Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 Explaining Police Behavior: Organizations and Context P olice behavior is affected by broad forces, including features of the organizations that hire, train, and supervise police, as well as the environment within which they work. Organizational influences are important to understand because factors reflecting police organization and policies present possible points of leverage in shaping police practice. Those seeking to change police, whether they are inside or outside the agency, often pursue organizational changes, including instituting new policies, forming new units, changing supervisory practices, instituting performance quotas, or even hiring a new chief. In contrast, many features of the environment are often beyond "easy human contrivance" (Bayley, 2002), and indeed, police practice is often constrained by the environment. The police may not be able to do much about the personal characteristics of residents, the economy, the distribu- tion of wealth, the influx of immigrants, and the cultural habits of the people who live and work in their jurisdiction. Yet these characteristics do power- fully shape the workload of the police, so understanding the influence of these factors and the boundaries they may set on police practice is therefore an important component of police research. Because the effectiveness of a given policy in promoting a desired practice may vary across different envi- ronments, police leaders may find it especially useful to know when and where those policies are most and least effective. Under some circumstances, environmental conditions can also be an object of policy manipulation by police. For instance, a department faced with an indifferent or hostile community may attempt to change the community's willingness to cooperate with the police in order to engage in more effective crime-control strategies. Or the department may wish to pro- 155
156 FAIRNESS AND EFFECTIVENESS IN POLICING mote the community's capacity for collective action in partnership with the police. Organizing neighborhood groups, working with churches and reli- gious organizations, and educating the public through citizen-police acad- emies are all ways in which the police attempt to change outcomes, such as the crime rate, by changing the environment of policing (Mastrosfki and Ritti, 2000:191-192). When they engage in such environment-changing strategies, they may also change the way in which the department practices. For example, a police force that works closely with community groups may find that it can engage in more intrusive actions in that community without raising the distrust of community leaders. Thus some environmental factors will interest police precisely because they want to try to change them. Research on these questions is not well developed. In order to under- stand the influence that organizational and environmental factors exert, the committee again presents its conclusions in the form of propositions that summarize large bodies of work. In terms of police organization, the com- mittee reviews research on important law enforcement actions that admin- istrative policy or structure might be expected to affect, such as the use of force, both lethal and nonlethal; citizen complaints and legal actions against the organization; arrest and clearance rates; and traffic citations. Each of these represents an important output in its own right, and each might also speak to and help depict a larger organizational style. The environment in which police organizations operate is challenging to parse into meaningful categories. Many external forces can influence how the police act: their immediate surroundings, the historical relation- ship between a neighborhood and the police, the political tenor of the com- munity, and even the regional culture in which they are situated. The com- mittee organized this research in concentric circles: first the neighborhood that an agency serves, then the city or metropolitan area in which it is located, the politicians and civilians that oversee the department, and politi- cal forces--such as federal legislation--that address police operations. Of course, one important external force, the magnitude of which is not yet fully manifest, could not be ignored: as part of its review, the committee examined the effect of the terrorist attacks of September 11, 2001. While little systematic social science data exists on this topic for obvious reasons, the committee reviews what is known and culls from other research litera- tures in order to deepen its analysis of this vital issue. We first consider some of the dependent variables used in this research. They reflect the activities and practices of policing that researchers have attempted to explain. Unlike research reviewed in the previous chapter, which concentrated on the behavior of individual officers, most of the re- search covered in this chapter focuses on police practices in the aggregate-- that is, at the organization level. The committee makes some recommenda- tions regarding measurement issues raised by this research; in addition,
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 157 recommendations have been made about some of them in other chapters of the report. We then turn to research on organizational and environmental factors and conclude with more research recommendations. MEASURES OF ORGANIZATION ACTIVITIES AND OUTPUTS Police organizations, like the people in them, vary considerably. The ability to measure and understand this variation depends on the frequency with which information is collected, the "coverage" or the number and variety of agencies from which collection takes place, the quality of data, and the degree to which the research framework that interprets these data is meaningful and valid. All of the activities or police outputs reviewed in this section are impor- tant to understating the operation of police organizations. Most are ob- tained directly from the agencies, either from surveys initiated by research- ers or by federal data-gathering systems. In a few cases the data come from other organizations, including those in the justice system. In many instances, very few data exist to measure activities or outputs of police organizations, even though they may be a regular feature of police work. Lethal Force During a five-year period in the early 1990s, officers in the Washing- ton, DC, Metropolitan Police Department shot and killed 57 people. Dur- ing the same period, police officers in Chicago--which had five times the population and three times the number of officers--shot and killed 54 people (Leen et al., 1998). What factors explain this considerable variation in the use of lethal force by police in these two cities? A report by the Washington Post examined and dismissed several possible explanations, including the size of the population, the amount of violent crime, the num- ber of homicides, the size of the police force, and the number of arrests for violent offenses (Leen et al., 1998). For instance, DC police averaged 2.3 fatal police shootings per 1,000 officers, compared with 1.5 in 26 other big city police departments and 1.7 in the nine U.S. cities with the highest homi- cide rates (Leen et al., 1998). There is no more important piece of data regarding police behavior than that on the exercise of lethal force and, indeed, data are available nationally and have been collected systematically by the Federal Bureau of Investigation (FBI, in its Supplemental Homicide Reports module) since 1968. Yet these data are not published, in part because the FBI "cannot vouch for its accuracy" (Fyfe, 2002:18). Data on police lethal force are also collected routinely by the National Center for Health Statistics in the Vital Statistics of the United States (Kania
158 FAIRNESS AND EFFECTIVENESS IN POLICING and Mackey, 1977). Sherman and Langworthy (1979) argue that estimates of justifiable homicide by police derived from vital statistics data deviate substantially from estimates derived from other data sources. They list six reasons for this finding, including vagueness in the instructions for complet- ing standardized death certificates, inconsistencies in the decisions made by coroners and medical examiners, the close relationships between local po- lice and medico-legal agencies, and logistical problems in coding, transport- ing, and entering data. Sherman and Langworthy (1979) offer mixed con- clusions about the utility of vital statistics data for explaining variation in lethal force rates across cities. On one hand, they claim that the data should not be used to compare cities because such comparisons "are likely to be dangerously misleading" (p. 559). On the other hand, they find that the data may be useful for correlational analysis of the factors influencing varia- tions in lethal force across cities (p. 559). Several studies have examined variations in lethal force over place and time. The most accurate data on lethal force that identify jurisdictions have been obtained by newspapers threatening or actually litigating under the Freedom of Information Act (Fyfe, 2002). Yet these data are not systemati- cally collected or available. Nonlethal Force The past decade has seen a pronounced increase in the level of research attention devoted to nonlethal force by police. While many individual po- lice agencies track use of force by their officers, they have only recently begun to collect and assemble these data in a format useful for comparing police organizations. Thus, findings in this area must be subject to more review and research. In 1992, Pate and Fridell (1993) conducted a national survey on police use of force. More than 1,100 agencies responded to the survey (response rate of 65.5 percent). Data from this study form the basis for much of what is known about use of force at the agency level and the kinds of relevant information that are collected by police organizations. Furthermore, data from this study continue to be used for scholarly inquiry in secondary analy- ses (e.g., Archbold and Maguire, 2002; Cao, Deng, and Barton, 2000; Cao and Huang, 2000). Pate and Fridell (1993) acknowledge that since some agencies do not require officers to report all use-of-force incidents, mea- sures computed from these data may be imprecise. Their findings show that less severe forms of force are most prevalent, with handcuffs being used 490 times per 1,000 officers in 1991, bodily force being used 272 times, and weapons being unholstered approximately 130 times. Lethal force was used an average of 0.9 times per 1,000 officers in 1991, vehicle rammings 1 time, and dog bites 6.5 times.
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 159 While the Police-Public Contact Surveys (discussed in chapters 3 and 8) have been very useful and have provided important information on a vari- ety of topics, their utility for exploring hypotheses about interagency varia- tion in the use of force is limited because they do not contain agency or jurisdiction-level identifiers. They are unlikely to become the instruments through which researchers learn the most regarding police use of force. Another major data collection initiative, funded jointly by the Bureau of Justice Statistics and the National Institute of Justice in 1996 and 1997, sought to create a voluntary, anonymous national reporting center for po- lice agencies to report their use of force data. Funding for the creation of a national data base was provided to the International Association of Chiefs of Police (IACP) in 1996 and 1997. In 1998, the IACP provided its own funding for the project, but the lack of outside support and low coverage convinced the organization to disband the project in 2001 (see Chapters 3 and 7 for further discussion of these data; Henriquez, 1999). Arrest Rates Arrest rates represent one of the most visible output measures of police agencies. They are collected consistently from most police agencies in the country through the Uniform Crime Reports (UCR). However, even this measure, which might appear on its face to be clear, has problems. Sherman (1980a, 1980b) found that the legal definition of arrest varied widely across agencies. He concluded that "differing arrest definitions make productivity comparisons between agencies impossible" (Sherman, 1980b:468). Sherman and Glick (1984) used a variety of research methods to study the quality of arrest statistics, including mail survey data from a sample of police agencies and state agencies responsible for compiling police data, brief site visits to 18 police agencies, and intensive case studies in four police agencies. Their findings suggest that state Uniform Crime Reporting agencies failed to en- sure quality control of arrest data, and that in some cases they failed to understand the rules themselves. Furthermore, some police agencies ne- glected to include citations, summonses, and citizens' arrests. Sherman's research on the quality of arrest statistics, which was com- pleted in the early 1980s, has not been replicated; researchers do not know whether the findings apply nearly two decades later. The Justice Research and Statistics Association, a national nonprofit organization linking the di- rectors of state statistical analysis centers, as well as other researchers and practitioners, serves as a professional vehicle for improving criminal justice data collected from states and localities. However, whether there has been any improvement in the quality of arrest statistics since Sherman's study is unknown.
160 FAIRNESS AND EFFECTIVENESS IN POLICING Clearances Like the arrest rate, the clearance rate is another measure of police output that is collected widely and frequently from police agencies around the nation. As discussed in Chapter 3, once reported, an offense is desig- nated as founded and it may be cleared in two ways: by arrest or by excep- tion (Federal Bureau of Investigation, 1980; Martin and Besharov, 1991). Clearance rates vary over time and place, and this variation is impor- tant to understand. For instance, Riedel and Jarvis (1999) report that clear- ance rates for homicide have been falling almost linearly over the past four decades, dropping from 92 percent in 1960 to 66 percent in 1997 (p. 279 and p. 301). Presumably, a larger proportion of murderers are now walking the street than four decades ago. Understanding the factors that influence clearance rates is more than just an intellectual game; it is a fundamental question in the study of the police. Nearly all clearance rate measures use reported crime as the denomina- tor (Cordner, 1989). Some use arrests in the numerator (Decker, 1981), while others use arrests and exceptions (Cordner, 1989). Clearance rates are routinely used as a proxy for investigative effectiveness or success, yet clearance rates are beset with measurement problems (Alpert and Moore, 1993; Riedel and Jarvis, 1999). For instance, Cordner (1989:146) argues that both the numerator and denominator used in computing the clearance rate are "susceptible to manipulation and measurement error." Similarly, in their report on the future of the Uniform Crime Reporting program, Poggio and his colleagues (1985) list a number of problems with clearance rates that reduce their utility for measuring police performance. Liska, Chamlin, and Reed (1985) examined variation in "certainty of arrest" or "certainty of punishment," which they operationalized as the ratio of arrests to re- ported crimes. While not strictly a clearance rate (as operationalized by the FBI Uniform Crime Reports), it is conceptually similar and avoids the po- tential for "bureaucratic manipulation" by the police (p. 123). Despite these problems with the measurement of clearance rates, they are reported rou- tinely by police departments and used routinely by researchers. Citations There are no national data on citations issued by police agencies. This situation may begin to change as police organizations continue to imple- ment data collection systems related to racial profiling. Because there are no data sets, there is little knowledge about departmental variations in the issuance of citations to citizens, and consequently few empirically supported explanations for these differences. Of the estimated 19.3 million drivers who were pulled over by police at least one time in 1999, about 54 percent received a traffic citation, about 26
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 161 percent received a warning, and only about 3 percent were arrested. Given the volume of citations issued by the police relative to other outcomes they produce, research and theory on what the police do should not neglect this fundamental task. Very little is known about interagency variations in the issuance of citations. Traffic tickets are not the only kind of citation used by police agencies. Many jurisdictions now rely on citations in lieu of arrest for certain misde- meanors. For instance, many states authorize the use of citations for posses- sion of small amounts of marijuana (Feeney, 1982:38). There are at least three general categories of citations (Feeney, 1982; Welsh, 1993). The names used to describe them vary widely across jurisdictions. The first, often re- ferred to as a field citation, is like a traffic ticket and is issued in lieu of a custodial arrest. The second, known as a station citation, is issued after a custodial arrest has been made and is used in lieu of detention. The third, known as a jail citation, is used by sheriffs' deputies or other jailers to release detainees. The field citation is the only one of interest in this report, since it represents an alternative measure of police output that does not involve a custodial arrest. Since the police issue field citations with such frequency, understanding the factors responsible for interagency variation in their use seems to be an important component of explaining what police organizations do. Civilian Complaints Citizen complaints arise when a citizen takes action in response to an encounter with an officer that generates a record of the grievance. These data have been used as indirect indicators of the behavior of the organiza- tion, as seen through the eyes of aggrieved parties. At first glance, the na- ture and pattern of citizen complaints might appear to be a useful indicator of police misbehavior during interactions with citizens. After all, police of- ficers who behave in a civil, professional manner are likely to generate fewer complaints than officers who behave in a rude, abrasive, unprofessional manner. However, aggregated data on citizen complaints have many shortcom- ings. Chief among these is that the number of complaints may say more about a department's openness in seeking out, welcoming, and investigat- ing feedback from the community than it does about police behavior on the streets (Adams, 1999; Walker, 1998, 2001). In addition, Walker (2001) provides numerous examples of police agencies failing to uphold their own complaint intake policies by thwarting citizens' efforts to file complaints. There is also doubtless substantial variation in the willingness of citizens to step forward and press complaints, linked to neighborhood, cultural and historical factors. If they reach the legal system by other channels, the will-
162 FAIRNESS AND EFFECTIVENESS IN POLICING ingness of municipalities to negotiate informal settlements of formal suits or complaints affects which ones become known to the public. These factors cloud the interpretation of complaints and other com- plaint-related measures of police behavior, including the rate at which civil suits are filed against departments and civil rights complaints are made against them. As Adams (1999:10) notes, "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." National studies of civil suits (Archbold and Maguire, 2002; Pate and Fridell, 1993) have not found them a persua- sive measure of police behavior. In two major surveys (Pate and Fridell, 1993; Worrall, 2001), complaint questions generated lower response rates than other items, suggesting that the data are not routinely collected and tallied by some agencies. Furthermore, even when agencies do provide nu- merical summaries, there are serious doubts among the academic commu- nity about the reliability and validity of the measures. Given the fundamen- tal importance of police-citizen relationships, however, the topic is important enough to warrant considerable effort in building a reliable body of knowledge. One potentially fruitful avenue to explore is the use of citizen surveys that ask respondents about recent contacts with police. Because these surveys are independent of the department's complaint process, they can provide a reliable measure of citizens' grievances whether or not they were filed or processed. NEGLECTED DIMENSIONS OF POLICE BEHAVIOR Despite their many limitations, the measures reviewed thus far have provided researchers and evaluators with data to assess how police exercise their formal enforcement powers. However, these measures are clearly in- sufficient for tracking police organizations' progress in achieving new no- tions of the police mission that became popular in the late 20th century and are likely to continue well into the 21st century. Indeed, research over the last 40 years clearly shows that most police resources are expended on ac- tivities that do not result in formal enforcement, and that citizens care a great deal about when, where, and how these informal activities occur (Mastrofski, 1983; Whitaker et al., 1980) Research also suggests that they may have a significant influence on the crime control effectiveness of the police (see Chapter 7). The overwhelming emphasis on recording only cer- tain formal enforcement activities in policing is analogous to a hospital's maintaining records only on the surgery its physicians perform, leaving un- recorded the many other important treatments administered by its staff. As Chapter 3 indicates, police have been shown to rely on formal en- forcement powers relatively rarely, and, just as importantly, new strategies such as community and problem-oriented policing call for police to engage
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 163 in a much broader range of activities to accomplish desired outcomes. And even more recently, concerns about a more effective police response to the risks of terrorism have illuminated just how little systematic information there is about police policies and practices concerning the gathering, pro- cessing, and dissemination of information (Ericson and Haggerty, 1997). Policy makers who want to measure police agencies' progress toward implementing practices consistent with these new directions need a new set of institutionalized measures that simply do not exist in any systematic form in agencies across the United States. At a minimum, the following types of indicators are needed to capture the wide range of approaches to police work with which local agencies are experimenting: applications of police authority other than formal enforcement of the criminal law, forms of po- lice assistance, indicators of the quality of the policing process that citizens experience, measures of mobilizing and working with the community, mea- sures of problem solving, and indicators of police as information brokers. Each of these categories is briefly discussed below. Applications of Police Authority That Do Not Invoke the Criminal Process Research shows that police relatively rarely resort to their formal au- thority to arrest or cite someone in dealing with the myriad situations they handle. This is not only because the police, more often than not, choose leniency when the evidence would support formal enforcement, but also because they must often deal with situations in which formal enforcement action is not legally warranted but they feel compelled to do something to correct a situation or prevent matters from deteriorating. Thus, they seek citizen compliance and order maintenance by such methods as simply mak- ing their presence or interest known to potential troublemakers, stopping and questioning them, persuading, advising, commanding, or threatening them, or referring problems to other agencies. Broken-windows policing relies heavily on these informal methods of enforcement, as well as arrests and citations for minor offenses (Wilson and Kelling, 1982; Kelling and Sousa, 2001), yet police departments do not routinely record most of these actions, making it difficult to determine how and how thoroughly this ap- proach has been implemented. Many departments do maintain records on certain field interrogations, which are stops of suspects that do not result in an arrest, and of traffic stops that do not result in a citation. However, the diligence with which these are recorded varies greatly and the comparabil- ity of record-keeping across agencies is unknown. The availability of useful cross-agency data may be changing, however, due to standardization im- posed by vendors of records management systems in multiple agencies. While the Uniform Crime Reports have done a relatively good job of
164 FAIRNESS AND EFFECTIVENESS IN POLICING indicating the formal aspects of police performance over time and across jurisdictions, there is no comparable body of data on the far more fre- quently employed informal methods by which police authority is exerted. The nation's accounting system for police authority leaves exertions of that authority unrecorded. Given the challenges of making the UCRs a reliable data source, the difficulties of developing a more comprehensive system for monitoring the informal exercise of police authority are truly daunting. Nonetheless, it might be possible to develop a uniform reporting system for certain aspects of police agencies' documentation of calls for service and field interrogations, one that would call for agencies to use a standardized system to note which informal, as well as formal, uses of authority were applied in each situation recorded by the agency. Such a system would en- able police leaders and policy makers to develop a far more comprehensive picture of policing as most people in their communities are actually experi- encing it. The committee recommends that support be given to the Bureau of Justice Statistics to develop and pilot test in a variety of police depart- ments a system to document informal applications of police authority. Police Assistance to Citizens Police in the United States have long performed a wide range of ser- vices, most of which receive little attention in the analysis of organizational performance. During previous eras of reform, these activities were discred- ited as "not real police work," which detracted from the ability of police to fight crime (Fogelson, 1977; Walker, 1977). Yet currently popular commu- nity policing and problem-oriented policing reforms encourage police to engage in an even broader range of services as a means of more effective crime prevention, enhancing the quality of community life, and securing greater community support for and participation in police efforts. Police are frequently called to exert their authority to protect one party from another (Mastrofski et al., 1996; Snipes, 2001) which is one kind of service, but they are also often asked to do things that could be performed by a private service provider, things that require no special police authority but are provided because of the availability of officers to respond in a timely fashion (Wilson, 1968). These range from pulling stranded cats out of trees, to providing crime prevention advice to neighborhood groups, to running midnight basketball leagues designed to keep youths from engaging in ille- gal and disruptive activities. Some departments engage in more of this activ- ity than do others, and the existence of this natural variation creates oppor- tunities to judge the effects of these efforts on a wide range of outcomes, such as crime, fear of crime, and police legitimacy. For example, the public in one community might find it useful to com- pare its department's level of nonenforcement service delivery to that of
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 165 other similar communities, just as consumers of privately delivered services find it useful to have such information on service providers. Such data, if available on a large number of police departments, would allow researchers to determine the contribution of nonenforcement service delivery to crime control and community satisfaction with the police. The committee recom- mends development of measures that better document at the jurisdiction level the nature and extent of nonenforcement services delivered by police. This program of development should consider the variety of current mea- sures available in U.S. police agencies, pilot test a system at several sites, and then propose a large, multiagency data collection system. Policing Processes There is value in measuring various aspects of the processes of policing, not only because the quality of those processes appears to have consequences for valued outcomes (such as reducing law violations), but also because the public highly values certain aspects of police behavior in and of themselves (such as being reliable, attentive, responsive, competent, respectful, and fair) (Mastrofski, 1999). As Chapter 8 shows, whether and how police perform these processes is at least as important to citizens as police effectiveness in crime control. We have noted some of the limitations of using complaints against the police and lawsuits to measure the quality of these processes. Another limi- tation is that complaints and lawsuits at best measure only bad perfor- mance and leave unattended acceptable or good performance. Chapter 8 recommends a way to generate useful information about these processes by conducting surveys of police "clients." Here we note that the character of those processes appears to be so important to securing police legitimacy and crime reduction that developing a national jurisdiction-level system of data collection on policing processes is essential to a comprehensive charac- terization of what police organizations do. Mobilizing and Working with the Community Community policing reformers have made it widely expected that po- lice in the United States work closely with community groups to prevent crime and enhance the quality of life. As discussed elsewhere in this chapter, a number of survey projects have attempted to measure this kind of activity, one of which has been institutionalized in the Law Enforcement Manage- ment and Administrative Statistics (LEMAS) surveys periodically conducted in the nation's larger police agencies. The problem with existing survey data is that they offer weak measures of the scope and intensity with which such activities are undertaken, noting usually only whether an agency is cur-
166 FAIRNESS AND EFFECTIVENESS IN POLICING rently engaging in such activities, but not how much police and community effort is exerted (Maguire and Mastrofski, 2000). For example, two depart- ments may report that they work with community groups to prevent crime, but there is a tremendous difference between one that meets frequently with many community groups and one that meets occasionally with a few. The current inability to measure the dosage of such efforts greatly limits the ability of researchers to distinguish substantial differences among agencies in how and how much they engage in these activities. The committee rec- ommends that the Bureau of Justice Statistics develop measures that pro- vide a more accurate indication of the extent to which community liaison and mobilization activities are engaged by individual police agencies. Problem Solving The desirability of a problem-solving approach to policing has taken hold in many police agencies in the United States (Scott, 2000). A confer- ence devoted to problem solving, sponsored annually by the Police Execu- tive Research Forum, is attended by members of hundreds of departments from around the nation. Surveys of police agencies suggest that the popu- larity of this approach is widespread (Maguire and Mastrofski, 2000), yet it too suffers from the same measurement difficulties as does community liai- son and mobilization. Some research suggests that great variation exists in the kinds of problems targeted for these methods, as well as in how prob- lem solving is actually conducted. Some have distinguished weak problem- solving efforts from stronger ones. The extent of problem analysis, the search for effective solutions, evaluation, and follow-up are important as- pects of the dosage issue for problem solving (Eck and Spelman, 1987). The committee recommends that the Bureau of Justice Statistics develop mea- sures that provide a more accurate indication of what problems and popu- lations are targeted for problem-solving efforts, as well as the intensity and quality of those efforts. Because problemsolving can occur at many levels and in many ways in a given police organization, a feasible measurement method will undoubtedly require sampling problem-solving efforts over a given time period in each surveyed department. Police as Information Brokers Some researchers have argued that by the late 20th century, the core police capacity had shifted from being purely that of coercion to that of information gathering, processing, analysis, and dissemination (Ericson and Haggerty, 1997). Certainly community and problem-oriented policing call for increased levels of this sort of activity, but the current domestic war on terrorism has heightened sensitivity to this responsibility for local law en-
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 167 forcement agencies. A later section of this chapter deals with the question of how the war on terrorism is affecting police structure and practices. Here we note the need to gather systematic data on what those structures and practices are. As with any broker, learning certain things about the information gath- ering, processing, and dissemination will be an important foundation to assess how well police agencies are performing this function, specifically: · nature of the information collected, · how the information is gathered and what other organizations con- tribute to it, · how the information is checked for accuracy, processed, archived, and retrieved within the police agency, · how and to whom information is disseminated inside and outside the agency, and · how information is used and how useful it is to the users. Police deal in far too much information to gather data on the above ele- ments for all types of information. Strategically focusing inquiry on police information brokering is necessary, and the committee recommends two areas that are especially timely: (a) information on criminal events and (b) intelligence on terrorist suspects and potential terrorist activities. In the first case, police departments around the country are undergoing major changes in the technology of gathering, storing, mapping, and ana- lyzing criminal events (Anselin et al., 2000). Advocates of such programs as CompStat claim that it is revolutionizing the management of police crime control and service delivery (Bratton, 1998; Silverman, 1999; McDonald, 2001), yet there is relatively little rigorous study of how, if at all, it is chang- ing police organizations, and that which is available has produced intrigu- ing findings. For example, a national survey of police agencies conducted in 1999 found that agencies that had implemented CompStat tended to place greater emphasis on empowering, rather than challenging, the traditional hierarchical paramilitary structure of municipal police organizations in the United States (Weisburd et al., 2003). The committee recommends that the National Institute of Justice support a program of rigorous evaluation of new crime information technologies in local police agencies. This program should go well beyond describing variation in technologies and whether they "work" in a technical sense; it should track the impact of changes on organizational structures and practices over time in selected departments and compare the impact of these technologies in a variety of departments. Such information will help police agencies establish which technologies, and which ways of organizing to cope with those technologies, will work best for them.
168 FAIRNESS AND EFFECTIVENESS IN POLICING In studying local police agencies as information brokers on terrorism, data collection issues are complicated by the recentness of local agencies attending to this function, the high level of interagency coordination en- tailed, and the need to maintain the necessary security for these information systems. Such information networks will obviously extend to federal agen- cies and departments, such as the FBI and the Department of Homeland Security. Some of the data elements bulleted above will involve working with classified information, so special arrangements will need to be made to obtain researcher access to data, and the distribution of results will be chal- lenging. Nonetheless, the importance of the terrorism intelligence function is too important to leave unexamined or reported only in classified docu- ments not available to the wide range of policy makers with responsibilities for funding, planning, and developing terrorism intelligence systems involv- ing local police agencies. The U.S. Department of Justice has already funded a small pilot project, run by the Police Executive Research Forum (PERF), exploring issues of federal-local coordination in dealing with information on terrorism. The committee strongly encourages using the results of this effort to develop a long-term national program for tracking and evaluating the performance of local police departments' efforts in gathering and han- dling intelligence on terrorism. INFLUENCE OF POLICE ORGANIZATION The ostensible function of an organization is to structure the activities of its members to accomplish goals. Policy makers and administrators at- tempt to manipulate various features of police organizations to accomplish such goals as crime control, legal compliance, and community contented- ness. In this section we consider several features of police organizations thought to influence police practices that are presumed relevant to the ac- complishment of these goals. We first consider the size of a police agency. Then we turn to structural features that have been frequent objects of policy manipulation: the degree to which decision making is centralized or decen- tralized, the degree of complexity (defined in various ways), and the extent to which policies are formalized. Finally, we consider CompStat, a cur- rently popular omnibus reform package that combines several of these struc- tural elements. Police Agency Size Proposition 1: The size of a police organization is a product of policy choices. Some claim that larger departments tend to yield more and better police activity, while others claim that large police departments
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 169 enjoy no particular advantage and that they may often suffer disad- vantages. The committee finds that the effects of police agency size are mixed and contingent on too many factors to conclude that police agencies of any size--small, medium, or large--are necessarily to be preferred. As suggested in Chapter 3, police service delivery in the United States is fragmented among thousands of police agencies, most of which are very small, the median local police department had only 7 sworn officers in 2000 (Bureau of Justice Statistics, 2002). There are three major reasons for this: the tradition of local government service delivery in a federal system, the proliferation of small suburbs that limited the growth of large center cities, and the general lack of enthusiasm among police leaders for consoli- dation of departments (Reiss, 1992:64-65). Since the 1930s, "good govern- ment" reformers have advocated consolidating small departments in metro- politan areas to facilitate police professionalism, improve productivity, and enhance service quality (Walker, 1977:141-146). Several blueribbon com- missions in the 1960s and 1970s endorsed this approach (U.S. Advisory Commission on Intergovernmental Relations, 1963; President's Crime Com- mission, 1967; Committee for Economic Development, 1972:31; National Advisory Commission, 1973), and progressive police leaders frequently as- serted its benefits for police efficiency and effectiveness (Altshuler, 1970:38). Consolidation, when it has occurred, has occasionally taken the form of integrating numerous agencies into a single county police agency; more of- ten it occurs when a small jurisdiction contracts with a larger one for police service, or when it disbands its police force and relies on the county sheriff or state police to provide services (Mastrofski, 1989:10-11; Reiss, 1992:66). Running counter to the consolidation movement was a call from the "community control" reform movement to break up large police depart- ments. It surfaced briefly in the late 1960s and early 1970s. Many police progressives of the time claimed that breaking up large police departments would threaten impartial law enforcement and return to the corruption and inefficiencies of machine-era policing (Mastrofski, 1981:44-45). Large de- partments could afford to implement the latest technological innovations, support a large supervisory hierarchy for quality control, hire and retain highly qualified people, and provide for more and better training (Wilson, 1968:284-299). The validity of these claims was questioned by some re- formers and researchers, who argued that "bigger was not better," and that in fact, smaller departments might well serve their communities better (Ostrom, 1973, 1976; Ostrom et al., 1973). Smaller communities would rely less on organizational hierarchy and more on elected officials and citi- zens to oversee police actions, thereby enhancing productivity and efficiency. Furthermore, some argued that any gains to be had in the greater profes-
170 FAIRNESS AND EFFECTIVENESS IN POLICING sionalism of larger police agencies were undermined by the costs imposed by heightened bureaucratization: greater alienation from the citizenry and less responsiveness to their needs and preferences, trends acknowledged by bigger-is-better advocates and that concerned them too (Schmandt, 1972; Murphy and Plate, 1977; Peterson and Pogrebin, 1977). Despite the popularity of the consolidation movement among blue rib- bon commissions and some big-city chiefs, it simply has not transformed the structure of the American police industry, which remains overwhelm- ingly populated by small agencies, while a major portion of the public is served by large agencies (Reaves and Goldberg, 1998). Of course, police agency consolidations do occur on a small scale, but they are rarely of spectacular proportion, such as merging many small and medium-sized agencies into a single county organization (Mastrofski, 1989:10). Conse- quently, most of what researchers know of the consequence of police agency size for police practice comes from comparisons of a cross-section of large and small departments at a given point in time. Much of this research fo- cuses on outcomes, such as crime rates, citizen satisfaction, and cost-effi- ciency. However, some research has examined the implications of agency size for actual practice, and it is this literature that the committee reviewed. Two major studies examined the consequences of police agency size for the ability of departments to mobilize police, that is, to put patrol officers on the street. Elinor Ostrom and colleagues examined patrol deployment of over 1,000 local departments providing patrol services in 1974-1975 in 80 metropolitan areas of the United States. More than half of the sample had no more than 10 full-time sworn officers, 45 had more than 150 (and only 10 of these employed more than 500 sworn). For municipal police (as well as other local police agencies) they found that the smaller the agency, the more officers per resident were on patrol. For example, they found that departments with only 5-10 sworn officers typically placed 4.2 per 10,000 residents on patrol at 10:00 p.m., while departments with more than 150 sworn officers managed only 2.3 per 10,000 (Ostrom et al., 1978:85-91). They found that smaller departments accomplished this by allocating a much higher percentage of their sworn force to patrol operations and much less to administration, auxiliary services, and other direct services. What that meant in practical terms was that the chief and supervisors often per- formed rank-and-file patrol duties as well as supervisory and administrative functions. In a follow-up, Langworthy and Hindelang (1983) compared the Ostrom et al. results to those of another data set comprised of a national sample of 69 larger departments (more than 134 sworn officers) surveyed in 1977. Langworthy found that the sample of larger departments showed a median level of patrol mobilization consistent with that found by Ostrom and colleagues. Additionally, he noted that, like the Ostrom data, his sample displayed considerable variability in both agency size and patrol mobiliza-
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 171 tion. Examining the relationship of agency size to patrol mobilization among these larger departments, he found a weak direct relationship between agency size and number of officers on patrol per citizen. Thus, while the Ostrom study indicated that the consolidation of small agencies would yield a lower level of patrol mobilization, Langworthy's analysis suggested that breaking up the largest agencies into smaller ones would not necessarily yield higher levels of patrol mobilization. Although citizens are certainly concerned with the level of police pres- ence on the street, they are equally concerned with what the police do while on the street. The entertainment media undoubtedly encourage the public to perceive small and large departments in different ways by stereotyping them. For example, Dragnet and Adam 12 depicted big-city police as legal- istic, technically sophisticated, and enforcement-oriented, while Andy of Mayberry showed small-town law enforcement as service-oriented, human- istic, and inclined to seek less coercive, informal solutions to problems. To what extent does research support these stereotypes? Some of the research suggests that the relationship between police de- partment size and actual practice is weak or nonexistent, but a fair amount of the available research does in fact tend to support the popular stereo- types (see Mastrofski, 1981:Ch. 3 for a review). On one hand, several stud- ies have shown that, taking other influential factors into account, large departments tend to have higher arrest rates (per officer or per capita), to have officers who are more aggressive and less selective in using their for- mal law enforcement powers, and to resort to force more frequently (Brown, 1981; Mastrofski, 1981:Ch. 6). On the other hand, smaller agency size is associated with officers who are more familiar with individual citizens and neighborhood organizations, provide more assistance to neighborhoods, and more follow-up to citizens' crime reports (Parks, 1979, 1980; Whitaker, 1979). However, these studies, which compared police practice in large center-city neighborhoods with that in comparable neighborhoods in nearby small cities, also found some significant exceptions. For example, citizens were more likely to report incidents of police mistreatment in the smaller departments of two of the studies, but in general the differences found were not large. Studies that looked at middle-sized departments sometimes showed a curvilinear relationship. The mid-sized agencies outperformed both the large and small ones when the bases of comparison were citizens' perceptions and evaluations of police service in general, not associated with a particular event. Because the specific pattern of results varied with the study, one of the evaluators concluded that while the evidence did not al- ways favor small departments, sometimes varying with the race or class makeup of the neighborhood, "Contrary to what most contemporary crit- ics of American policing have argued, we have not found performance ad-
172 FAIRNESS AND EFFECTIVENESS IN POLICING vantages for residential neighborhoods that received police services from large departments" (Parks, 1980:4-60). There are some interesting exceptions to the tendencies just discussed. The impact of agency size on enforcement levels, as measured by arrest rates, has been shown to vary with the type of offense. For example, one study found that small middle-class suburbs showed higher arrest rates for breaches of the peace than did the nearby large county police force, but the reverse was true when agencies were compared on arrest rates for theft (Wilson, 1968:211-215). The researcher hypothesized that smaller agencies were more responsive to the priorities of the citizens in those communities who would find breaches of the peace quite troubling. Furthermore, the smaller departments had fewer alternatives to arrest. In contrast, larger de- partments would be more buffered from the preferences of the neighbor- hoods and also had more nonarrest alternatives available. However, re- garding thefts, the issue is less one of community outrage than of rectifying the victims' losses. Here larger, specialized departments are presumed to have the advantage in solving crimes and catching thieves. Another exception running counter to the popular images of small ver- sus large departments is found in research on drunk-driving enforcement. Small departments have been shown to show a striking inclination to exer- cise more initiative and less discretion in enforcing drunk-driving laws than found in large departments (Mastrofski et al., 1987; Mastrofski and Ritti, 1990, 1992). A study of 735 municipal Pennsylvania agencies found that departments with 1-5 full-time sworn officers had a per-officer DUI arrest rate more than three times that of departments with more than 100 sworn officers (Mastrofski and Ritti, 1996). In general, drunk driving is at least as prevalent in jurisdictions served by large departments as those served by small ones. Researchers noted that small departments generally have a less intensive workload than larger ones, thereby giving officers more available time to initiate time-consuming DUI stops, and smaller communities may tend to see drunk-driving enforcement as a higher priority (Mastrofski et al., 1987). But they attributed much of the difference between small and large agencies to differing sets of opportunities available for these depart- ments to demonstrate their law enforcement prowess. Large departments have many ways of demonstrating their law enforcement proficiency by developing specialist units to deal with serious crimes. Small departments had fewer serious crimes and much less resource availability to devote to the kinds of specialist structures that signify law enforcement competence (see Chapter 8 for a discussion of this approach to enhancing organiza- tional legitimacy). Finally, some studies of clearance rates provide a picture of American policing at odds with the big city and small town stereotypes. A study of national UCR data found mixed results when examining the effect of agency
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 173 size on clearance rates for various offense categories (Cordner, 1989). In a national sample, agency size had a negative effect on the clearance rate, but in a supplemental sample of Maryland agencies, size had no effect. In the Maryland sample, agencies located in the Baltimore-Washington metropoli- tan area also had lower clearance rates, suggesting the likelihood of an urban-rural effect. The researcher concluded that the environment may have had a stronger influence on clearance rates than the organizational charac- teristics, which suggests that there are limits to what can be accomplished by manipulating agency size. A study of child sexual abuse cases found that large agencies had much lower clearance rates, although the researcher noted the susceptibility of these rates to earlier decisions to found cases or discard weak ones (Maguire, 2002c). It is hazardous to offer specific policy recommendations from this rela- tively modest body of research on the impact of agency size on police prac- tice, but several general lessons can be drawn. First, policy makers should not assume that consolidation of small police departments will yield greater productivity or changes in the quality of police practice. Many studies sug- gest that smaller agencies serve their jurisdictions as well or better than do larger agencies serving similar areas. Furthermore, one of the costs of achiev- ing economies of scale by consolidation may be reduced responsiveness to some of the jurisdictions served. Indeed, a case study of the feasibility of small-department consolidation revealed that not one of five different ser- vice arrangement options served each jurisdiction equally well (Mastrofski, 1989). Similarly, breaking up large agencies into smaller ones (something that occurs very rarely), also carries no guarantees of improved perfor- mance. Finally, it appears that the consequences of agency size for enforce- ment productivity may depend on the type of offense under consideration. Independent of the extent to which the problem exists, large agencies ap- pear more active in enforcing some kinds of laws than are small agencies, but the reverse is also true. Smaller agencies pay more attention to some offenses than do large agencies. This suggests the importance of taking into account what priorities for enforcement are communicated from a depart- ment's environment, and what mechanisms exist to channel pressures into or away from police agency policies and practice. Decentralization of Decision-Making Authority Proposition 2: Police reformers over the last three decades have sup- ported decentralization of decision-making authority in the police agency as a powerful tool to accommodate to varying needs and priori- ties in the agency's jurisdiction. This began as a movement toward ad- ministrative decentralization in the late 1960s that has continued under the banner of community policing. The committee found very little re-
174 FAIRNESS AND EFFECTIVENESS IN POLICING search on the topic and is therefore unable to draw conclusions about the extent to which decentralized organizational structures do produce practices that meet the needs and demands that vary within the jurisdic- tion served. Centralization is the extent to which the decisions in an organization are concentrated; decentralization is the extent to which they are distrib- uted more evenly throughout the organization. Decentralization of police departments has had its advocates since the late 1960s (Mastrofski, 1981), and it has remained a staple of police reform in the 1990s under both com- munity policing (Skogan and Hartnett, 1997) and CompStat (Bratton, 1998). Advocates of decentralization have argued for moving more deci- sion-making authority down the chain of command--from headquarters to the precinct station, and from the precinct station to the first-line supervi- sors and officers on the street. The justifications for decentralization have been several, but two stand out. One is that a decentralized structure is better able to deal with varia- tions in the department's work environment. For example, where neighbor- hoods vary in culture, wealth, and demographic profile, they are also likely to vary in the sorts of crime and disorder problems they experience, and their service needs may differ too. Advocates of decentralization argue that those closest to the neighborhood are in a position to know more about its needs and the preferences of its residents than those further up the chain of command. It therefore makes sense to give lower ranking personnel greater authority to decide what problems should take priority and how to handle them. A second, related argument for decentralization is that true profes- sionalism requires the delegation of authority to the professionals who must cope with the problem. Effective human services, especially those dealing with the diversity of problems that police must handle, require well-edu- cated and well-trained professionals who must be given a substantial degree of leeway to exercise their judgment. Uniform rules and procedures issuing from headquarters will restrict the exercise of this professional discretion and result in reduced effectiveness. How should decentralization affect police practices? Presumably, de- centralized structures will promote greater variation in how police do their work, assuming that there is substantial variation in their work environ- ment--say neighborhoods. But reformers expect more than variation in practice; they expect police practices to vary appropriately with the needs and demands of the environment. Thus, a neighborhood with high levels of street disorder would receive one kind of policing, while a neighborhood with low levels of crime and disorder would receive another. Of course, the specification of conditions might be more complex, also factoring in the neighborhood's capacity and willingness to act collectively in concert with
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 175 the police to handle problems. Presumably, research on what is most effec- tive under what circumstances would prescribe the appropriate approach, but the challenge here is to determine whether a decentralized structure is more likely to facilitate the selection and implementation of the best ap- proach than a more centralized structure. Research of the sort just described is generally not available, due largely to problems with measuring both decentralization and its effects on prac- tice. It is difficult to measure the degree of centralization or decentralization because they are aggregate organizational properties. In other words, mea- suring the level of centralization in an organization would mean determin- ing at what level a particular decision or set of decisions is made. This might be done by observation, but because that is very time-consuming, it is usu- ally accomplished by asking one or a few informants in the organization where the locus of decision making is. The problem with this is that infor- mants may have a biased or inaccurate view. Asking more people in the organization how much they participate in the decisions in question would yield a more robust, and undoubtedly more complex, picture of decentrali- zation, but that consumes a lot of resources, which means that research of this sort is usually limited to case studies of a single organization. One example of this is a long-running study of a community policing in Chicago, which has provided some evidence that greater decentralization has yielded more customizing of police strategies and tactics to the needs and preferences of neighborhood residents (Skogan and Hartnett, 1997). Cross-sectional research of large numbers of police organizations has relied heavily on using a single informant (the survey respondent designated to speak for the department) to characterize the degree of decentralization in the department by indicating the degree to which employees at various lev- els in the hierarchy are able to make important decisions. Computing com- posite centralization scores using this method, recent research has found that police organizations have, in fact, become more decentralized during the 1990s (Maguire et al., 2003). However, these researchers have not taken the next step, which is to correlate the degree of decentralization with mea- sures of police practice. The principal obstacle to measuring police practices relevant to decen- tralization is determining the criteria that establish what is the best police activity relevant to a given targeted population, such as a neighborhood. One way to deal with this might be to establish how much actual police practice (documented in agency records, officer surveys, or field observa- tions) deviates from the preferences of neighborhood residents (determined from opinion surveys). If decentralization advocates are correct, then greater decentralization should yield patterns of police behavior in greater congru- ence with the neighborhood's expectations than more centralized decision making.
176 FAIRNESS AND EFFECTIVENESS IN POLICING Organizational Complexity Organizations are more or less complex in terms of how they structure the division of labor among members (Hall, 1991:50ff). The police organi- zation can be more or less occupationally complex by adjusting the number of job specialists (e.g., detectives, crime analysts, juvenile officers, commu- nity policing officers). It can become more or less functionally complex by adjusting the number of divisions or units. It can be more or less vertically complex by manipulating the degree of hierarchy (e.g., levels of supervision and management). And it can increase spatial complexity by adding more geographic units or shifting job responsibilities from occupational specialty to spatial specialty (e.g., by assigning officers to permanent beats in which they "specialize" their activities). Job Specialization Proposition 3: Police organizations create job specialties and specialist units for a variety of reasons, including exerting greater control over the activities of personnel. The degree of job specialization appears to be increasing among American police agencies. However, there have been few studies of the effects of police job specialization of either form, and the results are mixed. One of the hallmarks of the professional or advanced police organiza- tion has traditionally been increased job specialization. This occurs when certain police tasks become the responsibility of individuals who specialize in performing that task. This occupational specialization is assumed to ben- efit the agency by providing for increased training and experience in a par- ticular type of work, for increased sharing of knowledge and communica- tion among experts in a particular line of work, and for focusing the mission and monitoring the performance of those specialists, whether they operate independently or in a specialist unit. When organizations are divided into separate units that perform different functions (e.g., patrol, detectives, juve- nile, crime analysis), they are said to be functionally differentiated. The general trend in policing for the 20th century has been to increase the de- gree of occupational and functional specialization in police departments, with civilians taking over many functions previously performed by sworn officers, and increasingly dividing and subdividing responsibilities among different organizational units, often at the cost of reducing the proportion of the organization devoted to the most generalist work, patrol (Reiss, 1992; Maguire, 2002b). Nonetheless, with the emergence of team policing as a reform ideal, a countermovement toward despecialization in policing took root in the late 1960s. For example, advocates found it desirable to return some of the follow-up investigative responsibilities that had devolved to
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 177 criminal investigations specialists to the patrol officer. Although team po- licing quickly faded in the 1970s as a passing fad, the benefits of despecializ- ing police became part of the mantra of community policing reform that emerged in the 1980s (Mastrofski, 1998:163). The handful of studies on community policing specialization suggest that specialists do spend more time on community policing activities than generalist officers, and they may be less likely to engage problem citizens, but they do not appear to show appreciable differences in how they deal with problem citizens when they do engage them in face-to-face encounters. Given the dearth of research in this area, the committee recommends analy- ses that estimate the relative merits of assigning community policing work to specialists, generalists, or both. Most studies measure functional differentiation by counting the num- ber of special units or functional divisions in an organization. Reformers have called for police organizations to despecialize, placing more authority in the hands of generalist officers, and encouraging them to provide a broader range of services. Although there have been numerous studies on the causes and correlates of functional differentiation (Langworthy, 1986; Maguire, 2002a), there is limited research on its effects on police practice. One reason for this is that researchers often include specialization as one indicator in a composite score measuring overall bureaucratization. Many scholars who study complex organizations, both inside and outside polic- ing, have argued that bureaucratization is not a viable concept--that study- ing its individual components, like functional differentiation, is far more meaningful. Two recent studies have examined changes in the functional differen- tiation of police organizations. The first (Maguire, 1997) found conclusively that during the time when community policing was growing in popularity, functional differentiation was increasing. The second found conclusively that it is not decreasing, but was unable to determine due to data problems whether it is increasing (Maguire et al., 2003). Either way, changes in func- tional differentiation are not following the reform prescriptions. This is probably because specialization appeals powerfully to police and other policy makers as a solution to organizational problems. First, it is believed to be an effective means of concentrating expertise in an organizational unit, focusing the mission of that unit on the proper exercise of that exper- tise, and rewarding its members accordingly. Thus, police leaders believe it to be an effective mechanism to get the organization to respond to its direc- tives, especially when the leadership wishes to launch a new and unfamiliar approach to policing. A second powerful attraction of functional specializa- tion is that it can work to satisfy demands for doing something about a problem while isolating the effects of organizational change to the bound- aries of a well-defined unit, leaving the remainder of the organization to
178 FAIRNESS AND EFFECTIVENESS IN POLICING operate with minimal disruption to organization routines (see Chapter 8 for a detailed discussion of the effects of specialization on police legitimacy). Both of these reasons have probably caused police departments to sus- tain or even increase the level of functional specialization, even in a time when community policing reform rhetoric calls for despecialization. What this has meant in practical terms is that police leaders have been deciding whether to implement community policing through special squads or to give all officers equal responsibility. Perhaps the most popular approach is a hybrid, which creates one or more specialist units responsible for con- ducting community policing activities exclusively (or in some cases, con- ducting traditional police patrol duties exclusively), while still training and encouraging generalist patrol officers to engage in community policing as well (Skogan and Hartnett, 1997). Ideally, researchers would assess the overall impact of generalist, specialist, and hybrid approaches by measuring the effects in the aggregate on a wide range of police activities, including those falling into both the traditional and community policing categories. This has not been done, but a handful of studies have compared the prac- tices of community policing specialists to generalists in the patrol division. This research, which may shed some light on the consequences of special- ization, is summarized below. One study, focusing on data from the Project on Policing Neighbor- hoods (POPN) for Indianapolis and St. Petersburg, compared community policing specialists (CPOs) and patrol generalists (beat officers) in two de- partments that used hybrid systems of specialization. In both cities, CPOs were freed from calls-for-service responsibilities so that they could devote more time to community policing activities. Patrol generalists, who were responsible for responding to calls for service, were expected to have their actions suffused with the department's community policing philosophy and to engage in community policing as time permitted. In both cities, commu- nity policing specialists spent less time dealing with the public face-to-face than did patrol generalists, and they tended to deal with problem citizens (those with elevated emotions) at a lower rate as well (Parks et al., 1999). In Indianapolis, the department where police leaders emphasized the need to deal with public disorders aggressively, CPOs were less likely to spend time on handling these problems than patrol generalists, even though the special- ists were freed from answering calls for service so that they could spend more time on these problems. In both departments, CPOs did tend to spend more time on community outreach and problem solving than did general- ists, although the difference was most pronounced in St. Petersburg, where these aspects of community policing were emphasized. Even here, however, CPOs were observed to spend on average only 17 percent of their time on problem solving, 4 percent attending community meetings, and 2 percent on casual chats with the public--leaving more than three-fourths of their
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 179 work time to other, noncommunity policing activities. A subsequent multi- variate analysis showed that when other factors were taken into account, the CPO specialists in Indianapolis spent no more time on problem solving than did beat officers, although the strong difference between specialists and generalists did remain in St. Petersburg (DeJong et al., 2001). Interest- ingly, St. Petersburg specialists were not assigned to a separate community policing unit, but rather were integrated into community policing teams, with the patrol generalists assigned to each neighborhood. Another study using the POPN data found that CPOs in these two jurisdictions were no more likely than beat officers to grant citizens' requests to control another person who was bothering them (Mastrofski et al., 2000). A comparison of CPOs and beat officers in Cincinnati also used POPN observation protocols, but in this city there was a sharper distinction drawn between the responsibilities of beat officers (traditional preventive patrol and calls-for-service response) and CPOs (mostly community policing du- ties) (Novak et al., 1999). The study employed a multivariate analysis that controlled for the effects of situational influences, officer characteristics, and neighborhood characteristics. The analysis showed that the nature of the officer's job assignment had no effect on whether the officer would resort to an informal, order maintenance solution to a crime or dispute (p. 178), and no influence on the likelihood that the officer would resort to arrest (p. 203). In the exercise of these traditional forms of police discre- tion, police observed in this study were not affected by the nature of their functional job assignment. The results of four studies in three police departments are insufficient to draw conclusions about the effects of functional specialization on police practice. When differences have been found, they have been measured in how and with whom officers spend their time, although even here, being assigned to a specialist unit is no guarantee that the officer will expend more time engaged in the activities designated for the job specialty than his or her generalist colleague. Where police discretion is measured in terms of a discrete decision about how to deal with a specific individual in a specific encounter, studies have found no appreciable difference in the exercise of discretion. This is as might be expected. Management may not wish for officers to exercise their discretion differently in disposing of particular per- sons causing problems, but managers may well expect to see differences in how officers choose to spend their time generally. Hierarchical Differentiation Proposition 4: In recent years, police reformers have regarded the many layers in the rank structure of the typical large or even middle-sized municipal police agency as responsible for organizational inefficiency
180 FAIRNESS AND EFFECTIVENESS IN POLICING and sluggishness. However, there is little information on the effects of the degree of hierarchical differentiation in police agencies, and what is available is mixed. The committee finds insufficient evidence to draw conclusions about the effects of hierarchy on police practices. As with functional specialization, police organizations during the 20th century have tended to become increasingly differentiated vertically. Many police agencies have grown in size and functional complexity, and as they have done so, policy makers have discerned a need to assign a larger share of personnel to supervise, manage, and coordinate others as their primary job responsibility (Reiss, 1992). But emerging in the late 20th century was a popular organizational development movement to reduce the degree of hi- erarchy in public and private organizations (Mickelthwait and Wooldridge, 1996), a trend that became part of the popular community policing reform movement (Mastrofski, 1998:163). Cutting out entire levels in the supervi- sory hierarchy, or "delayering," became especially popular (Greene et al., 1994; Robinette, 1989; Weatheritt, 1993). Reformers blame the police hi- erarchy for many of the ills of modern policing. It is conventional wisdom in progressive policing circles that the police rank structure is responsible for a variety of problems, ranging from poor communication and low mo- rale to sluggish and unresponsive organizations. Furthermore, the police rank structure is thought to be notoriously resistant to change. One popular article likened changing the police rank structure to "bending granite" (Guyot, 1979). As popular as this conventional wisdom about the problems of the police rank structure might be, it is backed up by virtually no re- search evidence. The committee was able to identify only a handful of studies on the effect of the police rank structure. For example, in an examination of Illi- nois police departments, Crank (1990) found that departments with taller hierarchies made more arrests. King (1998) found that taller police depart- ments had higher scores on a scale measuring "management oriented ad- ministrative innovations." Zhao (1996) found that hierarchy had no effect on the adoption of either the internal or external elements of community policing. Clearly, the research on the effects of the police hierarchy is insuf- ficient to support the conventional wisdom about its dysfunctionality. Although such research is rare in policing, King (2003) reviewed 18 studies on the effects of hierarchy in samples of nonpolice organizations. He found that rank structure impedes performance and internal group pro- cesses in small groups (of approximately five or less), but that in larger groups it has either no effect or a positive effect. Overall, King (2003) con- cluded that the evidence for the deleterious effects of rank structure in orga- nizations of all types is weak: some studies find negative effects, some find no effects, and some find positive effects.
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 181 One thing that is clear, however, is that in spite of reform rhetoric, there was not a wholesale reduction in the depth of police hierarchies in the 1990s. Based on an analysis of longitudinal data collected throughout the 1990s, Maguire and his colleagues (2003) found that vertical differentia- tion has not changed significantly during the community policing era. Some agencies eliminated rank levels, a smaller number of agencies added them, and most maintained their existing number of rank levels. Overall, changes in rank structure were not statistically significant. Future research could help to clarify the effects of the degree of hierar- chy by exploring the possibility suggested by King's review of literature on nonpolice organizations--that the effects of hierarchy may depend on other organizational features or task environments. For example, when account- ability up the chain of command is tightly coupled, an elaborated hierarchy may work to further the organization's objectives, such as reducing the incidence of inappropriate force by officers (Klockars, 1995). When higher ranks are not held closely accountable for the performance of their subordi- nates, the effects of management's policies may be quickly diffused. Geo-Focused Policing Proposition 5: Some police reformers have suggested that policing that is geographically focused on small neighborhood populations would facilitate greater officer familiarity with people in those neighborhoods and ultimately generate more service-oriented practice in those areas. The team policing movement of the late 1960s and 1970s embraced this view. The available research suggests that geo-focused structures may have facilitated greater police knowledge of the neighborhoods to which officers were assigned, but unless other organizational structures were implemented to reinforce the desired practices, the prospects of achieving those results were slim and the scope of the effects usually modest. Police organizations devoted much of the 20th century to reducing the significance of geographic assignment to police work (Reiss, 1992). As in- creasing numbers of officers took specialist job assignments, the geographic territories for which they became responsible increased in size, often to include the entire jurisdiction because beat assignments were frequently changed. In the 19th and early 20th centuries, foot patrol was the common assignment of most officers; they were physically restricted to small beats in which they became "specialists" over the years that they worked that same area. But when the automobile, telephone, and portable two-way radio revo- lutionized policing, the geographic scope of an officer's assigned territory was no longer so limited. As officers finished an assignment, they could be
182 FAIRNESS AND EFFECTIVENESS IN POLICING sent to another anywhere in the city to minimize the response time, rather than waiting for the officer assigned to that beat to become free to answer the call. This made it harder to develop knowledge of and contact in par- ticular geographic areas, a trend accelerated by the increasingly common practice of frequently rotating beat assignments precisely to prevent too much familiarity and corruption between the patrol officer and the resi- dents (Mastrofski, 1981:Ch. 2). By the late 1960s, team policing advocates began calling for a rever- sion to permanent beat assignments and limiting out-of-beat dispatches. These were promoted by such blueribbon commissions as the President's Crime Commission (1967) and the National Advisory Commission on Criminal Justice Standards and Goals (1973:113). Many expected that of- ficers with permanent assignments would be more friendly toward neigh- borhood residents (Wilson, 1968:290-296; Myren, 1972:721-722; Rossi et al., 1974:150-165; Gay et al., 1977:17-19). Some believed that these officers would become more protective of their turf and hence more ag- gressive in crime control (Davis, 1978:136). Later, under community po- licing, reformers went beyond permanent beat assignment, advocating that departments relieve officers of some of the burden of responding to calls for service so that they could spend time getting to know residents on their assigned beat and work with them to solve problems (Mastrofski, 1998:163). Permanent beat assignments became a fixture of community policing (Kane, 2000), and by 1999, 94 percent of local police departments with 100 or more sworn officers reported giving patrol officers responsi- bility for fixed geographic areas, and 47 percent reported assigning cases to detectives on that basis (Reaves and Hart, 2000:xii). Of course, those figures do not reveal the extent to which departments were successful in keeping officers in their assigned geographic areas and how long they were assigned to the same area before receiving a new assignment. However, there are a few studies that do examine the effect of increasing geographic specialization on patrol officer behavior. Most of the available evidence on the effects of geo-focused policing comes from evaluations of team policing and implementing stable beat assignments in the 1970s (see Mastrofski, 1981:75-79 for a review). Eval- uations are based on three types of data: officer surveys of attitudes and self-reported behavior, citizen perceptions of police practice, and field ob- servations of police practice. Although the nature and extent of effects vary with the particular measure, permanent beat assignments are more likely to result in officers who are more knowledgeable of or familiar with citizens in those areas, especially if the stable beat assignments are accom- panied by a program that emphasizes the importance of obtaining commu- nity support (Boydstun and Sherry, 1975:66; Schwartz and Clarren,
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 183 1977:34-39; Mastrofski, 1983). With such programmatic direction, offic- ers appear more positively inclined to be service-oriented, but without that kind of guidance, permanent beat assignments appear to give officers freer rein to follow their own preferences, which in one study meant more ag- gressive crime control methods (Bloch and Specht, 1973:67-75). Two stud- ies that examined the effects of permanent beat assignments over time indi- cated that whatever positive officer attitudes toward the community might develop could deteriorate over time (Bloch and Specht, 1973; Schwartz and Clarren, 1977). Ironically, even though team policing tended to in- crease officers' sense that citizens supported them, when police practices were viewed through the lens offered by citizen surveys, most studies showed that team policing produced few significant changes (Gay et al., 1977; Fowler et al., 1979:127-139). Finally, a study based on systematic field observation of 42 urban neighborhoods assessed the impact of focus- ing officers' beat assignments on very small to very large geographic areas (measured as the population of the officer's assigned area over a year's time). Of the 11 police behaviors measured, the researcher found that only 3 were significantly affected by the size of the officer's patrol assignment area (Mastrofski, 1981:190). Smaller areas were associated with higher numbers of public relations contacts, more comforting behavior toward victims, and more citizens contacted for whom the officer showed some familiarity. Unaffected behaviors included the frequency of service-oriented encounters, residential security checks, suspect and traffic stops, various forms of assistance, the use of force in nonthreatening situations, arrests, and showing hostility to citizens. One study has reported a more aggressive, crime-focused policing as a consequence of geographically focusing the patrol officer's work. Using an interrupted time-series design, a study of patrol in the Philadelphia Housing Authority found that officers who received permanent beat assignments did increase their level of (officer-initiated) field investigations after the imple- mentation of permanent assignments, leading the researcher to conclude that geo-focused policing had produced a greater sense of officer responsi- bility for the assigned area (Kane, 2000). The greater methodological rigor of this quasi-experimental design compared with the cross-sectional regres- sion analysis of the previous studies should be noted, but the comparability of a public housing area to a general police jurisdiction leaves open the question of the generalizability of these findings. In general, this body of literature suggests that geo-focused policing can alter police behavior in ways that have been seen as desirable by team polic- ing and later community-policing reformers, but without other organiza- tional structures to channel police behavior in a particular way, officers do not tend to show distinct or predictable patterns of behavior.
184 FAIRNESS AND EFFECTIVENESS IN POLICING Formalizing Police Policies Proposition 6: One of the frequent recommendations to effect police organization change is to establish new policies by formally instituting rules or procedures that restrict the range of judgment or discretion allowed to police. Limited available evidence suggests that formalizing policy in this way, especially in the area of use of lethal force, can reduce the frequency that officers resort to force. However, instituting a formal policy alone may be an insufficient step to ensure the desired impact. Police leaders may need to implement other changes, such as training, and may need to implement the new formal policy in a larger context of strong political pressure that will reinforce the need for change as felt both inside and outside the department. Formalization is the extent to which the organization imposes rules and procedures to govern the practices of its members (Hall, 1991:63-65). The more formalized an organization is, the less judgment or discretion its mem- bers are authorized to exercise. Formalization is often measured as the ex- tent to which rules, procedures, and practices are written, thus associating formalization with standardization and documentation. Police agencies re- sort frequently to formalization as a strategy to govern the practice of their officers, so there should be ample opportunity to judge the impact of for- malization on police practice. Few studies of the formalization of police policy have examined the impact of formalization generally across a broad range of policy concerns, and in the few cases when they have done so, they have not attempted to isolate the effects of formalization from a variety of other bureaucratic fea- tures thought to be strongly associated with high levels of formalization (centralization, specialization, hierarchy) (Maguire et al., 2003:48-59). Most of the studies of police agency formalization have examined the impact of formalizing a particular policy or changing the nature of an already estab- lished formal policy. Such studies help to determine whether the rules and procedures of policing matter. One policy that has received some research attention is the imposition of restrictive deadly force policies. For instance, Fyfe (1979) found that a restrictive shooting policy implemented in 1972 by New York City's police department produced a dramatic reduction in shootings of fleeing felons. In a later study, Fyfe (1982) attributed the difference in the number of shootings in Memphis and New York to the lack of internal agency con- trols and the absence of clear shooting guidelines in the former compared with the latter. In a review of his and four other studies, Fyfe concluded that "reductions in police shooting frequency and changes in police shooting patterns have followed implementation of restrictive administrative policies
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 185 on deadly force and weapons use" (1988:181). In this same article, he pre- sented evidence supporting this conclusion from three other large agencies. Research has suggested that the extent of the effects of restrictive deadly force policies depends on other factors, some organizational and some envi- ronmental. Training, for example, is held as an amplifier of the effects of restrictive use policies (Fyfe, 1988). Another analysis of restrictive use poli- cies in New York, Atlanta, and Kansas City attributed their success to their following on the heels of highly publicized "critical incidents," creating a political environment which enabled reform-minded chiefs to impose and sustain severe administrative and disciplinary changes (Sherman, 1983). This, the author noted, was similar to the effect of corruption scandals on implementing similarly severe policies to reduce corruption (Law Enforce- ment Assistance Administration, 1977) (see Chapter 7 for a discussion). In general, research on other aspects of formalizing specific policies suggests that, by itself, formalizing policy will often not suffice to secure the desired change in police practice. This was found in case studies on han- dling public drunks and making pornography arrests (Aaronsen et al., 1984:408-436) and the implementation of domestic violence policies (Sher- man, 1992:112). CompStat and Strategic Problem Solving Proposition 7: CompStat is a recent reform that many police leaders feel will revolutionize the management of police organizations. Very little objective empirical research has been done on CompStat or simi- lar management strategies. What is available suggests that it has changed the work practices of middle managers responsible for districts or precincts. However, the research also suggests a number of forces at work that limit the effects of CompStat and may even work at cross purposes. Much more research on the implementation of CompStat is needed before conclusions can be drawn. The previous discussion has focused on the impact of discrete structural features of police departments, but contemporary reform efforts often en- tail changing a number of features as a package of "organizational develop- ment." In recent years, police leaders in the United States have shown tremendous interest in adopting an innovation called CompStat, which sur- faced in New York City in 1994. A 1999 national survey of departments with 100 or more sworn officers showed that a quarter of the sample re- ported implementing a CompStat-like program and another third of the sample was planning to do so (Weisburd et al., 2003). CompStat's origina- tors promised to reduce crime by enabling police organizations to think and act strategically--putting resources where crime problems were emerging
186 FAIRNESS AND EFFECTIVENESS IN POLICING and doing so on a timely basis, before problems became insurmountable (McDonald, 2001; Bratton, 1998; Maple and Mitchell, 1999; Silverman, 1999). An interesting feature of this reform is that the primary targets of change are the middle managers of the police organization, especially those charged with commanding the precincts and districts into which the territory of large urban departments are divided. They become the mechanism by which the police organization is made responsive to the leadership of top manage- ment, and they do so by taking responsibility for closely monitoring the state of crime in their assigned areas and identifying patterns in crime prob- lems as they emerge, of devising and implementing solutions to those prob- lems, and of following through to make sure that those solutions have worked. Top management and the rank and file in the precincts both have a role to play. The former is responsible for clarifying the organization's mission and holding middle managers accountable; the latter are respon- sible for following the direction provided by the middle managers. CompStat programs around the nation go by many names, but they share the following elements (Weisburd et al., 2003): · Mission clarification. A crime control mission that is clarified, sim- plified, and measurable, such as reducing crime by 10 percent in a given year. · Internal accountability. Middle managers (district commanders and heads of specialist units) whose job assignments and career prospects de- pend on knowing what is going on with crime problems in their areas, discern patterns and likely causes of crime, devise effective solutions to those problems, act decisively to implement them, and follow through to make sure that they are successful. · Geographic organization of command. An organization that com- mits most of its resources to the direction of commanders whose responsi- bilities are defined geographically, that is, districts or precincts. · Organizational flexibility. A commitment to flexibility in allocating and deploying resources, so that they will be available when and where they will be most effectively applied to crime problems as they are identified. · Data-driven decision making. The timely analysis of accurate data on crime, so that the organization can mobilize responses rapidly to those areas most needing attention, making heavy use of computerized data sys- tems and crime mapping to identify and monitor hotspots. · Innovative tactics and strategies. Customizing the police response to crime problems as they arise by carefully analyzing the underlying causes and selecting the responses with the best prospects for success rather than resorting to a standardized and uniform response; innovation is prized.
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 187 These features were made most visible in a periodic meeting in which top executives, required middle managers to discuss crime in their areas of re- sponsibility, using computerized pin maps projected onto large overhead screens and lengthy printed reports detailing past and current crime pat- terns and agency activity. During these events, individuals commanders were held accountable, but all others present were expected to contribute infor- mation and ideas in a group problem-solving effort. The national survey showed that departments indicating that they had adopted a CompStat pro- gram showed distinctly different patterns in the policies and practices they described in the areas of mission clarification, internal accountability, and using data for decision making (Weisburd et al., 2003). However, they were indistinguishable from non-CompStat departments on the other features, suggesting that CompStat is not so much generating a new reform wave as riding one that was already begun. The important issue for this chapter's purpose is whether and how the behavior of the police has changed as a consequence of CompStat. There are very few objective analyses of CompStat's impact on organizational practice. A case study of New York City's CompStat program, published a few years after the program was initiated, considered it a resounding suc- cess in all of the above areas (Silverman, 1999). A later field study of CompStat implemented in Lowell, Newark, and Minneapolis suggested a mixed result (Willis et al., 2003). Across the three sites, researchers found the most powerful and consistent pattern of effects was manifested in the behavior of district commanders. These middle managers keenly felt the pressure of their responsibility to know about crime in their districts and act quickly to intervene where spikes in crime statistics and hot spots emerged. Virtually all of these middle managers reported that CompStat had changed the way they spent their work time, devoting many hours to studying crime reports and crime maps to prepare for CompStat meetings. Before CompStat, they would have only occasionally referred to crime sta- tistics, and seldom in a way so focused as individual hot spots. After CompStat, they were obsessed with data. Although some of their motiva- tion clearly derived from a desire not to lose their prestigious job assign- ment by looking bad to top management, the most powerful motivation appeared to be not looking bad before the entire audience, especially their peers. And CompStat meetings increased the frequency with which middle managers consulted each other on their problems. This formal, routinized exchange of information replaced an informal, haphazard pattern of inter- action. Yet the field research also revealed a number of limitations and unin- tended consequences of CompStat at these three sites. First, the intense sense of accountability for results did not trickle down to the rank and file, who
188 FAIRNESS AND EFFECTIVENESS IN POLICING were largely ignorant of and unaffected by what went on at CompStat meet- ings. None of these departments routinely held district-level CompStat meet- ings, during which district commanders held their subordinates account- able, and the absence of these meetings may have limited the diffusion of CompStat's effects throughout the organization, an effect reported in New York City, where such meetings were held (Kelling and Sousa, 2001). Sec- ond, the data analysis employed tended (a) to react to short-term changes in crime patterns while avoiding analysis of broader long-term trends, (b) to focus on identifying the location of hot spots but not figuring out what to do about them, (c) to focus disproportionately on what was happening with individual cases rather than larger patterns, and (d) to be underused for evaluating the effects of police interventions. Third, although district com- manders did enjoy a larger measure of authority delegated from headquar- ters, they remained constrained by civil service rules, union contracts, and local politics in where and how they could mobilize their personnel. Fourth, the pressure on middle managers to come to CompStat meetings with prob- lems already identified and solutions already implemented tended to make irrelevant the suggestions of colleagues, who were also reluctant to offer radically different suggestions, since this would reflect badly on the com- mander who had already committed to a course of action. This preference for decisiveness over rumination mirrored the culture of the officer on the street, albeit at a higher level in the agency. Finally, resort to innovative strategies was the exception, not the rule; most responses involved the tra- ditional responses of increasing levels of police surveillance and enforce- ment activities. The researchers concluded that some of the limitations on CompStat derived from inadequate changes to fundamental structures needed to sup- port a fuller implementation of CompStat, such as the absence of in-depth management training in data analysis and its uses and inadequate staffing of the crime analysis operation. They noted that, although CompStat pro- grams promulgated a narrowing of the police mission to focus on crime control, none of the departments undertook a significant restructuring of their workload to eschew tasks not focused on crime control. Thus, in many respects, CompStat reforms appeared to have been transplanted onto tradi- tional policing structures, rather than having replaced them. The research- ers also suggested that CompStat has some built-in internal contradictions that cannot be overcome, such as the pressure for individual accountability versus the pressure for collaboration or the desire to give district command- ers control of more resources versus the need for flexibility in moving re- sources from district to district as the need arises. A national survey and a handful of case studies are not adequate to draw conclusions about the impact of CompStat on police practice in the United States. The above findings are at best suggestive. Future research on
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 189 CompStat should take special pains to track its impact on decision making at the district level. What are the mechanisms of accountability between the district commander and his or her subordinates? To what extent do they use the crime analysis tools and services of crime analysts to identify prob- lems and devise solutions? How do management practices change when (a) CompStat meetings are routinely held at the district level? (b) middle man- agers are selected and trained to be adept in the use of data analysis, and (c) crime analysis units are given substantially larger staffs with the requisite expertise and resources to provide more cutting-edge support services? INFLUENCE OF EXTERNAL FORCES Police and police organizations are often targeted by outside individu- als and groups who seek to influence the way policing is practiced. Some- times the effects of outside forces are passive, meaning that there is no purposive effort to manipulate police practice, but there is an effect none- theless. For example, the creation of a tourist attraction in a community alters who will be present and for what purposes, thereby changing the kinds of problems requiring police attention that will arise. In this section the committee considers some of the wide variety of external forces that may play on police. The section begins with a discussion of two types of external influences in which it is assumed that the police react to the demo- graphic, social, and economic features of the community at both the neigh- borhood level and then at the city, metropolitan area, or regional levels. Subsequently the chapter turns to purposive efforts to shape policing, issu- ing from political oversight, interorganizational effects from other criminal justice and human service agencies, the state legislature, and the federal government. Neighborhood Characteristics Proposition 8: A common social science research hypothesis is that neighborhood factors directly influence both individual and organiza- tional behavior, including that of the police and police agencies. The relatively small body of research that examines neighborhood effects on police practice shows with fairly high consistency that disadvan- taged and higher crime neighborhoods are more likely to receive puni- tive or enforcement-oriented policing, other things being equal. How- ever, the strength of these effects is small compared with the effects of situational factors, such as those characterized as legal concerns in the previous chapter. The idea that police behavior is affected by neighborhood conditions in
190 FAIRNESS AND EFFECTIVENESS IN POLICING addition to legal, victim, offender, and situational factors has been discussed at length by Klinger (1997). In his ecological theory of police response, Klinger argues that officers learn through experience that crime, especially violent crime, is disproportionately found in areas characterized by socio- economic disadvantage. In such areas, police officers are busier and lack the resources for attending to other service requests. Thus, officers must make triage decisions regarding which actions they will attend to and which they will disregard relative to other matters in their own beat or district. Through work group practices, behaviors that might be defined as deviant in some districts come to be regarded as normal in high-crime areas, and officers understanding of district deviance become unique in each neighborhood. It has also been hypothesized that officers' experiences in high-crime neighborhoods tend to produce a higher level of cynicism and the view that higher proportions of victimizations are at least partially deserved. The fact that offenders are more likely to be victimized than nonoffenders means that officers also learn to regard certain events as expected; again, a greater proportion of deviance in these places is viewed as normal and pursued with less vigor. In addition, higher levels of dangerousness are anticipated in these areas, and consequently greater use of physical restraint, coercive authority, and arrests is more likely. The ecological theory of police response suggests numerous hypotheses about behavior, and the broadest notion is that encounters will produce greater similarities in officer behavior within neighborhoods than between them. According to this theory, differences in police behavior should be found when one community (or district or beat) is compared with another, but relatively few differences should be found among officers working within these areas. To date, these hypotheses have not been tested with empirical data. Nevertheless, a relatively small body of research has examined whether neighborhood characteristics are related to police practices and thus far suggests only weak links. Motivated by concerns about equity, this research has studied whether police respond in similar ways to incidents that occur in different kinds of communities. For example, do police devote the same level of attention to crimes in disadvantaged areas as they do to crimes in wealthier areas of the city? Are police more likely to arrest or use force against suspects in certain types of neighborhoods? Do the characteristics of suspects or victims matter in some places but not others? If certain types of communities are neglected by the police or, alternatively, receive overly aggressive policing, then fundamental concerns about justice are raised. Research in this area typically has used observational data on police- citizen encounters to assess whether neighborhood conditions are indepen- dently related to officers' behaviors. Once situational, suspect, complain- ant, and officer characteristics are taken into account, if neighborhood
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 191 factors are found to be significantly related to police behavior, it suggests that area conditions may be influencing police decision making. Similarly, data from these kinds of studies have also been used to examine whether suspect or complainant influences might have more subtle influences--for example, whether force is more likely to be used against minority suspects in certain types of neighborhoods. Much of the research on community influences has relied on data from the 1977 Police Services Study (PSS). Generally, this research has found that a variety of neighborhood charac- teristics are related to police behavior, although the direct effects are much smaller than the effects of legally relevant characteristics (Smith, 1984, 1986, 1987). In analyses of police activity during unassigned time, Smith (1984, 1986) found that assistance was somewhat more common in areas of lower socioeconomic status and greater racial heterogeneity and less common in neighborhoods with greater proportions of single-parent families with chil- dren. However, police assistance was not significantly related to neighbor- hood levels of mobility, percentage nonwhite, income heterogeneity, turn- over, or age composition. Proactive investigation was somewhat more common in areas with greater proportions of nonwhites, racial heterogene- ity, and people over age 65, and somewhat less common in areas with more people living alone. The broader conclusion inferred from this mixed pat- tern of weak relationships was that officer-initiated encounters with citi- zens during discretionary time are relatively unrelated to neighborhood characteristics (1986:325). In contrast, given a contact between an officer and a citizen, neighbor- hood factors were found to be significantly associated with the likelihood of arrest, the use of coercive authority, and the filing of official reports. Net of individual and situational factors, arrest was more likely if the encounter occurred in a neighborhood of lower socioeconomic status. For example, approximately 12 percent of interpersonal violence incidents resulted in arrest in low-poverty places, while nearly 45 percent of similar incidents in high-poverty areas led to arrest. However, rates of arrest also varied signifi- cantly according to race: white suspects were approximately three times more likely than black suspects to be arrested once controls for community and incident characteristics were considered. Accordingly, Smith suggests that this is primarily a reflection of the victim's status. He also found that police were less likely to arrest the suspect when the victim was black or female. Thus, in incidents of interpersonal violence, police decision making appeared to vary according to community context and victim characteris- tics. Similarly, coercive authority was used more often if the encounter was in a nonwhite or racially mixed neighborhood. Finally, community crime victimization rates and stability influenced official reporting. Police were less likely to file reports in high-crime areas, but more likely to file in un-
192 FAIRNESS AND EFFECTIVENESS IN POLICING stable neighborhoods (Smith, 1986:330-332). In addition, Smith found that police were more likely to file reports for incidents occurring in black neigh- borhoods, and they were less likely to do so for black victims in predomi- nantly black or white areas. A popular conditional effect hypothesis is that black suspects are more likely to be arrested or treated with coercive authority if the encounter oc- curs in a predominantly white neighborhood. No such pattern was observed in the decision to arrest; however, there was a significant interaction be- tween suspect's race and the racial mix of the community for the use of coercive authority. All else equal, police were more likely to use coercive authority against blacks in predominantly black areas than in white areas. The extent to which these findings may be generalized to other types of places or to contemporary policing strategies is not altogether clear. The PSS data were collected over 25 years ago, prior to the implementation of community policing reforms and mandatory arrest policies in domestic vio- lence incidents. In addition, the demands of observational research limited the sites to three metropolitan areas. Hence, the extent to which these find- ings could be generalized to places outside urban areas or of more recent history is unknown. While differences across neighborhoods motivate con- textual hypotheses, the time and expense of directly observing police activ- ity necessarily limit the number of research sites and (perhaps) the general- izability of the findings. Moreover, the contingent nature of many of the findings suggests that neighborhood influences are very complex, or per- haps very sensitive to model specification. Findings based on the more recent Project on Policing Neighborhoods offer an assessment of neighborhood influences that better reflects contem- porary patterns. Mastrofski et al. (2000) studied the extent to which police officers fulfilled citizens' requests for action, such as advising or warning the offender or arresting the offender. The factors that were considered include legal considerations, offender characteristics, victim-offender rela- tionship, and need. These results showed that police were most likely to comply with a citizen's request for a particular action if there was evidence to support the charge, the citizen was credible, the action requested was not the most severe, and the citizen was not requesting action against someone with whom they had a very close relationship. The citizen's race and social status were not significantly related to outcome; however, younger police officers committed to community policing ideals were more likely to com- ply with requests. While community characteristics were not the focus of this research, it was noted that, despite differences between the two cities in poverty rates and race and family composition, there was very little varia- tion in the predictors of these police behaviors across the two sites. Hence, this research suggests that at a broader level of definition, community fac- tors are less important than legal and situational considerations.
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 193 Using POPN data, Mastrofski and colleagues (2002) also examined the effects of concentrated disadvantage in a neighborhood on the inclinations of officers to behave disrespectfully toward suspects. Their hierarchical analysis found that where concentrated disadvantage was high, there was a significantly greater likelihood that officers would behave disrespectfully toward suspects, taking into account a number of characteristics and be- haviors of the suspect and the encounter. Neighborhood differences have been uncovered in the POPN data in police use of physical restraints. Reisig and Parks (2000) examined the rela- tionship between area levels of homicide, concentrated disadvantage, and physical restraint and found that officers are more likely to take such ac- tions in communities with higher homicide rates and in areas with concen- trated disadvantage. However, because this research was designed to in- vestigate whether the use of restraints was the source of greater citizen dissatisfaction with the police in these areas, no multivariate assessments of restraint were provided. Thus, while it can be concluded that such activity is more common in disadvantaged neighborhoods, it is not clear whether this reflects police response to higher levels of danger or to some other community condition. Influence of City and Metropolitan Area Proposition 9: Some "conflict" social theorists argue that American police use their enforcement powers in ways that disproportionately benefit privileged social and economic groups by maintaining the status of those groups--using force against lower and marginalized social and economic groups. They argue that if their proposition is correct, then the coercive aspects of police work should be higher in cities or areas where there is greater inequality and racial diversity, other things being equal. The available research does tend to show that this relationship holds, but there are other theoretical interpretations for these results besides conflict theory. The same social groups that may experience disproportionate coercion from police may also experience dispropor- tionate benefit, if those who are most susceptible to victimization also come from these disadvantaged and marginalized groups. Until re- searchers can estimate the full range of control and benefits distributed by police, there will not be an adequate test of this proposition. A second research tradition of exogenous factors that influence police has examined how city or metropolitan-area characteristics are related to police activity. The general approach of this research has been to use census data and police department information to examine whether such factors as levels of poverty, economic inequality, minority population size, and segre-
194 FAIRNESS AND EFFECTIVENESS IN POLICING gation are correlated with arrest rates or use of force rates once levels of crime are taken into account. Much of this research has been prompted by conflict theory hypotheses, which assert that the police serve as agents of the privileged classes and that economic inequality is maintained by force or the threat of force by agents of the state (Jacobs, 1978a, 1978b; Jacobs and Britt, 1979; Williams and Drake, 1980; Liska et al., 1981, 1985; Loftin and McDowall, 1982; Liska and Chamlin, 1984). If this hypothesis is cor- rect, then police use of arrest, force, or coercion should be higher in areas with greater economic inequality, controlling for violent crime rates. Similar kinds of research have been prompted by the "racial threat" hypothesis, which asserts that once racial or cultural minorities reach a certain proportion of the population (e.g., 20-30 percent; Liska et al., 1985), they are perceived as a threat to the existing political and social order. According to this perspective, racial segregation is one way of reducing this threat, because problematic groups can be controlled more easily if they are spatially segregated. Thus according to this theory, controlling for crime rates, greater policing resources, and higher rates of arrest, force, and coer- cion should be found in metropolitan areas with higher levels of racial in- equality. In addition, the percentage minority in a metropolitan area should have a weaker relationship with policing activities in places where segrega- tion is greatest. Empirical support for many of the above hypotheses has been obtained from analyses of city- or state-level data. For instance, Jacobs and Britt (1979) found that states with higher levels of income inequality tend to exhibit greater police use of lethal force than states with lower levels of income inequality. These models controlled for state differences in violent crime rates, police officers per capita, percentage black, percentage urban, population growth, and region of the country. In these analyses, violent crime rates and population growth rates were also significant. Because most policing jurisdictions are not state-level agencies, the study of differences across metropolitan areas or cities offers more refined conclu- sions. Jackson and Carroll (1981), Jacobs (1978), Liska and Chamlin (1984), and Liska et al. (1981) found that the racial and economic charac- teristics of cities and metropolitan areas are related to police department size (resources dedicated to social control), controlling for reported crime rates. In other words, cities with greater proportions of blacks or nonwhites had more police officers per capita and thus a greater capacity for crime control. Similarly, earlier research demonstrated that fear of crime is higher in cities with greater proportions of nonwhites (controlling for crime rates) (Lizotte and Bordua, 1980; Liska et al., 1981). In two of the most complete investigations of variation in arrest rates, Liska and colleagues studied how urban characteristics influenced police crime control activity (i.e., the ratio of arrests to number of reported crimes). Economic theories of crime control hypothesize that there should be a posi-
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 195 tive relationship between arrest rates and such factors as number of police per capita (Liska and Chamlin, 1984; Liska et al. 1985). This research found that income inequality, percentage nonwhite, and levels of segregation were significantly related to property and personal crime arrest rates, while po- lice size was not. Moreover, evidence about the role of segregation suggested that both the threat and a "benign-neglect" hypotheses were supported. In segregated cities with fewer nonwhites, arrest rates were lower, while in less segregated cities with greater proportions of nonwhites, arrest rates were significantly higher. Thus, these findings support conflict hypotheses by demonstrating that arrest rates are greater in cities in which minorities are more visible. They also challenge the simpler economic hypothesis. Arrest rates are re- lated to policing resources, not because greater resources increase the likeli- hood of arrest, but because both resources and arrest rates are to a large extent influenced by the percentage nonwhite, the segregation patterns, and the economic inequality level of urban areas (Liska and Chamlin, 1984; Liska et al., 1985). A recent study by Holmes (2000) investigated interagency variation in civil rights complaints. He obtained data from the Justice Department's Civil Rights Division on civil rights complaints against the police from 1985 to 1990. Testing a conflict theory explanation, Holmes (2000) found that "measures of the presence of threatening people (percent black, percent Hispanic [in the Southwest], and minority/majority income inequality) were related positively to average annual civil rights criminal complaints" (p. 343). Holmes's study is another in a long line of research evidence support- ing conflict theory explanations for police organizational outcomes; as noted earlier, the committee recommends more research in this area. There are a number of limitations to the research that attempts to test conflict theory applied to policing. First, the results are subject to alterna- tive interpretations. Police arrests may be targeted against persons who live in areas where there are large numbers of low-income or minority residents, but the police may be doing much of their enforcement work at the request of low-income and minority persons who also reside in those same areas (Vold et al., 2002:243). Thus, it is not clear that it is the privileged classes who derive exclusive, or even the most, benefit from these practices. Sec- ond, because these studies use aggregate data, they are unable to specify precisely who (that is, which race or social class) receives police enforce- ment attention in an area. This is especially problematic where there is considerable variability in a geographic area of study, or where there is a high level of movement of people through the area who do not reside there. Finally, the studies do not consider the full range of services police provide, many of which may be viewed as a form of social control but nonetheless highly desired by those subject to them (handling disputes, for example).
196 FAIRNESS AND EFFECTIVENESS IN POLICING Political Oversight: Local Elected Officials Proposition 10: Whether and in what ways politics should influence policing have been topics of debate since Americans formed full-time police departments. Good or appropriate influence is often character- ized as "accountability," while bad or inappropriate influence is termed "interference," "corruption," or just "politics." Making a judgment about the rightness of the influence belongs to the realm of political philosophy, but there are interesting theories available to account for the circumstances under which political influence of whatever sort is most or least likely to have an effect. Based mostly on a few case stud- ies, the available evidence suggests that when local officials, such as the mayor, exercise influence over the department, they tend to do so most potently through executive appointment, job review, and threat of re- moval from office. However, to assess the relative strength of influence exerted by those with official responsibility for political oversight of police, researchers need to consider the relative strength of a wide range of alternative political influences as well, such as collective bargaining units, courts, and civil service commissions. Depending on one's perspective (and assumptions about the character and motivations of elected officials), a demonstration that the preferences of mayors or other local officials have an impact on police programs and practices is either good news or bad news. A substantial portion of the literature on policing views the political environment as potentially cor- rupting of local law enforcement and, like the trend toward professionalism in policing, assumes that a dichotomy between politics and administration (in this case, police administration) is a good thing. Those holding to a classic overhead model of democracy assume that democratic accountabil- ity is possible only if police departments are responsive to the will and political preferences of the officials that the public has elected. Political Principals and Their Agents Cross-cutting this difference in normative perspective is the empirical question of the extent to which elected officials actually do (or even could) influence police programs and operations. This is a specific instance of the more general question of the extent to which there can be political control of the bureaucracy--a question that has yielded a very substantial research literature based on principal-agent theory. This theory, originally devel- oped in economics to deal with contractual arrangements between buyers and sellers in a marketplace, has been extended to the question of political control of the bureaucracy. In the language of this theory, bureaucratic
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 197 officials are the agents who are tasked with carrying out the preferred poli- cies of political principals, such as legislators. However, because specialized administrators have information advantages compared with their princi- pals, due to their expertise and to their more direct involvement in bureau- cratic processes, and because there may be goal conflict between the agent and the principal, the principal is faced with a problem. Agents may well shirk their duty--that is, act contrary to the principal's preferences. Much of principal-agent theory is then focused on mechanisms for ensuring that agents follow their principals' preferences--such mechanisms as the ap- pointment of agency leaders who share the political principal's goals, moni- toring of bureaucratic activity, and bureaucratic design features that pro- vide incentives. The principal-agent problem is not necessarily solved simply because of the existence of these mechanisms. This point is well made by Guyot (1991) in her review of some of the key mechanisms that one might have assumed would be useful in maintaining the accountability of police departments to their political overseers. Appointing a new police chief, Guyot observes, is sometimes so fraught with multiple political considerations having nothing to do with substantive police policy issues that it becomes ineffective as a tool to ensure accountability (1991:237). Elected officials control police department budgets, but this does not constitute a strong tool for accountability because (1) so much of the police department budget is in personnel, and both civil service and union rules constrain redirection of these resources; (2) the nature of the policing func- tion means that the kind of program budgeting that would be needed to make budgetary changes toward policy redirection is not possible; and (3) the more general constraints on the budgeting process--that is, tight time lines and decision making in the absence of long-range planning--also limit the ability of political leaders to use budgetary decisions as a mechanism for controlling the policy direction of the police department. Once hired, a police chief exercises considerable control over police functions not covered by legal, including contractual, obligations. Waterman and Meier (1998) have recently expanded principal-agent theory by noting that the principles of information asymmetry and goal conflict on which it is based should be treated as variables rather than con- stants. That is, in some situations, bureaucratic agents may have substantial information advantages over their political principals, but in other situa- tions, both bureaucratic agents and their principals lack information, and in still other (admittedly rarer) situations, political principals may have the information advantage. Similarly, Waterman and Meier argue that in some circumstances, agents and principals may share goals rather than exhibiting goal conflict. Taken together, these considerations suggest several possible scenarios that must be incorporated into future research.
198 FAIRNESS AND EFFECTIVENESS IN POLICING In addition, Waterman and Meier (1998) offer several important de- ductions that are particularly relevant to policing. First, they suggest that in the situation in which the principal and the agent both lack information yet have goal consensus--a situation they claim to be characteristic of less technical, morality-based matters, such as drunk driving and drug con- trol--there will be a tendency for the political principal to "grab whatever ideas are floating around and adopt them. This permits the politician to take credit for combating the problem. Bureaucrats in these policy areas become advocates--or perhaps more harshly, cheerleaders--for the principal's proposed solutions" (p. 181). By contrast, in those rare situa- tions in which there is goal consensus and the principal has the informa- tion advantage, presumably because they have access to big picture infor- mation that agents lack, the principal-agent problem is solved in favor of the principal's preferences, and agents have minimal discretion. This sce- nario is hypothesized to characterize the handling of such issues as nude dancing, enforcement of dog licensing laws, and parking on main streets in relatively small communities. While they do not offer such pointed applications to policing, Water- man and Meier's other scenarios can be used to suggest still other hypoth- eses relevant to police policy and practice. For example, when there is goal consensus and the agent has information advantages over the principal, the classic model of a politics-administration dichotomy is said to arise, with the principal delegating maximal responsibility for administration to the agent, monitoring loosely through regular reports and intervening only in crises. This scenario may be hypothesized to characterize many of the core functions of policing, such as basic crime control. No work as yet system- atically tests the hypotheses derived by (or derivable from) Waterman and Meier's research (1998). Case Studies Not surprisingly, most studies of the extent to which political officials influence policing are based on case studies or comparative case studies. This is because there are no systematic data bases available to measure relevant independent variables across a large, representative sample of ju- risdictions. Such case studies do, however, have the advantage of allowing a richly textured treatment of political leaders' attempts to influence policing, as well as the capacity to use a historical approach to examine change in a single department (or a small number of departments) over time. By and large, such studies tend to find that political officials have a limited influence on police policy and practice, or that the influence that political officials do have is pernicious. For example, in his case study of the politics of policing in Cedar City from 1960 to 1980, Scheingold (1991)
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 199 emphasizes the police department's "formidable capacity to resist and di- vert" the forces pushing for change. Despite the mayor's aggressive use of appointment power in an attempt to redirect the department by bringing in a series of new police chiefs, the department lurched unevenly in a more progressive direction, influenced more by departmental reaction to scandal and the chief's ability to make use of federal funds for hardware acquisition than by mayoral initiatives. In this case, the limited ability of political offi- cials to influence policing stemmed in part from the relatively low salience of street crime in the community. In her case study of the Troy, NY, police department in the late 1970s to early 1980s, Guyot (1991) explicitly introduces normative criteria to distinguish political officials' involvements in policing that are acts of forc- ing accountability from those that are acts of political interference. She uncovers some instances in which elected officials "appropriately" used their powers to make (or at least try to make) the police department accountable to policy priorities. These include redirecting departmental attention to the problem of excessive use of force, an initiative to establish a police-commu- nity relations board, and the redirecting of priorities toward downtown safety. However, the bulk of the city council's actions with respect to the police were judged by her to be acts of interference, such as the institution of neighborhood safety centers, efforts to establish foot patrols downtown, and the introduction of team policing (1991:244-250). Guyot's results might be interpreted somewhat differently if the question is simply an empirical one of the extent to which police policy and practice are influenced by elected officials (regardless of the efficiency, effectiveness, fairness or pro- priety of that influence). Some of the acts that she judged to involve politi- cal interference would be categorized as examples of a lack of influence, because the attempt at interference was rebuffed by the department (i.e., the police chief's refusal to institute foot patrols); similarly, some acts she judged to involve accountability would have to be reclassified into examples of a lack of influence, again because the attempt at political control failed (i.e., the attempt to institute a civilian review board). A number of studies suggest that mayors and other local elected offi- cials do have an influence on police policies, and that they are more likely to affect policies than practices. In a study of the determinants of lethal force, Jacobs and O'Brien (1998) find that, while the rate of police killing of black citizens is positively related to both the size of and the recent growth in the city's black population, it is negatively affected by the presence of a black mayor. Chaney and Salzstein (1998), in the study referenced earlier, found that the existence of a local-level mandatory arrest ordinance had even greater impact than a state-level mandatory arrest law, thus suggesting that through the enactment of direct orders (here mandatory arrest in domestic disturbance situations), elected officials can influence the behavior of the
200 FAIRNESS AND EFFECTIVENESS IN POLICING police. However, Chaney and Salzstein also found that the presence of a female mayor did not significantly affect arrest patterns for domestic distur- bances. And Salzstein (1989), using a cross-sectional design based on a na- tional sample of police departments, found that the presence or absence of a black mayor did not affect police practice in handling public disorder incidents, such as vagrancy, loitering, public drunkenness, noise complaints, and the like. These null findings can be interpreted as showing a disconnect between elected official preferences and police practice, but they included no actual measures of officials' policy preferences. Instead, these studies assumed that female mayors have preferences for arrests in domestic disturbance situa- tions, and (perhaps with more validity) that black mayors prefer fewer po- lice killings of black civilians. Salzstein (1989) acknowledges that, because the black community has mixed preferences with respect to police handling of public disorder incidents, minority mayors may not be attempting the sort of strong and consistent policy direction in this area that they do with respect to such matters as police use of force against minorities and the need for civilian review boards. Empirical studies showing a connection between mayors or city council members and police policy seem to be more prevalent, encompassing effects on police policy ranging from the hiring of minority police officers to adop- tion of civilian review boards to police corruption. Salzstein (1989) found that the presence of a black mayor is a significant predictor of two police policy outcomes important to blacks: black representation among sworn officers and the adoption of civilian review of the police. That finding par- allels an earlier finding of Browning, Marshall, and Tabb, in a comparative time-series study of 10 California cities from 1960 to 1979, which found that the extent to which racial minorities are "incorporated" in local elected office (i.e., either capture the mayoral office or are part of the dominant coalition of the city council) is a key predictor of the adoption of civilian review boards. In particular, they found that in both cities in which minor- ity incorporation was strong, a civilian review board had been adopted by 1980; in all three cities in which minority incorporation was moderate, such a review board was considered but not approved; and in all five cities in which minority incorporation was weak, a civilian review board was not even considered (Browning, Marshall, and Tabb, 1984:153). Interestingly enough, the same Salzstein (1989) study that showed a connection between black mayors and both minority hiring and civilian review boards showed no evidence that black elected leadership was linked to the adoption of another policy outcome--community-oriented policing. In addition to these studies, using systematic data to analyze variation across multiple cities, there are a number of case studies that demonstrate the impact of elected officials, and especially mayors, on police policy and
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 201 practice. For instance, Greene (1999) offers a case study analysis of how Mayor Rudolph Giuliani was able to substantially alter police policy and practice in New York City by appointing Police Commissioner William Bratton, an individual whose zero tolerance approach emphasizing large- scale arrests of petty criminals and crackdowns on threats to public order meshed with the quality of life issues stressed by the mayor in his campaign. Negative Influence While studies such as these provide evidence that political officials in- fluence police policy in a way that is consistent with notions of democratic accountability, there are also studies suggesting that political officials influ- ence police policy in a much more negative fashion. In his systematic analy- sis of four police departments' experience with corruption scandals, Sher- man (1978) makes clear that corruption can at times become the "policy" of some police departments and that political control by corrupt politicians makes this possible. By the same token, Sherman also makes clear that reform of corrupt police departments is initiated by political officials' re- placement of corrupt police leadership with new leadership. Crucial deci- sions about new policies for internal control of corruption were largely left to the new police leadership rather than being made by external political leadership. Thus, political leaders exert indirect influence (through the ap- pointment process) in policy development for reform of corruption. Still other studies reflect elected officials' impact on both police policy and practice. In a time-series analysis of two police departments, Meehan (2000) showed the "degree to which gang problems are an artifact of politi- cal interests and their effect on police practices." Using a combination of data sources--transcripts and logs of telephone calls to the department 911 number; gang car patrol logs and taped officer responses to dispatch; field notes; newspaper accounts; and an aggregate data set of citizen calls for police service in the city over a six-year period--Meehan shows how the mayor's determination to show high-profile responsiveness to neighbor- hoods during an election year led the police chief to create and publicize a citywide program to curtail street gangs, including the institution of a spe- cial "gang car," and how this high-level policy led to the classification of dispatch calls as gang-related in an unusually high number of cases during the election year. In the two preceding nonelection years, police officials had indicated that gang-related problems were not a priority; in both those years and the subsequent nonelection year, the number of dispatch calls classified as gang-related was lower than in the election year. In short, while limited to a pair of case study communities, this research quite explicitly suggests something like the policing equivalent of what political scientists have called the "political business cycle," which involves the timing of tax
202 FAIRNESS AND EFFECTIVENESS IN POLICING reductions and government benefit increases so as to provide credit-claim- ing opportunities for politicians during election years. Whether positive or negative, the nature and extent of political influ- ence on police policy and practice appear to be contingent, not surprisingly, on the character of the political environment. That environment can change over time in a given jurisdiction, and it can vary at any given time among jurisdictions. A good example of this perspective is a comparative case study analysis of New York, Chicago, and Philadelphia in the 1960s and early 1970s (Ruchelman, 1974). This study considered the ability of mayors to shape police policy and practice in the face of the then growing strength of the police bureaucracy, defined largely in terms of organized power exer- cised through collective bargaining units. In New York and Philadelphia, political parties no longer exerted the kind of influence over municipal poli- tics generally, and policing in particular, that they once had in the heyday of machine politics. In New York, the mayor faced a powerful police bureau- cracy that won almost as many political contests as did the mayor. In Phila- delphia, a powerful police commissioner, who enjoyed the support of the policemen's association, was able to coopt the mayor, who had limited power as a lame duck, as well as a greatly weakened party organization. However, in Chicago, where the mayor headed a still robust local Demo- cratic party, the police department was very responsive to his bidding. Cir- cumstances have changed in all three cities, so it is important to keep in mind that the nature and extent of mayoral influence in each city may also have changed. This study also suggests that a comprehensive analysis of contemporary political influences on police should extend beyond "princi- pal agents" who have official responsibility for governance; other political actors, such as unions, civil service boards, and courts can exercise signifi- cant influence. However, based on the limited evidence available, the oversight exer- cised by local government officials over the police seems to be most potent in the act of executive appointment and, correspondingly, job review and the threat of job loss, an argument made by James Q. Wilson (1968) in his comparison of policing styles in eight local jurisdictions in the 1960s. Political Oversight: Citizen Review Boards Proposition 11: When police departments face charges of police mis- conduct, there is often pressure to institute citizen review boards, which in various ways involve nonpolice in gathering and assessing facts about allegations, review of investigative documents, hearing appeals of com- plaint dispositions, and reviewing police policies and practices. Despite the popularity of citizen review boards, very little research is available
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 203 to indicate the nature and extent of their effect on police practices. The committee recommends a program of evaluation research to determine the nature and extent of such influence on actual police practices. Citizen oversight of the police, and particularly citizen participation in the review of complaints against police officers, is a common prescription for the ailment of police misconduct (Skolnick and Fyfe, 1993). Citizen in- volvement in complaint review takes many different forms in the United States. Some efforts have been made to reduce this complexity in order to describe the main features of citizen review. One such effort describes four models of citizen review: · complaint review that provides for fact-finding investigations of com- plaints by persons who are not sworn police officers (i.e., external investi- gations), reports of which are reviewed by other civilian officials, who make recommendations about disposition to the police executive; · complaint review that provides for investigations conducted by po- lice personnel (internal investigations) with citizen input in the review of investigative reports, whereby citizens may monitor police handling of com- plaints; · complaint review that provides for citizen involvement in an appel- late capacity, with a board to which complainants may appeal if they are dissatisfied with the outcomes of their complaints; and · complaint review that provides not only for citizen involvement in the review of police investigations of individual cases, but also for a citizen role in reviewing police policies and procedures and making recommenda- tions for change (Walker, 2001:61-63; also see Kerstetter, 1985, and Perez, 1994). Citizen involvement in complaint review is thought to have a number of salutary effects. It may improve the perceived receptivity of the com- plaint review system to complaints; the perceived efficacy of the complaint review system; the rate at which perceived misconduct is reported to au- thorities; the depth and thoroughness of complaint investigations; the satis- faction of complainants with their experiences with the complaint review system; the procedural fairness of complaint review, as it is judged by com- plainants; and police performance in interactions with citizens. Complaint review is therefore about much more than the accurate and fair disposition of individual allegations of police misconduct. Little empirical evidence has been produced about the extent to which citizen review meets these expectations. Some evidence suggests that most citizens who believe that they have reason to complain do not complain. Walker and Graham (1998) analyzed data collected through a survey of
204 FAIRNESS AND EFFECTIVENESS IN POLICING 12,000 respondents in 3 metropolitan areas encompassing 24 jurisdictions in 1977. About 6 percent said that they had a reason to complain about an aspect of police service in the preceding 12 months, and of those, 36 percent had complained (half of those had reportedly called the police department). But the study could not examine whether the reporting rate was higher in jurisdictions that provided for greater citizen involvement in complaint re- view. The popularity of citizen review and optimism about its benefits far outstrip the empirical evidence that citizen review has salutary effects. The committee recommends that a program of rigorous evaluation re- search consider a wide variety of citizen review boards, assessing their im- pact on a range of police practices (but especially those features that are frequent targets of citizens' complaints), controlling for the influence of other institutions that might also be responsible for overseeing police integ- rity (e.g., the police organization itself and the courts). A comprehensive evaluation would go beyond examining the effect of civilian review on com- plaints filed but would consider other data sources, such as citizen surveys of high-risk populations (e.g., arrestees and those interrogated during field stops) and direct field observation. Interorganizational Effects: Other Criminal Justice and Human Service Agencies Proposition 12: It is widely assumed that the police are part of a crimi- nal justice system in which policies and practices in one part of the system reverberate in many others. Changes in prosecutorial practices in charging and plea negotiations, penal policies, and correctional re- lease decisions all presumably have important implications for the po- lice work environment, some intended and some not. The available research is too scant to make generalizations about the nature and ex- tent of these influences. Most examinations of criminal justice influence focus on prosaic, but nonetheless powerful, effects. In his 1969 book, Traffic and the Police, Gardiner explored interagency variations in traffic citations, finding sub- stantial variation in citations for moving violations (as opposed to parking violations) in his two samples: a national sample of 508 police agencies and a sample of 180 Massachusetts cities and towns. He found that local court rules about the appearance of officers at traffic ticket hearings had an effect on ticketing rates. Those jurisdictions requiring officers to appear at the next available court hearing date had lower ticketing rates than those juris- dictions with more flexible courtroom procedures. This is one indication of an interorganizational effect on police outputs, an effect that Feeney (1982)
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 205 was unable to detect in his analysis of the relationship between the use of field citations and court orders to reduce jail crowding. Relying on data from 58 California counties, Welsh (1993) tested the hypothesis that court orders to reduce jail overcrowding would expand the use of citations and reduce the number of custodial arrests. While these changes occurred in some cases, Welsh was unable to find evidence of an aggregate change in police arrest and citation behaviors based on the court- ordered reductions in jail crowding. The scant research on this important question suggests that interorgani- zational effects appear to depend on the particular link in question. For instance, in Gardiner's study, court acquittal rates for traffic offenses are not associated with police ticketing rates. But Van Dijk (1988) reports that in Holland a change in prosecution policies regarding bicycle theft rever- berated through the system, subsequently changing police arresting and charging decisions and, later, victimization rates. Research in this genre needs to take into account the dynamic nature of systems, including feed- back processes. Intergovernmental Influences: State Legislation State legislation is logically an obvious source of influence on police programs and operations, in the sense that it provides mandates for action (e.g., state laws requiring arrests in domestic violence incidents), opportuni- ties or tools for action (e.g., state laws allowing for on-site revocation of licenses of drivers found to exceed legal blood alcohol levels), and con- straints on action. That state legislation influences what local police do may therefore be so obvious that it has not motivated as much research as has other topics. However, there is some systematic, empirical research sup- porting the logical conclusion that state law influences what the police do. For example, Chaney and Saltzstein (1998), in a cross-sectional analy- sis of a national sample of municipal police departments, examined the extent to which state and city laws requiring the arrest of perpetrators of domestic violence actually affected policing activity in that regard. They found that the existence of mandatory arrest laws do influence police arrest patterns, at least in those situations in which violence was threatened. Simi- larly, Haider-Markel (2001), in a study based on survey data from police departments and district attorneys, found that the existence of a state hate crime law influenced the perceived likelihood of police arrest in a hypo- thetical hate crime incident, as well as district attorney pursuit of such cases, although it did not influence the perceived likelihood that police officers on the scene would classify the crime as a hate crime. The proposition that changes in the criminal justice system influence police behavior, sometimes in unintended and unanticipated ways, has not
206 FAIRNESS AND EFFECTIVENESS IN POLICING received sufficient research attention. For example, one of the most com- mon assumptions in this regard is that laws passed that allow police to keep some portion of assets seized in drug arrests have served as an incentive for police to increase their attention in this area. Sollars, Benson, and Rasmussen (1994) found that local police in Florida "substantially increased arrests for drug offenses relative to arrests for property and violent crimes" in response to drug asset forfeiture laws. Blumenson and Nilsen (1998) argue broadly that the drug war has transformed the criminal justice, in- cluding law enforcement practice, and credit this change to forfeiture laws. Given the importance of this question, it is surprising how little research has addressed this issue. There is clearly scope for a great deal of additional research on the impact of legislation on the activities of the police and other criminal justice agencies. When states change (and usually tighten) the rules surrounding drunk driving, domestic violence, child abuse, hate crime, gun carrying, and the like, the effects should reverberate through the system. Where there are data, multistate interrupted time-series and quasi-experimental research designs show great promise for estimating the effects of legal interventions on both the problems of interest and the efforts of the criminal justice sys- tem against them. And because there is likely to be variability in both the efforts and effectiveness of individual agencies within states, much more can be learned about the actual implementation of law at the operating level. Intergovernmental Influences: Court Rulings Proposition 13: The American system of government relies heavily on the judicial branch to oversee police practices, and many legal scholars and policy makers assume that most police dutifully follow court rul- ings when made. The available research suggests that police compliance with restrictive court rulings tends to be gradual and incomplete at first, but it increases over time. However, police often seek ways of getting around the requirements of the ruling, often in legal ways. In the United States, the judiciary has a major responsibility for inter- preting laws that specify how the police should practice certain aspects of policing. The most examined of these practices relate to police adherence to constitutional standards of due process in the areas of search and seizure, in-custody interrogation, and use of force. This literature is reviewed in detail in Chapter 7, so in this section we draw general implications from that review and refer the reader to Chapter 7 for details. The body of empirical research on the impact of U.S. Supreme Court rulings suggests that when the Court issues a ruling that restricts or con-
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 207 strains police powers, the police reaction is not one of automatic compli- ance. At first a considerable degree of noncompliance may be expected, as was recorded in studies following the Miranda ruling, which required that the police advise suspects of their right not to self-incriminate and their right to an attorney. Researchers attributed police unfamiliarity with the requirements of the new ruling, but it is also likely that much of the non- compliance issued from reluctance to alter customary practices. Subsequent research, many years after the Miranda standards were established, has noted a high degree of compliance with Miranda, but the police attempt to subvert the protections it affords the suspect by offering perfunctory read- ing of the rights and attempting to persuade the suspect to waive his or her rights. Studies of applications for search warrants show a similar tendency to get around the standards for issuing a search warrant, which include shopping for a judge sympathetic to loose standards, distorting facts, and sometimes even lying. Studies of search and seizure vary considerably in their estimate of the degree of police compliance--due no doubt in part to differences in methods of collecting data--but there is reason to believe that police compliance is much higher when the police desire to seek a successful prosecution, but noncompliance with legal requirements is considerably less likely when the police purpose is to disrupt illicit practices or harass the suspect. This, of course, is a product of the restricted domain of effects that the exclusionary rule can have, since it is only relevant (if raised) when an arrest is made and the prosecutor files charges. The available research suggests that it is safe to conclude that restrictive appellate court rulings will meet with various forms of resistance, subver- sion (both legal and illegal), and just the inertia of old habits dying hard. Resistance and noncompliance out of ignorance give way to more subtle manipulations of the rules as police become accustomed to the new way of doing things over time. Interestingly, researchers have not paid much atten- tion to what happens when the courts loosen restrictions on police. Whether police eagerly take advantage of the loosened rules or whether they also tend to continue past practices is an unanswered question. Intergovernmental Influences: Incentives and the Federal Government There is a substantial, albeit largely descriptive, case study literature suggesting that intergovernmental incentives have a significant impact on both police programs and practices. The most obvious manifestation of this influence, of course, is when the federal government provides grants-in-aid to encourage particular programmatic initiatives. With respect to commu- nity policing, for example, there is at least some evidence that Community Oriented Policing Services (COPS) grants made a difference in the adoption of the community policing model (Moore et al., 2000).
208 FAIRNESS AND EFFECTIVENESS IN POLICING Handling Protests There is also empirical evidence of the federal government's influence on policing even when grants-in-aid are not involved. For example, McPhail, Schweingruber, and McCarthy (1998) have documented a major shift in how police agencies handle protests in the United States and link that shift to several different actions by the federal government--actions that sub- stantially changed the incentive structure. Specifically, they document a shift from an "escalated force" policy of policing protests to a "negotiated man- agement" policy. In contrast to an escalated force policy, a negotiated man- agement policy emphasizes the First Amendment rights of protesters, toler- ates a substantial amount of community disruption, entails substantial communication between police and protesters, aims to minimize both the extent and aggressiveness of arrest tactics, and involves a minimal use of force rather than the principle of using a dramatic show of force (1998: 52-54). The shift from an escalated force to a negotiated management policy of policing protest is attributed to three key federal actions: (1) the critical commentary on the escalated force policy produced by several presidential commissions investigating the riots of the 1960s, (2) a series of Supreme Court decisions establishing a settled body of "public forum law" that in- structs police agencies on the need for maximal tolerance of First Amend- ment rights in spaces that have come to be traditionally used as places for expressive activity, and (3) the civil disturbance orientation course (or SEADOC) developed by the U.S. Army's Military Police School (1998: 54-63). In related research, McCarthy and colleagues (McCarthy, McPhail, and Crist, 1995) estimated that "as many as ten thousand police adminis- trators, police officers, and other public officials may have gone through the SEADOC courses" (p. 62). In sum, through court decisions, presidential commissions, and training courses, the federal government changed the le- gal and informational contexts in ways that significantly altered the incen- tives of police departments with respect to the policing of protests. Another facet of the federal influence on local policing may be the FBI National Academy, which receives nominations from departments across the country for potential attendees. A selected number participate in an 11- week training designed for upper and mid-level law enforcement officers, administered to law enforcement managers in four sessions each year at the FBI training academy complex in Quantico, Virginia. It was established in 1935, and since then has graduated more than 32,000 students. While most are from law enforcement agencies in the United States, more than 2,000 students from 128 foreign nations have graduated from the academy as well. A 1999 internal review of graduates showed that more than 5,700 were currently serving as agency heads, including nearly 4,900 local chiefs of police (Federal Bureau of Investigation, 2002). There is not a single other
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 209 centralized source through which so many police executives in the United States pass. The FBI National Academy is an important source of innova- tion diffusion, both in the United States and abroad, about which virtually nothing systematic is known by social scientists (for basic descriptive statis- tics, see LaSante and Scheers, 1988). Handling Terrorist Attacks Proposition 14: The terrorist attacks of September 11, 2001, may have altered police structure, behavior, and policing style in the United States. It takes only a cursory review of headlines to note that the events of September 11, 2001, altered expectations about policing in the United States. Now there are calls for police to respond to suspicious situations, uncover terrorist networks, and work with other agencies and jurisdictions in an unprecedented way. In the event of an attack, police would find them- selves as first-line emergency responders, perhaps faced with biological or radiological hazards. However, little is known about the capability of espe- cially local police to handle these weighty responsibilities. The most important public study to date was conducted in 1992, by researchers from the RAND Corporation. They surveyed several hundred state and local law enforcement agencies about their level of preparation to deal with terrorist incidents (Riley and Hoffman, 1995). At the time, the respondents envisioned terrorism as encompassing both domestic and inter- national groups. Although surveys were administered to samples drawn from three different kinds of agencies (state emergency planning agencies, state police agencies, and local police agencies), we restrict our discussion to the findings for local police agencies. Half of the sample was selected to represent all police agencies, while the remainder was selected because they were deemed "more likely to be targets of terrorism" (Riley and Hoffman, 1995:6). Therefore, as the authors acknowledge, the findings drawn from this study are clearly not generalizable to the population of police agencies in the United States. The response rates for the local police agencies were low; 53 percent for the stratified random sample and 46 percent for the targeted sample. Preparedness and Response. Like any study of agency responses to terror- ism, the RAND project began with their level of preparedness. One-third of those surveyed had identified terrorist groups within their jurisdiction. About a quarter of respondents had conducted an investigation or con- ducted surveillance on suspected terrorist groups in the past five years, though only 8 percent reported participating in a terrorism-related prosecu- tion. Nearly 23 percent had conducted a threat assessment of potentially
210 FAIRNESS AND EFFECTIVENESS IN POLICING vulnerable targets. Just over 35 percent had a "special unit, section, group, or person that is specifically concerned with terrorism" (Riley and Hoffman, 1995:53). Nearly three-quarters of responding agencies indicated that at least one empoyee had received special training in antiterrorism or counter- terrorism. Almost a decade before the deadly attacks on the Pentagon and the World Trade Center on September 11, 2001, terrorism was clearly on the minds of some police agencies. Relationships with Outside Agencies. The ongoing threat of terrorism, par- ticularly international terrorism, will require police to work in cooperation with other agencies at different levels of government. The RAND study found that, despite popular impressions about jurisdictional squabbles, the relationship between police and other investigative agencies appeared to be fairly good. The departments surveyed reported having the best relation- ship with other local agencies (91 percent were rated as "good" or "very good," followed by state agencies (81 percent) and the federal government (71 percent). In an effort to improve this state of affairs in the wake of September 11, FBI Director Robert Mueller created the Office of Law En- forcement Coordination and selected a retired police chief to head the of- fice. Follow-up evaluations could determine whether the FBI's efforts are having the intended effects. Outlook for the Future. Prior to September 11, some viewed the likelihood of terrorist attacks in the United States as negligible (e.g., Crock, 2000). The nation's police, however, were less optimistic. In 1992, nearly half viewed it as "very" or "somewhat" likely that a major terrorist attack would take place in the United States during the next 10 years. Eerily, more than 72 percent viewed it as likely that terrorists would attack a domestic com- mercial airliner. However, only 1 percent of respondents viewed it as very likely that they would experience a terrorist attack in their own jurisdiction in the next 10 years, with another 26 percent viewing such an attack as somewhat likely. Finally, and tellingly, only 1 percent of respondents viewed their agencies as very well prepared for a terrorist incident; another 11 percent considered their agencies to be well prepared and 57 percent "some- what prepared." Little systematic knowledge exists on local police preparedness for ter- rorist incidents in the wake of September 11. A recent book on disaster preparedness in general (Tierney, Lindell, and Perry, 2001) laments the lack of research on the role of police in disaster situations. They conclude that "almost nothing new has been learned about police and fire department disaster preparedness" (p. 51). In general, their review of research suggests that police agencies are ill-equipped to deal with large-scale disasters, whether natural or man-made. Wenger, Quarantelli, and Dynes (1989), on
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 211 the basis of many years of research at the University of Delaware's Disaster Research Center, report that police agencies tend to plan for disasters in isolation from other community agencies and to think of disaster response procedures as being a simple linear extension of the procedures they ordi- narily use. According to the authors, police departments "tend to devote few resources to emergency planning.... Larger departments are more likely to plan than smaller ones. When they do plan, police agencies tend to plan internally, in isolation from other community organizations; few have adopted an interorganizational approach to the disaster problem. The po- lice appear to believe that disasters can be handled through the expansion of everyday emergency procedures--that is, they do not consider the quali- tative (as opposed to the quantitative) difference between disasters and `ev- eryday' emergencies" (Wenger, Quarantelli, and Dynes, 1989:51). Of course, the apparent tendency for police to plan in isolation is not unique. A recent report on chemical and biological terrorism by the Na- tional Research Council (1999:30) suggests the need for "an institutional- ized linkage between the law enforcement and medical communities." The report suggests that the inclusion "of key medical personnel in anti-terrorist intelligence activity would no doubt be facilitated by their willingness to undergo training on the needs of the law enforcement community, espe- cially procedures for proper preservation of evidence" (p. 30). They con- clude that few local medical communities have effectively developed struc- tural linkages with law enforcement. Local response teams should include a law enforcement officer whose responsibility is to "establish relationships with the local FBI office and other law enforcement agencies sufficient to ensure that the team has the maximum prior warning of potential nuclear, chemical, or biological incidents" (p. 31). The role of local police in prevention and response to terrorism remains unclear. The police are just one element of a vast interorganizational net- work of agencies with an interest in terrorism. These networks are both horizontal and vertical. Horizontally, a variety of agencies performing dif- ferent functions--medical, fire, civil defense, etc.--must all work together in cooperation with the police. Vertically, the response to terrorism takes place a many different levels, including local, county, state, federal, and international authorities. The proper role of police in these networks is still emerging. Although many agencies play a role in the response to terrorism, at the local level, police agencies are now dealing with a host of expanded respon- sibilities. A post-9/11 survey of 192 cities by the U.S. Conference of Mayors found that the 45 percent of total spending on "unanticipated, additional security and readiness costs incurred between September 11 and December 31" was concentrated on the police, with fire, emergency medical services, public health, and other agencies also incurring expanded costs. Respon-
212 FAIRNESS AND EFFECTIVENESS IN POLICING dents estimated that 31 percent of additional costs in 2002 would be ex- pended on the police. Anecdotes provided by survey respondents suggest that most police agencies have been experiencing a variety of increased bur- dens since September 11, including: · Responding to suspicious packages in the wake of the anthrax scare. · Providing extra security for buildings, critical infrastructure (such as water treatment plants), and events. · Reassigning personnel to a variety of new assignments, such as re- gional task forces and special units. · Creating new units and positions to respond to security concerns. · Rapid expansion in hiring. · New training needs. · Assessing local security risks. · Increased planning and policy development. · Dealing with a shortage in staff due to the deployment of National Guard and military reserve units. Only five of the 192 cities surveyed by the U.S. Conference of Mayors re- ported that their communities had not incurred additional costs for security in the wake of September 11. Research Issues. Understanding how police agencies in the United States are preparing for and responding to the terrorist threat presents a special set of challenges for researchers. The 1992 RAND survey described earlier ob- tained a response rate much lower than is typical for establishment surveys of police agencies (see Maguire et al., 2003). The researchers suggest that the sensitivity of the topic probably had an influence on the response rate. The infrequency of actual terrorist incidents on American soil means that random samples of agencies would include few or even none that had expe- rience in dealing with terrorism. Security concerns have probably made rep- etition of the RAND study even more difficult, for police agencies are ner- vous about sharing their preparation strategies, perhaps fearing this will increase their vulnerability (Pluchinsky, 2002). In the interest of national security, many federal and state agencies are now paying more attention to the need for protecting sensitive but unclassified information (Parker, Johnson, and Locy, 2002; Sammon, 2002). The burden is on the research community to ensure continued access to police agencies in a manner that accommodates these concerns. One possible, if more subtle, effect of the attacks of September 11 could be the acceleration of an already existing organizational posture that some observers have dubbed "militarism." As early as 1970, Bittner noted that "the conception of the police as a quasi-military institution with a warlike
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 213 mission plays an important part in the structuring of police work in modern American departments" (Bittner, 1970:52). The roots of militarism extend well back into the history of policing, and Klockars (1988) argued that the emergence of the military analogy served three important functions: (1) it conferred honor and respect on policing as an occupation, (2) its rhetorical focus on fighting a war on crime enabled police to attach a moral urgency to their work, and (3) it sought to adjust the relationship between police chiefs and local politicians to resemble more closely the relationship be- tween military generals and national politicians. Yet it is also the case that the level of militarism is distributed un- equally, so that some police organizations are more militaristic than others (Kraska and Cubellis, 1997; Kraska and Kappeler, 1997), and the overall level of militarism in policing, both in the United States and abroad, is expanding (Kopel and Blackman, 1997; Weber, 1999; Kraska, 1996, 2001; McCulloch, 2001). Kraska and his colleagues have noted an increase in the number of agencies with police paramilitary units. About 89 percent of the large agencies they surveyed reported having such a unit, up from 59 per- cent in 1982 and 78 percent in 1990. Furthermore, this trend is now mak- ing its way into routine patrol operations and into small town police de- partments (Kraska and Cubellis, 1997; Kraska and Kappeler, 1997; Weber, 1999; McCulloch, 2001). Police are not only relying increasingly on the "lethal artifacts" of the military, but also on a variety of advanced nonlethal technologies (Haggerty and Ericson, 1999). These include sophisticated communications and sur- veillance technologies designed for the battlefield, but adapted for law en- forcement. Haggerty and Ericson note that the diffusion of military tech- nologies to civilian law enforcement defies an unconscious trickle-down explanation. Rather, the diffusion is due to a conscious, direct effort on the part of "powerful military/corporate interests...to justify the `dual-use' sta- tus of their technologies " (p. 248). This is one of many substantive areas in which private-market vendors appear to be influencing policing. Some sug- gest that the militarization of criminal justice agencies is a by-product of the emergence of the "criminal justice industrial complex," a term coined by Richard Quinney (1980) (Haggerty and Ericson, 2001; Kraska, 2001). Haggerty and Ericson (2001:54) describe Quinney's concern as with "the alliance being forged between the criminal justice establishment and corpo- rations that require markets for technologies of social control." The effects and extent of militarism must be examined much more closely, since the attacks of September 11 suggest that this trend will con- tinue. Police organizations around the nation are struggling to prepare them- selves for the possibility of future attacks, and some of these responses are likely to be paramilitaristic.
214 FAIRNESS AND EFFECTIVENESS IN POLICING In light of the dearth of research in this area, the committee recom- mends research on the organizational demands of responding to terrorism. What has been done suggests that responding to terrorism places new de- mands on municipal police agencies. It requires them to coordinate their efforts with multiple levels of government; to plan in coordination with public health and medical organizations and with the military; and to learn to safeguard their own employees from new chemical and biological risks. They must continue to maintain open communication with the communi- ties they serve and their commitment to lawful conduct, while they are faced with new information and intelligence needs. From the perspective of local departments, more research is needed on how to respond to these organizational challenges. CONCLUSION The oft-repeated theme of this chapter is that the available evidence is insufficient to draw conclusions and that more research is needed. The defi- ciencies in the research on organizational and environmental effects on po- lice practice fall into four categories. One is simply the absence of studies entirely or of studies more rigorous than the case study. This can be said of virtually all of the propositions considered, but is an especially severe prob- lem for those pertaining to the effects of decentralization, hierarchy, CompStat, oversight by local elected officials, citizen review boards, other criminal justice and human service agency practices, and terrorism since September 11, 2001. In these areas research offers at most interesting specu- lation about what might be happening. A second deficiency is that even the research that goes beyond the case study design is overwhelmingly limited to cross-sectional comparisons that statistically control for the effects of other variables. Studies using quasi- experimental and experimental designs are quite rare. Consequently, it is difficult to have confidence that relationships shown in the data are attrib- utable to causation. A third problem is that much of the data currently collected on police practice have significant problems that limit their utility for applied and scholarly research. The areas of lethal and nonlethal force should be espe- cially high priorities for improvement, since they generate so much concern among the public and are the cause for the more extreme legal measures taken to seek civil remedies for individuals, as well as to enjoin and alter police practices. A fourth deficiency is that measures of organizational practice tend to be heavily biased in favor of measures of policing that are now regarded as traditional such as arrest, use of force, and case clearances. These are, of course, important aspects of what police do, but the data that are collected
EXPLAINING POLICE BEHAVIOR: ORGANIZATIONS AND CONTEXT 215 on a widespread, national basis neglect police practices that reflect the de- sires of many citizens, politicians, and the police. Police handle most disor- ders and minor crimes without invoking the criminal process, yet the Uni- form Crime Reports, the principal national data resource, report only rates of arrest. Police provide an array of other services that often are poorly documented on the agency's records, not to mention their absence from a nationwide database. The same goes for problemsolving, mobilizing and working with the community, and gathering or brokering information func- tions that will be absolutely critical to the successful waging of a war on terrorism. The committee judges that the federal government can and should play a key role in stimulating data collection and research that enables research- ers to fill the many gaps discussed in this chapter. The ready availability of standardized data, collected in many agencies repeatedly over an extended period of time, provides the raw material needed to facilitate a proliferation of quality studies of the impact of police organization and environmental features. However, unlike the top-down approach used in the development of the Uniform Crime Reports, the committee recommends that the federal government begin with a bottom-up approach. Rather than impose new data systems on police agencies or try to persuade them to adopt them, the federal government should begin by supporting the development and test- ing of a variety of different data collection systems in various police depart- ments around the nation. This will have two benefits. First, it will stimulate competition, thus increasing the quality of results, and it will increase the likelihood of uncov- ering systems that will work for the broadest possible array of local depart- ments. Second, such an approach is more likely to result in systems that local departments themselves can and will use for their own practical pur- poses. The ultimate goal, of course, is to have in place systems that large numbers of departments will commit to using over long time periods, but unlike the Uniform Crime Reports, it may be less important to promote comprehensive adoption than it is to ensure that adoption be sufficiently widespread that policy makers can evaluate the effects of organizational innovation and environmental influences over a wide range of conditions. The committee judges none of the propositions considered in this chap- ter to have conclusive findings, but certain of them point in directions that have implications for public policy. First, it appears that simple recommen- dations about increasing or decreasing the size of the police organization to produce a particular pattern of behavior are unlikely to yield desirable re- sults with a high degree of predictability. The effects of size appear to be contingent on many other factors, although current research is unable to specify what those conditions are. Second, when police organizations create new job specialties, such as community policing officer, how those officers
216 FAIRNESS AND EFFECTIVENESS IN POLICING spend their time may change in the direction anticipated by the department, but patterns in how the officers interact with the public do not appear to be different from those of generalist patrol officers. Third, when patrol offic- ers' responsibilities are focused geographically by permanent beat assign- ments, they may increase their knowledge of the geographic area, but there do not appear to be clear changes in patterns of officer behavior unless they are reinforced with other structures, such as training, supervision, and in- centives. Fourth, formalizing restrictive policies on the use of lethal force can reduce the frequency with which police use it, but the strongest effects may occur only when the policy change is stimulated by a crisis. Fifth, how police behave toward the public does appear to be conditioned by the char- acter of the neighborhood, disadvantaged and high crime areas receiving more punitive policing than other areas. Similarly, at the level of city and metropolitan area, police deliver punishment to those communities that are more disadvantaged. Although these results are consistent with a conflict theory perspective, they are also consistent with the notion that the police are serving the disadvantaged persons in those areas. Finally, police appear to respond to restrictive court rulings initially by resisting them. However, over time they do acclimate to them, increasing the frequency of compli- ance. Yet they also find ways to get around the purpose of those rulings, suggesting that whatever the courts' expectations, police are not inclined to accept the goals of efforts to change their practices through litigation.