of police use of force that are serious and consequential. Also, police injuries to civilians are easy to count because they are relatively unambiguous and because medical records, generated independently of police records, can be used to investigate these events. Finally, another perspective is to view the excessive force problem as part of a much broader set of police misbehavior. It may be that officers who engage in excessive force are "problem" officers in a much more general sense. On the job, they may disregard procedures, disobey rules, show poor judgment, and have bad attitudes. Off the job, they may drink too much, abuse their spouses, and get into fights and traffic accidents. An officer's use-of-force difficulties may be part of a larger constellation of problem behaviors that become manifest in citizen complaints, disciplinary actions, and poor performance evaluations. These relations, which should be investigated, are theoretically important because they may require that we broaden our explanations of excessive force to include other problem behaviors. Such theoretical explanations are policy relevant to the extent that efforts to deal with use-of-force problems require a full understanding of police misbehavior in order to be successful. Finally, as a practical matter, these relations can be used to identify problem officers early on, before their problems swell and get out of hand. Overlap and convergence Research should focus on relations among various use-of-force problems, emphasizing areas of overlap and convergence and concentrating on issues of measurement and prediction. Some questions that need to be answered are as follows: --What is the extent of overlap among rates of use of force, excessive force, and civilian injury? --How stable are these relations? --To what extent can one measure substitute for the other? --Can overall rates of force be used to identify police departments or police officers that are likely to have problems with excessive force? Can rates of civilian injury be used in the same manner? --Can problem behaviors in the workplace be used to identify officers who are likely to have problems with excessive force? Can behaviors outside the workplace be used in a similar manner? Improving Measurement Measurement problems are common in the social sciences. The quality of data typically can stand major improvement. This too is the case with police use of force. Although it is desirable to have a sound, unambiguous, accurate measure of excessive force, for example, in police departments throughout the country, it could take a decade or more of work before we begin to approximate this goal. In the near term, we will have to make do with what we have and work for small incremental improvements in data availability and quality. Multiple indicators of use of force In situations in which data contain a fair amount of "noise" (meaning error) that masks the true picture of what is being measured, and in which significantly improved measurement is not feasible, perhaps due to the expense, time, or effort that is required, use of multiple measures and triangulation of data is an advisable strategy. Multiple measurement involves use of several measures that capture different perspectives on the object being measured. The idea is that several measures, when compared with and contrasted against each other, will offer a more complete picture than any one measure, which, standing alone, is known to be incomplete. Triangulation is an analytic approach based on multiple measures. The approach capitalizes on the fact that each measure is imperfect in different ways and that each offers a different perspective on the problem. If all measures point in the same direction, one's confidence in the results of statistical analyses is increased. The underlying phenomenon is robust enough that a consistent "picture" emerges through the noise. If, however, measures point in different directions, then one must consider the specific limitations of each measure in making an assessment. Sometimes it is possible to combine a variety of measures into an index that is robust because the combined errors of the individual measures tend to cancel each other out. The various data sources that are available to investigate issues of police use of force already have been discussed elsewhere in detail.[1] The focus below is on steps needed to improve data sources so that we can move toward a multiple-indicator approach. Official records For several reasons, official records are and will continue to be the major source of information on police use of force generally, as well as on excessive force. Although official records provide less-than-perfect data, from a practical viewpoint the advantages strongly outweigh the disadvantages. Among the benefits of a good system of official recordkeeping are wide geographic coverage, collection of detailed information, the possibility of linking various record systems for a more comprehensive picture, and up-to-date statistics. Also, because data collection takes place at the local agency level, there is the possibility of distributing the burden of data collection across many organizations. The drawbacks of official record systems typically are lack of standardization across agencies and poor data quality. These problems are exacerbated in the context of national coverage, which involves coordination of many different agencies that are geographically dispersed. For this reason, simple, easy-to-use data collection procedures are to be preferred over complex, difficult-to-master procedures, particularly in the early stages of establishing a reporting system. Use-of-force reports Many police departments require a use-of-force report to be filed any time an officer uses force against a civilian in the line of duty. The report describes the circumstances of the event and the nature of the force that was used. These reports can be extremely valuable on a number of counts. Use-of-force reports allow police administrators to monitor the number of incidents by individual officers, geographic areas, or organizational units. They provide a basis for periodic administrative review to determine whether proper procedure is being followed when force is used. They offer the possibility of supplementing raw counts of incidents with detailed information on officers, suspects, and the circumstances of encounters. Finally, documentation contained in use-of-force reports may prove useful in defending the department's actions in litigation. Although police officers now spend a lot of time on paperwork, use-of-force reports are not unduly burdensome to officers because they use force infrequently in their work. Garner and his colleagues surveyed all arrests in the Phoenix Police Department for 2 weeks regarding use of force.[2] During this time, 1,585 adult arrests were surveyed, of which about 1 in 5 involved any type of physical force by police. Thus, a requirement that every use-of-force incident be documented translates into roughly 150 reports per week for a police department with more than 3,000 employees. Since most of the force that police use is at a relatively low level (i.e., holding, grabbing), the number of reports can be reduced by raising the threshold for reporting. All police departments should be required to maintain use-of-force reports. If the officer's use of force is serious, it is to the organization's benefit to document its occurrence. Furthermore, these reports should be subject to periodic review and scrutiny, both in terms of general trends and on a case-by-case basis. Finally, police departments should be required to report periodically their statistics on use of force to a State or Federal agency. Although some effort will have to be made to standardize reporting procedures, this suggestion could be implemented fairly quickly. As computer use becomes more widespread among front-line police officers, reporting requirements should become less burdensome. Detailed suggestions for a national reporting requirement can be found elsewhere.[3] Civilian injury and hospitalization records Another source of information on police use of force is medical records on injuries that civilians receive at the hands of police. These records focus on injury and treatment, and they are maintained by health-related agencies rather than police agencies. Medical records capture the most serious use-of-force incidents, including the most serious instances of excessive force. All police departments should be required to maintain and publish statistics on civilian injuries caused by police officers. Police agencies should work in cooperation with hospitals and emergency rooms to develop these statistics. Compiling and publishing statistics on civilian injuries should not place a heavy burden on police agencies, again because these events do not happen very often. Indeed, one might think along the lines of a supplemental Uniform Crime Reports reporting system parallel to that on law enforcement officers killed in the line of duty but concentrating on civilian injury. The data can involve simple counts of events, broken down so that criminal suspects can be distinguished from other civilians and the method of injury can be identified. Court records Lawsuits involving police use of force tend to contain allegations of excessive force, although court records are poor measures of how often such incidents occur, because the allegations probably constitute a highly biased and selective subset of police-citizen encounters. Use-of-force incidents that lead to lawsuits typically reflect some combination of a highly motivated plaintiff, egregious action by police, and a tenacious lawyer. The occurrence of police misconduct is only one element in the litigation equation and it may not be the most important. However, lawsuits are directly relevant to the issue of how much use-of-force transgressions by police cost society. Knowing the costs incurred under current policies and practices, as measured by monetary awards to plaintiffs, can help put the price of reform into broader financial perspective. These cases, which merit systematic study, show that there are some costs to having police use force, although it is not clear how much of these costs can be attributed to excessive force. Presently, information about lawsuits against police is distributed across several government agencies, and there is little coordination and feedback. City attorneys may not routinely notify police administrators about the outcome of litigation, so they are not always in a position to follow up with appropriate administrative action. Financial liability is an important aspect of the use-of-force problem, and citizens are entitled to know what police misconduct costs them. Thus, every police department should make available on a periodic basis data on the number, types, and outcomes of lawsuits filed against the department, separately identifying those that involve allegations of excessive force as well as other use-of-force issues. Survey methods Survey methods tend to be relatively inefficient at capturing police use-of-force incidents. The yield depends to some extent on the definition of force and on the timeframe being referenced. Definitions that include verbal threats will capture more incidents than definitions limited to physical contact. Likewise, questions about lifetime experiences will capture more incidents than questions about the past 6 months or the past year. Two strategies for addressing the inefficiency of survey methods in studying police use of force are using targeted samples to increase the yield of incidents and supplementing survey research projects on other topics with use-of-force questions. Both these techniques, which offer cost-effective ways of studying infrequent events, are underutilized and could be used to greater advantage in research on use of force by police. Targeted sampling. Targeted samples trade breadth of coverage for a focus on high-risk persons or situations. Rather than covering the entire country, a survey might be restricted to big cities. Instead of surveying an entire city, only high-crime or low-income areas might be surveyed. In lieu of sampling all adults, only young adults or males might be sampled. Instead of questioning all civilians, only persons who have been arrested might be queried. If survey efforts concentrated on high-risk persons or situations, the number of use-of-force incidents captured in a study could be increased, which would mean that statistical analyses would be more reliable and potentially more sophisticated. The data can be used for a variety of purposes. They can be used in epidemiological fashion to generate more reliable point estimates of police use of force across various conditions. They can be used to gain a more complete theoretical understanding of use-of-force incidents using forms of elaboration analysis that examine subgroups within the sample. They can be used to study important subpopulations within society, both in terms of point estimates and theoretical analyses. Depending on the purpose of the investigation, the characteristics of the sample being targeted and the types of questions being asked will vary. If the scope of persons included in a survey is restricted, some questions will go unanswered. For example, if a survey concentrates on males, information on females will not be collected. The danger here is that by restricting inquiry, we may unintentionally confirm our misconceptions. Nonetheless, targeted samples may be a sensible tradeoff when research funds are modest and when information is available to weigh the cost of the tradeoff. In the Phoenix study, Garner and his colleagues surveyed all persons arrested over a 2-week period.[4] A reanalysis of the data could tell us if the number of use-of-force incidents captured for the same expenditure of resources would be greater by limiting coverage to Friday and Saturday. The data also could tell us how use of force differs by weekday and weekend, thus informing us about the biases of a targeted "weekend" survey. The compromise, of course, is that the results would only be generalizable to weekend arrests. The supplement approach. Another strategy for increasing return on research investments is "piggybacking" one's research onto another project. The strategy is an opportunistic one that involves locating another research project that can be supplemented to suit one's needs. The "other" research project might be in the early stages of planning or initiation, or it might be an ongoing longitudinal effort. If one is starting a research project, finding "partners" for combining several research agendas into a single research vehicle is economical because development and implementation costs can be shared. Working out differences among partners can be difficult, however, and can lead to compromises that seriously detract from the utility and integrity of the research when viewed from the perspective of one or all of its component parts. With regard to an ongoing research project, a secondary or supplemental project can capitalize on the efforts of a primary project in such areas as sample selection and training of interviewers. However, the secondary project will have little control over almost all aspects of the research design. This is fine if the design of the primary project meets the needs of the secondary project. To the extent that this is not the case, the piggyback approach extracts a cost from the secondary project in terms of scientific value that needs to be debited against the economic savings. Also, researchers may give only limited opportunity to others to become involved in their ongoing research because they are concerned about protecting the integrity of their project. The Police-Public Contact Survey (see chapter 2) by the Bureau of Justice Statistics (BJS) is a good example of the "supplement" strategy.[5] BJS added a set of questions about police contact and police use of force to the National Crime Victimization Survey (NCVS), which is a 3-year, seven-panel rotating longitudinal design carried out by the Bureau of the Census. In this collaboration, which involved a pilot test of a survey questionnaire, BJS used only one of the NCVS panels. As a vehicle for testing new survey instruments, NCVS is very cost efficient, functional, and expeditious. The BJS supplement also highlights two important aspects of research on police use of force. First, surveys of the general population are inefficient at capturing encounters in which police use force against the public. Of the 6,421 survey respondents nationwide, only 14 reported that police used or threatened to use force against them in the past year, leading to an annual estimate of 500,000 persons nationwide. Calculation of the margin of error around this estimate was not attempted for the pilot survey, but it is likely to be large because of the small number of cases involved. BJS will address these issues in a second pilot survey, which will use a sample more than 10 times larger than that used in the first pilot test. Another important feature of the BJS survey is that it illustrates the complexity of use-of-force encounters. In the BJS pilot test, respondents who indicated that police had used force against them could have been asked more than 100 questions, depending on their experiences. Recommended changes to the survey, for the most part, include the addition of more questions to describe use-of-force incidents in greater detail. Although the length of the survey is not problematic, given that the average time to complete it for persons who had contact with police was 10 minutes, the survey makes it evident that a substantial amount of information has to be collected in order to eliminate ambiguity and ensure that the data are most pertinent to policy issues. Observational methods Observational methods are highly inefficient at capturing use-of-force incidents not only because these are infrequent events but also because a researcher can observe only one or two officers on assignment at a time. Bayley and Garofalo had six observers spend a total of slightly more than 2,000 hours observing police officers in the field.[6] These observations identified 37 use-of-force incidents by police, the majority of which involved relatively low levels of force. A development worth mentioning is the use of video cameras in police cars. In many cases, the camera is set to operate automatically when an officer leaves the car. The purpose is to capture police-civilian interactions so that they are available for later review in case there is a complaint about the officer's behavior. Although some details or aspects of the encounter may be missing, incidents so recorded can be considered as observational data. These data can provide information on the frequency and characteristics of various types of police-public encounters within the limited context of automatic recording from a patrol car. Police management might use the recordings in its efforts to oversee the behavior of front-line officers. These video recordings also might be used in training programs and in research, when a high degree of verisimilitude is required. A limitation of this strategy, however, is that it can be fairly labor intensive to review the many videos. Identifying Important Variation and Correlates The main purpose of scientific research is to identify variation in things of interest and link this variation to other factors or variables, which are called correlates. In the beginning stages of research, scientists usually focus on accurate measurement of variation. Thus, priority should be given to development of tools that allow us to determine how often excessive force occurs. Social science research, however, moves down several paths at the same time. While researchers are busy trying to measure variation, they quickly jump to questions of what factors are associated with higher or lower levels of excessive force. Correlates, if reliable and substantial, allow us to predict. If, for example, the incidence of excessive force is positively related to the incidence of violent crime, we know where problems of excessive force will be greatest. Another practical side of correlates is that they can allow us to take steps to mitigate problems. Thus, if the incidence of excessive force is negatively correlated with levels of police officer supervision, increasing supervision of line officers may reduce or eliminate the problem. Finally, correlates can help us to understand why things happen as they do, and so they are the building blocks of theories. Again, if we know that the incidence of excessive force is highest among new officers, we might explain this relation by pointing to a lack of experience and training. The correlates of excessive force might be quite different from the correlates of excessive use of force, however. Geographic and temporal variation In searching the list of possible correlates of excessive force, some factors stand out as potentially more important than others. In particular, variations in the incidence of excessive force that are associated with time and place are especially significant. Geographic variation, in terms of region of country or size of place, allows us to focus on problems of excessive force and excessive use of force in concrete terms. There is a tendency in police research to concentrate on a handful of departments in large metropolitan areas. There are several practical reasons for this. Big cities are hotbeds for all sorts of social problems, and their large populations make them good places to study infrequent events. Also, police departments in large cities tend to have resources that facilitate research, such as computerized recordkeeping systems. For these reasons, which are largely matters of practicality and convenience, police departments in medium- and small-sized cities often are overlooked as potential research sites. Larger metropolitan areas will contain a greater number of use-of-force incidents, and probably excessive force incidents, than smaller areas. However, this may be the result of more police-public interactions or more police officers. When measured as rates per 1,000 arrests or per 100 front-line officers, use-of-force problems actually may be greater in smaller areas. Also, taken as a group, the many medium and small cities that dominate our country may constitute the greater part of the use-of-force problem. Thus, there is a need to broaden the scope of police research to include departments in medium and small cities. Without this research, we will not know the answers to many important questions about the scope of the excessive force problem and about the changing nature and distribution of the problem across the country. Studying temporal variation is important because it allows us to determine whether use-of-force problems are getting better or worse. This information permits forecasts about the future, which may be important in galvanizing the perceived need for action. It also may provide a general sense of how well efforts to curb use-of-force problems are doing. An ability to investigate geographic and temporal variation requires the collection of standardized data across places and over time. The IACP project (see chapter 3) is a move in this direction, although the voluntary nature of the reporting system limits the utility of the data. A voluntary system--even one incorporating anonymity--allows the worst departments to avoid participation, leading to underestimation of the problem. The need for a fully nationwide reporting system on police use of force already has been discussed at length, and viable recommendations on how to implement the system already have been made.[7] Significant improvement in understanding police use-of-force problems depends in large measure on significant advancement toward this goal, which has been slow to date. Correlates and use of force There are correlates that span individual, situational, and organizational factors that may be related to the incidence of excessive force. At the individual level, personal characteristics of police officers and civilians involved in use-of-force incidents, viewed separately and in combination, may be important. Age, race, gender, education, and economic status are typical correlates of social behavior, and information on these factors will help fill out the use-of-force picture in terms of social classifications. Criminal justice-related factors also are relevant. Researchers would be interested in work-related variables, such as officers' experience, training, rank, and nature of work assignments. Also of interest would be use-of-force histories, civilian complaints, disciplinary actions, and work performance ratings. Relevant criminal justice information about suspects would include prior criminal record, history of violence, and gang involvement. Situational factors deal with the social and physical context in which use-of-force events take place. Some important factors to be investigated are characteristics of the physical setting, circumstances regarding the initiation of police-civilian encounters, possible drug and alcohol use by the suspect, presence of weapons, number and types of primary actors involved in the situation, number and types of secondary actors (e.g., bystanders or witnesses), and immediate actions that precipitate use of force, with a focus on criminal activity, provocation, and threats, if any. Organizational factors concentrate on aspects of police departments. Although formal aspects of the organization are easier to document, informal aspects of the organization are relevant as well. Some factors to examine are: --Number of employees. Do larger departments tend to have more use-of-force problems? --Number and types of administrative units, with a focus on departments that control misconduct and violence. Does an ethics unit or an integrity unit make a difference? Does civilian review matter? --Personnel allocation. Is the ratio of supervisors to line officers related to the frequency with which force is used? --Expenditures. Do departments with higher pay scales have lower rates of use-of-force problems? --Workforce characteristics. Do departments with a large proportion of minority officers, female officers, experienced officers, or college-educated officers have lower rates of excessive force? --Training and supervision policies. Are the frequency, length, type, and quality of training programs related to the incidence of excessive force? --Departmental philosophy. Does community policing bring fewer use-of-force problems? Do broken-windows or zero-tolerance philosophies increase these problems? Under the heading of organizational factors might be included items external to the organization that shape the work environment. These factors include crime rates and characteristics of the population being served, such as poverty rates. Alternatively, these might be labeled community characteristics. Evaluating Efforts to Manage Use-of-Force Problems The last part of the proposed research agenda focuses on evaluation of attempts to manage police use of force. Here the concern focuses on activities that are specifically targeted at dealing with use-of-force problems. These activities might include training, technologies that provide police officers with new options for safely and effectively controlling recalcitrant suspects, methods for identifying rogue officers in advance, management strategies for controlling officer behavior in the field, or programs designed to change officer attitudes and behaviors in ways that reduce the use of excessive force. Technologies Among the new technologies developed for controlling criminal suspects, pepper spray and stun guns have received considerable attention. These technologies often are promoted as alternatives to lethal force, although they have utility in their own right as tools for behavior control. Research by Kaminski and colleagues suggests that pepper spray is effective in controlling disorderly suspects and may reduce injuries to both officers and suspects.[8] In addition to confirming these findings, research should investigate where these technologies fit into the use-of-force continuum. To what extent are these new technologies used as alternatives for higher or lower levels of force? Also, what evidence is there to suggest that these new technologies are abused and thus are contributing in some way to the excessive force problem? The Christopher Commission report, through its "list of 44" problem officers, drew attention to the fact that a handful of Los Angeles police officers had extensive histories of use-of-force problems and that management often was unresponsive to this situation.[9] Since that time, there has been an increased interest in information systems that can identify officers who are at high risk for using excessive force. In essence, these systems are prediction devices that are updated with new information on a regular basis. Criminal justice researchers have a long history of experience with prediction devices as applied to offender behavior, and many of the same problems and issues that surfaced in that context are germane here. Practical questions (such as availability and quality of data), methodological issues (such as selection of statistical techniques and prediction error), and ethical issues (such as consequences of potentially noxious classification and privacy of information), all remain to be explored in relation to using information systems to help manage use-of-force problems. Another technology that may influence use of force is video cameras mounted inside police cars to record police-public interactions. It was mentioned previously that video recordings could be used for research and administrative purposes and as evidence in lawsuits. A possible consequence of this technology is that it may make officers more circumspect in their use of force, which may show up as fewer use-of-force incidents and fewer complaints about excessive force. Comparisons between departments that are using this video technology and those that are not could address this question. Training Some attempts to deal with use-of-force problems are based on strategies that target the problem at a broader level. A good example of this approach is training programs designed to promote ethical behavior, enhance police-community relations, or train officers in verbal techniques for achieving compliance from civilians. Excessive force issues often are addressed in these programs. In some instances, training programs that address use-of-force issues are part of a more general effort to change police organization and culture. Perhaps the most popular of these efforts is community policing, which emphasizes joint working relationships between police and community to maintain order and promote justice. To the extent that community policing leads to a sense of partnership that includes feelings of fellowship based on a sense of common humanity, antagonism between police and the public should be mitigated and propensities for police to abuse authority or for the public to perceive abuse by police should be reduced. Part of the agenda for community policing evaluation research should be to determine whether a reduction in use of force and complaints of excessive force occurs. Research also should investigate whether changes in attitudes about the use of force occur for police and civilians. Summary This chapter outlines a research agenda on police use of force, giving special attention to problems of excessive force. A variety of questions are raised, both reflecting the complexity of use-of-force issues and the relative paucity of our knowledge about use-of-force transgressions. Four sets of research objectives are identified. First, more work is required on what various people have in mind when they refer to excessive force and on how they adjudge specific instances of police behavior when questions of excessive force arise. We need to study the general public, members of minority groups, police administrators, patrol officers, judges, and offenders, among others, to understand how they think about excessive force. This research is important because social problems often require shared solutions, and shared solutions require a common basis of understanding and mutual respect for differences in views. Second, more and better data on police use of force is needed. Most discussions about police use of force occur in an empirical vacuum where arguments are made without the benefit of solid, useful information. Various suggestions for a national reporting system on police use of force have been made, but progress toward this goal has been slow, even though the information collected by such a reporting system could be extremely valuable. As an interim step, it is recommended that all police departments maintain records documenting all instances of police use of force and all injuries to civilians, and that this information be reported regularly to the public. Maintaining these records will not pose a great burden to police agencies because use of force is infrequent. Furthermore, the knowledge gained should be well worth the effort. Third, also required is research on how use of force by police varies across time, across cities, and across individual police departments. Further, we need research on individual, situational, and organizational factors related to variations in use-of-force levels, along with excessive force levels. In the process of identifying correlates of use of force and excessive force, we will increase our theoretical understanding of these events and advance our ability to predict problematic situations. At an aggregate level, research also should concentrate on the relation between excessive use of force, meaning types of situations in which force is used with disturbing frequency, and excessive force, meaning types of instances in which police use more force than necessary. Finally, we need to identify, document, and evaluate interventions, changes, and reforms that may mitigate police use-of-force problems. Although our understanding of these problems is incomplete, the urgency of the situation requires that we seek out actions that are likely to have a beneficial impact. In some cases, such as community policing, where a broad spectrum of change is involved, what is required is sensitivity to use-of-force issues in evaluation studies. In other cases, such as with police ethics and integrity units, there may be a track record of impact on the organization that is waiting to be investigated. In addition, the effectiveness of activities that specifically target use-of-force issues, such as officer training programs and information management systems, needs to be determined before we can genuinely advocate widespread adoption of these programs. In other chapters of this report, recently completed and ongoing research projects on police use of force are described. As a group, these research projects illustrate rigorous methodologies, sophisticated measurement strategies, cost-effective design features, and the utility of comparisons across several police departments. These projects provide a good foundation for developing a research agenda on police use of force. Yet, substantially more research is needed. Useful, reliable, sound knowledge provides the best venue for society's attempts to deal with the pernicious consequences of use-of-force transgressions by police. Notes 1. McEwen, Tom, National Data Collection on Police Use of Force, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, April 1996, NCJ 160113; and Adams, Kenneth, "Measuring the Prevalence of Police Abuse of Force," in And Justice for All: Understanding and Controlling Police Abuse of Force, ed. William A. Geller and Hans Toch, Washington, DC: Police Executive Research Forum, 1995: 61-97. 2. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding the Use of Force By and Against the Police, Research in Brief, Washington, DC: National Institute of Justice, November 1996, NCJ 158614. 3. McEwen, Tom, National Data Collection on Police Use of Force; and Geller, William A., and Hans Toch, "Improving Our Understanding and Control of Police Abuse of Force: Recommendations for Research and Action," in And Justice for All: 277-337. 4. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn, Understanding the Use of Force By and Against the Police. 5. Greenfeld, Lawrence A., Patrick A. Langan, and Steven K. Smith, Police Use of Force: Collection of National Data, Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, November 1997, NCJ 165040. 6. Bayley, David H., and James Garofalo, "The Management of Violence by Police Patrol Officers," Criminology, 27(1)(February 1989): 1-27; and Bayley, David H., and James Garofalo, "Patrol Officer Effectiveness in Managing Conflict During Police-Citizen Encounters," in Report to the Governor, Vol. III, Albany: New York State Commission on Criminal Justice and the Use of Force, 1987: B1-B88. 7. McEwen, Tom, National Data Collection on Police Use of Force; and Geller, William A., and Hans Toch, "Improving Our Understanding and Control of Police Abuse of Force: Recommendations for Research and Action." 8. Kaminski, Robert J., Steven M. Edwards, and James W. Johnson, "Assessing the Incapacitative Effects of Pepper Spray During Resistive Encounters With Police," Policing: An International Journal of Police Strategies and Management, 22(1)(1999): 7-29; and Kaminski, Robert J., Steven M. Edwards, and James W. Johnson, "The Deterrent Effects of Oleoresin Capsicum on Assaults Against Police: Testing the Velcro-Effect Hypothesis," Police Quarterly, 1(2)(1998): 1-20. 9. Independent Commission on the Los Angeles Police Department, Report of the Independent Commission on the Los Angeles Police Department, Los Angeles: Independent Commission on the Los Angeles Police Department, 1991. ------------------------------- Sidenotes Kenneth Adams, Ph.D., is Associate Professor and Chair of the Criminal Justice Faculty, School of Public and Environmental Affairs, Indiana University-Indianapolis. "Excessive force is almost always a matter of degree...and circumstance. Excessive force complaints frequently arise in situations where the officer, armed and obliged to confront criminality, was doing his job; where he was duty-bound to intervene; where danger was present and some force was necessary; or, where witnesses are limited to the victim and the officer and there is a marked divergence in their views about what happened." --Cheh, Mary M., "Are Lawsuits an Answer to Police Brutality?" And Justice for All: Understanding and Controlling Police Abuse of Force, ed. William A. Geller and Hans Toch, Washington, DC: Police Executive Research Forum, 1995: 235. The Phoenix study was supported by the National Institute of Justice under grant 92-IJ-CX-K028. Among the observational methods alluded to on this page is systematic social observation, a field research method used to study police. "Researchers record events as they see and hear them and do not rely on others to describe or interpret events. The researchers follow well-specified procedures that can be duplicated." --Mastrofski, Stephen D, Roger B. Parks, and Albert J. Reiss, Jr., et al., Systematic Observation of Public Police: Applying Field Research Methods to Policy Issues, Washington, DC: U.S. Department of Justice, National Institute of Justice, December 1998: vii. ------------------------------- Bibliography The publications below are, in whole or in part, related to police use of force and are among those generated by research supported by the National Institute of Justice or the Bureau of Justice Statistics either through grants to outside researchers or through studies by inhouse staff. For information on the availability of a publication whose listing includes an NCJ or FS number, please contact the National Criminal Justice Reference Service (see inside back cover). Alpert, Geoffrey P. Police Pursuit: Policies and Training. Research in Brief. Washington, DC: U.S. Department of Justice, National Institute of Justice, May 1997. NCJ 164831. (The related 60-minute videotape: Alpert, Geoffrey P. Police in Pursuit: Policy and Practice. Washington, DC: U.S. Department of Justice, National Institute of Justice. NCJ 161836.) ------. Helicopters in Pursuit Operations. Research in Action. Washington, DC: U.S. Department of Justice, National Institute of Justice, August 1998. NCJ 171695. Edwards, Steven M., John Granfield, and Jamie Onnen. Evaluation of Pepper Spray. Research in Brief. Washington, DC: U.S. Department of Justice, National Institute of Justice, March 1997. NCJ 162358. Fridell, Lorie A., and Antony M. Pate. "Death on Patrol: Killings of American Law Enforcement Officers." In Critical Issues in Policing: Contemporary Readings, ed. Geoffrey P. Alpert and Roger G. Dunham. Prospect Heights, IL: Waveland Press, 1997. Garner, Joel, John Buchanan, Tom Schade, and John Hepburn. Understanding the Use of Force By and Against the Police. Research in Brief. Washington, DC: U.S. Department of Justice, National Institute of Justice, November 1996. NCJ 158614. (The related 60-minute videotape: Garner, Joel H. Use of Force By and Against Police. Washington, DC: U.S. Department of Justice, National Institute of Justice. NCJ 159739.) Geller, William A., and Hans Toch, eds. And Justice for All: Understanding and Controlling Police Abuse of Force. Washington, DC: Police Executive Research Forum, 1995. Greenfeld, Lawrence A., Patrick A. Langan, and Steven K. Smith. Police Use of Force: Collection of National Data. Washington DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, November 1997. NCJ 165040. Jefferis, E., R.J. Kaminski, S. Holmes, and D. Hanley. "The Effect of a Video-taped Arrest on Public Perceptions of Police Use of Force." Journal of Criminal Justice, 25(5)(1997). Kaminski, Robert J., Steven M. Edwards, and James W. Johnson. "The Deterrent Effects of Oleoresin Capsicum on Assaults Against Police: Testing the Velcro-Effect Hypothesis." Police Quarterly, 1(2)(1998). ------. "Assessing the Incapacitative Effects of Pepper Spray During Resistive Encounters With Police." Policing: An International Journal of Police Strategies and Management, 22(1)(1999). Kaminski, Robert J., and Eric Jefferis. "The Effect of a Violent Televised Arrest on Public Perceptions of the Police: A Partial Test of Easton's Theoretical Framework." Policing: An International Journal of Police Strategies and Management, 21(4)(1998). Mastrofski, Stephen D., Roger B. Parks, Albert J. Reiss, Jr., Robert E. Worden, Christina DeJong, Jeffrey B. Snipes, and William Terrill. Systematic Observation of Public Police: Applying Field Research Methods to Policy Issues. Research Report. Washington, DC: U.S. Department of Justice, National Institute of Justice, December 1998. NCJ 172859. McEwen, Tom. National Data Collection on Police Use of Force. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics and National Institute of Justice, April 1996. NCJ 160113. National Institute of Justice. Oleoresin Capsicum: Pepper Spray as a Force Alternative. Technology Assessment Program. Washington, DC: U.S. Department of Justice, National Institute of Justice, March 1994. ------. Preliminary Investigation of Oleoresin Capsicum. Law Enforcement and Corrections Standards and Testing Program. Washington, DC: U.S. Department of Justice, National Institute of Justice, April 1995. NIJ Report 100-95. ------. Positional Asphyxia--Sudden Death. National Law Enforcement Technology Center Bulletin. Washington, DC: U.S. Department of Justice, National Institute of Justice, June 1995. ------. High Speed Pursuit: New Technologies Around the Corner. National Law Enforcement and Corrections Technology Center Bulletin. Washington, DC: U.S. Department of Justice, National Institute of Justice, October 1996. ------. Pursuit Management Task Force. Research Preview. Washington, DC: U.S. Department of Justice, National Institute of Justice, August 1998. FS 000225. Pate, Antony M., and Lorie A. Fridell, with Edwin E. Hamilton. Police Use of Force: Official Reports, Citizen Complaints, and Legal Consequences. Vols. I and II. Washington, DC: Police Foundation, 1993. Pinizzotto, Anthony J., Edward F. Davis, and Charles E. Miller III. In the Line of Fire: Violence Against Law Enforcement. Washington, DC: U.S. Department of Justice, Federal Bureau of Investigation and National Institute of Justice, October 1997. Scrivner, Ellen M. The Role of Police Psychology in Controlling Excessive Force. Research Report. Washington, DC: National Institute of Justice, 1994. NCJ 146206. ------. Controlling Police Use of Force: The Role of the Police Psychologist. Research in Brief. Washington, DC: U.S. Department of Justice, National Institute of Justice, 1994. NCJ 150063. The following are among the objectives of ongoing NIJ-supported use-of-force research projects, which may result in future publications: --Measuring use of force relative to suspect resistance in two police agencies. --Identifying the frequency, rate, and type of force used in a county police department, including whether the introduction of pepper spray affected use of force or the extent of injuries to officers and civilians. --Determining, through a national survey of police agencies, how they organize their critical-incident response capabilities and what types of actions, including use of force, that officers and civilians take during the incidents. --Surveying a national sample of law enforcement agencies to determine the prevalence, distribution, and principal features of early warning systems as a response to the problem officer and evaluating the effectiveness of such systems in three police departments. --Examining the use of force by police when encountering persons with impaired judgment. --Examining serious assaults on police in two jurisdictions to identify neighborhood risk factors and areas of significant spatial clustering. ------------------------------- To obtain an electronic version of this report, please visit the NIJ World Wide Web site at http://www.ojp.usdoj.gov/nij or contact: National Criminal Justice Reference Service Box 6000 Rockville, MD 20849-6000 Phone: 800-851-3420 Phone: 301-519-5500 E-mail: askncjrs.org ------------------------------- NCJ 176330