Title: Measuring What Matters: Proceedings From the Policing Research Institute Meetings (1-64). Series: Research Report Author: NIJ Published: National Institute of Justice, July 1999 Subject(s):community policing, problem-oriented policing, and police planning and management 96 pages 244,000 bytes ------------------------------- This is an ASCII plain-text file. To view this document in its graphic format, download the Adobe Acrobat graphic file available from this Web site or order a print copy from NCJRS at 800-851-3420 (877-712-9279 for TTY users). ------------------------------- U.S. Department of Justice Office of Justice Programs National Institute of Justice Measuring What Matters Proceedings From the Policing Research Institute Meetings Research Report Cosponsored by the National Institute of Justice and the Office of Community Oriented Policing Services COPS Office of Community Oriented Policing Services ------------------------------- U.S. Department of Justice Office of Justice Programs 810 Seventh Street N.W. Washington, DC 20531 Janet Reno Attorney General Raymond C. Fisher Associate Attorney General Laurie Robinson Assistant Attorney General Noel Brennan Deputy Assistant Attorney General Jeremy Travis Director, National Institute of Justice Joseph E. Brann Director, Office of Community Oriented Policing Services ------------------------------- Office of Justice Programs World Wide Web Site http://www.ojp.usdoj.gov National Institute of Justice World Wide Web Site http://www.ojp.usdoj.gov/nij Office of Community Oriented Policing Services World Wide Web Site http://www.usdoj.gov/cops ------------------------------- National Institute of Justice Measuring What Matters: Proceedings From the Policing Research Institute Meetings Edited by Robert H. Langworthy Cosponsored by the National Institute of Justice and the Office of Community Oriented Policing Services July 1999 NCJ 170610 ------------------------------- Jeremy Travis Director, National Institute of Justice Joseph E. Brann Director, Office of Community Oriented Policing Services Opinions or points of view expressed are those of the authors and do not necessarily reflect the official position of the U.S. Department of Justice. ------------------------------- The National Institute of Justice is a component of the Office of Justice Programs, which also includes the Bureau of Justice Assistance, the Bureau of Justice Statistics, the Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime. ------------------------------- Contents Introduction Measuring What Matters: A Policing Research Institute Robert H. Langworthy Meeting One: November 28, 1995 Measuring What Matters in Policing Alfred Blumstein Great Expectations: How Higher Expectations for Police Departments Can Lead to a Decrease in Crime William J. Bratton Measuring What Matters: A New Way of Thinking About Crime and Public Order George Kelling Measuring What Matters: Crime, Disorder, and Fear Wesley G. Skogan Measuring What Matters Darrel W. Stephens The Incivilities Thesis: Theory, Measurement, and Policy Ralph B. Taylor Meeting Two: May 13, 1996 Constituency Building and Urban Community Policing David E. Duffee, Reginald Fluellen, and Thomas Roscoe Community Policing: What Is the Community and What Can It Do? Warren Friedman and Michael Clark Americans' Views on Crime and Law Enforcement: A Look at Recent Survey Findings Jean Johnson, Steve Farkas, Ali Bers, Christin Connolly, and Zarela Maldonado To Whom Do We Answer? Johnnie Johnson, Robert Berry, Juanita Eaton, Robert Ford, and Dennis E. Nowicki The Police as an Agency of Municipal Government: Implications for Measuring Police Effectiveness Mark H. Moore and Margaret Poethig The Police, the Media, and Public Attitudes Aric Press and Andrew Benson Constituent Expectations of the Police and Police Expectations of Constituents Stuart A. Scheingold Meeting Three: December 4, 1996 Some Really Cheap Ways of Measuring What Really Matters Carl B. Klockars What Matters Routinely? Robert H. Langworthy Appendix: Author Biographies ------------------------------- Measuring What Matters: A Policing Research Institute Robert H. Langworthy In 1992, a paper by George Kelling appeared in The City Journal titled "Measuring What Matters." In this paper, Kelling raised the perennial specter of police performance measurement, but this time with a new twist. His discussion focused on the organizational performance measurement demands of community-oriented policing. In essence, Kelling's argument was that our traditional yardstick was outdated and needed to be changed. The National Institute of Justice (NIJ) and the Office of Community Oriented Policing Services (COPS) also recognized that our historic measures of police organizational performance were outmoded. To address this issue, NIJ and COPS collaborated on a first-of-its-kind Policing Research Institute that focused on "measuring what matters." The Policing Research Institute examined the implications of community policing for measuring organizational performance and helped move the industry toward a new, more relevant set of assessment criteria. To accomplish this task, police executives, researchers, scholars, and others interested in police performance measurement were invited to Washington, D.C., to address a range of measurement issues. Measuring What Matters consisted of three meetings, each focusing on a particular set of topics. Each meeting considered a set of discussion papers commissioned by NIJ and COPS and prepared by selected Institute participants. The meetings produced: o Heightened awareness within the police and research communities of changing measurement needs associated with the shift to community policing. o Better informed Federal research and development grant programs on measuring police performance (the NIJ Measuring What Matters research solicitation, issued in May 1997, was shaped in part by these discussions). o A series of papers, designed to reach a wide audience, chronicling the Institute proceedings (see, Brady, 1996, for the first in this series). o This compilation of revised papers. The first Institute meeting, held on November 28, 1995, focused on two questions: How do we measure the amount of crime, disorder, and fear and their effects on the quality of community life? and Should we expect police activities to impact on measures of crime, disorder, and fear and how will we know if they have? Discussion papers regarding the first question were prepared by Darrel Stephens, then Chief of the St. Petersburg, Florida, Police Department; Wes Skogan, Professor at Northwestern University; and Ralph Taylor, Professor at Temple University. The second question was introduced by papers prepared by William Bratton, then Commissioner of the New York City Police Department; Al Blumstein, Professor at Carnegie Mellon University; and George Kelling, then Professor at Northeastern University. In essence, these discussions focused on how to measure police organizational performance and whether we can reasonably and unambiguously attribute changes in crime, fear, and disorder to it. The second session, held on May 13, 1996, focused on police constituencies' expectations and, perhaps more importantly, what police could expect of different constituencies in a partnership. Seven discussion papers were presented at this meeting. Jean Johnson, of Public Agenda, addressed public attitudes toward the police. Aric Press, then of Newsweek, and Andrew Benson, then of the Cleveland Plain Dealer, collaborated on a discussion paper that explored the relationship between the police and the media--particularly the print media. David Duffee, Professor at the University at Albany, and Stuart Scheingold, Professor at the University of Washington, independently considered alternative police constituencies and the implications for community policing partnerships. Warren Friedman, of the Chicago Alliance for Neighborhood Safety, and Michael Clark, of the Citizen Committee for New York City, collaborated on a paper that explored the community and police partnership from the perspective of "what's in it" for each of the partners. Mark Moore, Professor at Harvard University, discussed police organizations as instruments of local government with a particular focus on the nature of interagency partnerships. Finally, Johnnie Johnson, Jr., then Chief of the Birmingham, Alabama, Police Department; Dennis Nowicki, Chief of the Charlotte-Mecklenburg, North Carolina, Police Department; and Robert Ford, Chief of the Port Orange, Florida, Police Department, collaborated on a paper that addressed their experience in identifying important constituencies, what those constituencies expect of the police, and what the police can expect of those groups. This session was designed to address a salient community policing problem--police do not deal only with one community but simultaneously with many publics, often with competing expectations and differing capacities to be partners in a community policing enterprise. The title of the discussion paper prepared by Carl Klockars, Professor at the University of Delaware, captures the focus of the final Institute meeting, held December 4, 1996. His paper, "Some Really Cheap Ways to Measure What Really Matters," was intended to lead into a discussion of indexes and instruments that police agencies might consider to assess organizational competence, skill in the use of force, and integrity. The format of this session departed from the previous sessions by dividing the participants into small groups to discuss economically feasible and meaningful measures of police organizational performance. These breakout sessions considered a discussion paper I prepared while working with NIJ on a sabbatical from the University of Cincinnati. The five breakout groups were each assigned a conceptual domain and asked to focus their discussions on that topic. The domains were: o The impact domain--how might intended police effects on the environment be measured. o The process domain--how might police know if they are doing their work as they should. o The community assessment domain--how might public assessment of police performance be monitored. o Organizational health--how might police departments know if their employees are satisfied with their work. o Community context--how might police organizations monitor changes in the work environment that impede or promote their ability to achieve organizational goals. The aim of this meeting was to initiate discussion of organizational performance measurement systems that could provide information to organizations that they can use to monitor and contextualize their performance. Community policing, with its emphasis on problem solving and community restoration, significantly expands the police domain and demands that organizational performance be reconceptualized. It is no longer sufficient to measure organizational crime-control prowess (which we never did very well). Now we must address crime control plus the expectations created under the rubric of community policing. The Policing Research Institute improved our capacity for "measuring what matters" in the context of this new policing paradigm. This collection of papers was instrumental in shaping those conversations. References Brady, Thomas. Measuring What Matters, Part One: Measures of Crime, Fear, and Disorder. Research in Action. Washington, DC: U.S. Department of Justice, National Institute of Justice, 1996, NCJ 162205. Kelling, George. "Measuring What Matters: A New Way of Thinking About Crime and Public Order." The City Journal (Spring 1992): 21-33. National Institute of Justice. Measuring What Matters in Community Policing: Fiscal Year 1997. Solicitation. Washington, DC: U.S. Department of Justice, National Institute of Justice, 1997. ------------------------------- Meeting One: November 28, 1995 Measuring What Matters in Policing Alfred Blumstein The police and measurement of their impact The most traditional measure of police effectiveness is typically reflected in some measure of the aggregate crime rate or, possibly, in its disaggregation into crime types about which the public may be most concerned. When the crime rate is increasing, the public might demand police accountability for the rise. Usually, however, the police are quite effective in fending off those challenges, and thus we more often consider the rise to be attributable to demographic shifts or changing social conditions. When the crime rate is declining, the situation is usually quite different. It is common for the more aggressive police officials to seek to claim credit for the decline, usually attributing that decline to the latest operational innovation they have introduced. I have seen declines attributed to a new K-9 corps, new management practices, or a special action force designed for rapid response. Thus, we have one of the important measurement dilemmas on the effect of policing on crime--the asymmetric nature of police officials' claims of credit for their control over crime cycles: They claim credit for the decline, but they avoid any blame when crime is on the rise. A second issue closely related to crime measurement is that of arrest, and here we have a similar situation. Many police see their primary function not to be as closely related to crime as to the arrest of those who violate the law. Until recently, with the advent of community policing, arrest was their primary interaction with the community. Since most arrests result from onsite detection or witness or victim identification, shifts in the arrest rate for any particular kind of crime can also be affected by police policies or practices (e.g., setting up speed traps, cracking down on prostitution, setting up a burglary sting) or exogenous events involving changes in the composition of crimes (e.g., growth in the fraction of homicides involving strangers, which are more difficult to solve than those involving intimates). Here, again, it is important to distinguish the contribution associated with more effective policing from that associated with shifts external to policing. Closely related to crime is the issue of the fear of crime, and there is little question that anything that can be done to reduce that fear contributes to an improvement in the quality of life in a community, even if there is no impact on the crime rate itself. Also, since the police are one of the few agencies that are on the street all the time, there are many other aspects of quality of life to which they can contribute (ranging from rescuing the proverbial cats from trees to the settling of disputes that might escalate to serious violence). Even though the connection of these activities to crime may often be indirect, they clearly contribute to the community's support of the police in their crime-related work. In addition, there are many other community-related activities the police engage in that may be seen as ends in themselves but that also contribute to improved ability to prevent crimes or solve them once they occur. This is one of the basic principles underlying problem-oriented policing and community policing. Crimes can be prevented if the conditions leading to them can be identified and the potential offenders dissuaded from pursuing the crime. Also, connection to the community and its information networks provides important opportunities to learn of the perpetrator of a crime and enhance the likelihood of an arrest. Since arrest probabilities are so small, this potential for enhancing the intelligence capability represents a far more significant means of increasing general deterrent effectiveness than any of the changes that might be considered downstream from arrest in the criminal justice system. Aside from these activities in which a common interest exists between the police and the community, there is another aspect of policing that must be considered in any measurement of police performance. Policing inherently involves conflict between the police and at least some members of the community who may be--or may be suspected of--violating a law. Interacting with such suspects often involves the use of force in ways that may be seen as excessive by the suspect, bystanders, or viewers of a videorecording of the encounter. For a variety of reasons that could be legitimate (e.g., greater hostility to police based on past encounters or by oral history in the community) and illegitimate (e.g., racism by individual police officers), these situations occur disproportionately with minority suspects, and they represent a major problem in policing in minority communities where strong positive connections between the police and the community are most needed. Here, again, these problems could be attributable to police performance (e.g., inadequate training leading to premature invocation of excessive force) as well as outside the control of the police (e.g., when the community rallies around a legitimate arrest because emotions have been aroused over a previous questionable one). Thus, in addressing the issue of measuring police performance, we have two primary challenges: (1) identifying the variety of ways in which the police contribute to or detract from community well-being, and (2) partitioning both blame and credit for such changes, at least in a binary way between police and nonpolice factors. In this paper, we begin by addressing the issue of crime and arrest, partly because of its traditional relationship to policing and partly because it is one aspect that is regularly measured and reported to the Federal Bureau of Investigation (FBI) for the Uniform Crime Reports (UCR), thereby permitting comparison across police departments. These data, with local augmentation, provide a base for empirical analysis that enables a police department to identify where it is being effective or ineffective. That information and its analysis should be used for the basic purpose of continuous improvement, which should be far more important to effective management than the short-term political benefit of overblown claims of performance successes. Factors in crime and arrest Perhaps the most important indicator to the public about police performance is its effect on the crime rate; the magnitude of that effect is widely debated. Some argue that social and economic conditions, demographic shifts, and individual choices unaffected by police activity represent the total influence on crime rates. Others--notably police officials during crime downturns--argue that the credit fully belongs to the police. Of course, there are many points between 0 and 100 percent, and so a more meaningful partition somewhere in this range would generally be desirable. There seems to be wide agreement that a large fraction of the crime rate--and particularly the violent crime rate--is largely immutable and unresponsive to anything the police might do short of a massive intensification of police presence in the community and in everyone's lives. But there is also little doubt that more aggressive or targeted police tactics (e.g., intensive patrol or focused stop and frisk to confiscate guns in high-violence areas) or changes in police strategy (e.g., use of community policing to develop community ties to identify problems before they become crimes and obtain critical intelligence information on potential or actual crimes) can have a sizable effect on suppressing some crimes. It would appear to be valuable for most police departments to develop a tight feedback measuring capability enabling them to observe the influence of changes in tactics (typically short-term response) or strategy (where the response is expected to take longer and will not be seen as quickly) on crimes or arrests. The jargon for this approach has recently emerged almost as a religion in industry under the name "total quality management." This requires maintaining detailed and high-frequency information on crime measures. But it also requires keeping careful logs of police operations, particularly noting those locations and situations where there has been a change from what was previously standard or routine. This latter aspect is necessary to permit the linkage between operational actions and their consequences. Attributing the changes to "better policing," without being able to identify what aspect of "better policing" to apply elsewhere to achieve comparable success, may have its political and public-relations values but does not directly improve the effectiveness of police management. Of course, the problem is complicated by the fact that changes in the crime rate will often be generated by factors exogenous to anything the police might do. This could occur, for example, with the appearance of a new gang, the initiation of a new drug market, or the outbreak of warfare between two rival gangs. Although police efforts could well contribute to suppressing that increase once it occurs or keeping it from escalating, it is quite difficult to anticipate its emergence. But displaying speed and effectiveness in responding to its emergence can also be a factor inhibiting its appearance in the first place. Isolating how police contribute to upward or downward shifts in crime or arrest rates requires that information be maintained on key factors that might explain the shift. These should include at least the following: o Precinct or other spatial units, especially to distinguish those places where special effort or changed tactics or strategy are applied. A geographic information system (GIS) can be particularly helpful in maintaining and displaying such information. o Age, particularly because different criminal justice approaches are applied to different age groups. Incarceration and its associated incapacitative effects are most likely to influence older groups; younger groups are more likely to respond to changes in socialization and family structure patterns. o Drug markets, since so much of crime can be linked to drugs. The mores and practices that surround drug markets can easily contaminate the communities in which they are located. In addition, it is important to maintain other baseline data against which to relate the changes, such as locations in which officers are assigned at different times and shifts or those areas where innovative or experimental operations are introduced. Basic demographic information by location on socioeconomic conditions, family structure, and age and race composition are needed to provide a basis for measuring rates. In addition, the analysis should include intelligence information on the emergence of gangs and their criminogenic activities and on markets for drugs and guns and other criminogenic products. Whatever is used as a performance indicator poses the danger that operating officers will work at manipulating the measure itself rather than the underlying process being measured. This is of particular concern with respect to crime statistics, which are principally generated by the police. Intensive emphasis on crime statistics provides an undue incentive to distort the recording and reporting of the phenomenon being observed. Some homicides could be classified as suicides, robberies as larcenies, aggravated assaults as simple assaults, and auto thefts as joyriding. There could be a greater degree of unfounding of marginal crimes. And any police officer with sufficiently strong incentives who controls recording and classification can make the results look more favorable merely by changes in recording or classification practices.[1] The similar phenomenon with arrest statistics and clearance rates has been pointed out by Skolnick[2] in his classic work. Measures beyond crime and arrest Although crime is certainly a salient measure, it is clear that police have--or should have--a responsibility for other facets of the quality of life in a community. Some of these relate to fear of crime (which may or may not respond to shifts in actual rates of crime or victimization); some relate to affecting police ability to deal with crime (e.g., connections to the community and associated access to intelligence regarding crime). In this period of distrust and hostility between police and certain sectors of the community, especially in minority communities, it is important to measure the state of those relationships. These issues are addressed in this section. Fear of crime Fear of crime does not derive from a careful reading of UCR or National Crime Victimization Survey statistics. Rather, it is stimulated by dramatic incidents (the Polly Klaas murder and its impact on the passage of "three strikes" laws is a prime example), repetition of highly visual stories about crime on TV news programs, and reports of incidents involving individuals one knows or hears about. Thus, the time trends in fear could easily move in the opposite directions from crime trends. Indeed, even though there seems to be strong evidence of a growing fear of violence in the United States, most Americans would be surprised to learn that the homicide rate trend in the United States has been flat for the past 20 years, has not been increasing at all, and has been decreasing since it peaked in 1991.[3] It would be desirable to have a regular measure of fear in any community, particularly to see how that level of fear shifts with individual crime events, changes in the reporting of crimes, changes in police deployment tactics, and any of the other activities police engage in, whether intended to deal with fear or with crime itself. That might be done through periodic surveys of the community. But generating sample sizes of sufficient frequency with the potential for small-area estimation would probably make the cost of such surveys prohibitive for other than special measurement associated with a particular experiment or innovation. It would be much more desirable to have unobtrusive measures (see Webb et al.)[4] of public fear. That could be reflected in the number of people who are willing to walk in the street at night and in the use of places like public parks that may be viewed as inherently dangerous. One interesting such measure that has previously been reported on is the sale of the early evening edition of the Daily News in New York City, a reflection of the willingness of people to go out at night to buy the paper. These measures have the advantage of reflecting behavior rather than attitudes, they can be easily and cheaply obtained, they can be a good reflection of the state of fear in a neighborhood or community, and they involve no distortion of the behavior through the process of measurement. Finding such measures is an important challenge. Citizen cooperation with the police and use of excessive force Citizen cooperation with the police is a critical aspect of policing. It will be reflected in improved intelligence information for policing and a generally supportive and prosocial attitude within the community. Various indicators of this might be reports of citizen intelligence, surveys of the community, improvement in crime clearance rates, and various related measures. One of the most important factors inhibiting citizen cooperation with police is the tension, particularly in minority communities, between the police and the community. Because such communities tend disproportionately to be the locus of serious crime, it is critical that effective management control be maintained over excessive use of force. This requires a mixture of training, discipline, and punishment for blatant violations. Measurement of the level of such violations can be very difficult. For example, as the public comes to perceive police management as being more responsive to these concerns, it is possible that this increased sensitivity could stimulate reporting of incidents that might not otherwise have been reported and so give rise to an increase in the reporting of incidents. Thus, some kind of calibration is necessary to assess the threshold of incidents being reported by location and nature of the encounter. State of disorder One important indicator of a sense of disorder in a community is the "broken windows" theory highlighted by Wilson and Kelling.[5] This does seem to be an important issue for indicating both the quality of life in the community to its residents and the care with which policing is being done. Research possibilities These issues of measurement of police contribution are certainly important. In light of the large expenditure (in the order of $50 billion) throughout the Nation on policing, it is striking how little effort has been devoted to measuring police performance and using such measurements for the purpose of continuous improvement. In the military, beginning more than 50 years ago, operations research groups were assigned to many operating units to perform exactly that function. It would be extremely useful for the National Institute of Justice (NIJ) to identify a number of police departments that would value such service and establish pilot units to carry out measurements and report on the results of those measurements directly to top operating officials. This kind of activity is particularly useful when there are regular repetitions of the same kind of operations (e.g., police patrol). In establishing such groups, it is important that they maintain scientific integrity and their results not be oriented toward the public relations effort for the department. If that becomes the case, then there will be strong pressures to distort the results. The danger of these distortions could be reduced by establishing an external audit overseeing the work of these pilot programs. Aside from this more general assignment of operations research groups, it would be desirable to pick several cities that are willing to engage in careful and detailed incident-based data collection (e.g., through the National Incident-Based Reporting System) on crime and arrests to perform the partitioning and attribution discussed earlier in this paper. In the process, new methods of measurement and analysis are likely to be developed, and those results are likely to be generalizable to other jurisdictions, particularly to the operations research groups assigned to a number of departments. Approaches such as this would bring the competence that has been extremely important in enhancing military and business performance into the world of policing. It has the potential to significantly enhance the professionalism and effectiveness of management, not only in the jurisdictions where the studies are pursued but in others to which their results might be generalized. This is clearly an important mission for NIJ and would cost a tiny fraction of the operating cost of policing. Notes 1. My own experience highlights some of these possibilities. I was in New York (well before William Bratton was commissioner of the New York Police Department) and experienced an event at 5 p.m. on a summer Sunday afternoon in a crowded part of midtown that was a cross between a mugging and a pickpocketing incident. I asked the police officers who came to my aid following the incident if they wanted to take a report, and they replied, "Nah, that kind of thing happens here all the time." In another incident in Pittsburgh, when I tried to report an attempted larceny, I was bounced from central headquarters to the local precinct, where they tried to bounce me back to headquarters. When I told precinct staff I had already spoken to someone at headquarters, they told me to come into the police station to file the offense report--which I never did. Although this may be fairly common police practice, intensive evaluation of a unit on the basis of the crime reports on its beat could easily be seen to shift the frequency with which crime reports are discouraged or rejected. 2. Skolnick, Jerome H., Justice Without Trial: Law Enforcement in a Democratic Society, New York: John Wiley, 1966. 3. See, for example, Blumstein, Alfred, "Youth Violence, Guns, and the Illicit-Drug Industry," Journal of Criminal Law and Criminology 86 (1) (Fall 1995): 10-36. 4. Webb, Eugene J., Donald T. Campbell, Richard D. Schwartz, and Lee Sechrest, Unobtrusive Measures: Nonreactive Research in the Social Sciences, Chicago: Rand McNally, 1966. 5. Wilson, James Q., and George L. Kelling, "Broken Windows: The Police and Neighborhood Safety," Atlantic Monthly (March 1982): 29-38. ------------------------------- Great Expectations: How Higher Expectations for Police Departments Can Lead to a Decrease in Crime William J. Bratton Police management I have been asked to write on the question: "Should we expect police activities to impact on measures of crime, disorder, and fear, and how will we know?" I'd like to begin by turning the question around: If we don't expect police activities and police departments to have an impact on crime, disorder, and fear, they almost certainly won't. By accepting the prevailing image of police departments as slow moving and relatively ineffectual bureaucracies, and by assuming that nothing can be done to change them, we are, in effect, making a self-fulfilling prophecy. No organization, whether it is a police department or a private business, is going to achieve high-performance results in an atmosphere of such low expectations. I am a police manager, not a criminologist. I tend to think about crime not as a sociological problem but as a management problem. The scholarship about the underlying causes of crime is very interesting, but it is of limited utility to someone charged, as I am, with public safety in a large city. The fact that many criminologists have argued that police don't have much impact on crime adds to my management problem. My job is to direct police resources and motivate 38,000 police personnel. I cannot afford to subscribe to a system of belief that tells me the police can't accomplish our primary mission of controlling and preventing crime. Instead, like many police managers, I've turned to modern business theory and the study of how to make large organizations work more effectively toward their goals. Goals, it turns out, are an extremely important part of lifting a low-performing organization to higher levels of accomplishment and revitalizing an organizational culture. Goals become a means not only of measuring success but of replacing unproductive or counterproductive behaviors with effective, goal-oriented activity. Goals can be used to inspire an organization, long dominated by negativism and faultfinding, toward positive cooperative efforts and, therefore, toward success. As a police manager, I have learned how to set ambitious goals for police departments as the first step toward achieving ambitious results. In this paper, I will describe two police management stories: the New York City Transit Police since the early 1990s and the New York Police Department (NYPD) in the past 2 years. I think I can make a strong case that management changes and goal setting in both organizations were the primary catalysts for the steep decline in subway crime, beginning in 1991, and in citywide crime, beginning in 1994. I use the word catalyst intentionally. In organizations as large and complex as the Transit Police and the NYPD, no management team can claim sole or even primary credit for success. The role of top management is to motivate and support the organization as a whole, driving it to work to its full potential, but the credit for ultimate success belongs to the cops, detectives, supervisors, and precinct commanders who take our plans into the real world and make them work. Following the general police management discussion, the second part of this paper will discuss what we are doing in New York in terms of the relevant criminological theory about police departments and crime. It also considers some of the other possible factors, besides the NYPD, that might be causing the decline in New York City crime. In certain quarters, there seems to be a near-absolute certainty that police did not and could not have caused the steep drops. Scholars are ready to attribute these declines to demographics, social causes, supposed changes in the drug market, and unsubstantiated speculations about drug gangs making peace--in short, to any possible cause except police work. I think most of these alternative explanations can be easily discounted. They are simply not supported by the facts in New York City, where the number of youths between the ages of 15 and 19 has increased slightly rather than decreased, the economy is relatively stable, drug-use patterns are relatively unchanged, and small drug gangs continue to fight over turf in a number of locations throughout the city. I am hopeful this symposium will begin to change some of the preconceived notions about policing and crime. Better management, better strategies, higher expectations, and more effort on the part of police departments can do far more than just affect crime rates at the margins. We have in the Nation's police departments an enormous untapped potential. If we can bring just a portion of that potential into play, we can have a swift and decisive impact on crime. If we start to use police resources strategically and efficiently, we can cut crime by 20, 30, or even 50 percent in the space of several years. Consider the following story. A series of robberies is taking place in a neighborhood and giving the local area a steeply rising crime rate. It just so happens that this neighborhood has enough political clout to have an elite police unit, expert at apprehending robbers, assigned to the problem. With its special skill, the unit identifies the robbery patterns, deploys its resources, and systematically apprehends the members of two loosely knit robbery gangs. The robbery rate and the crime rate in the neighborhood plummet. Did the police cause the drop in the local neighborhood crime rate? Of course they did. But I can hear the arguments now. A police department could never apply that level of skill and resources to an entire city. Neighborhoods without clout--poor and minority neighborhoods especially--would be slighted. Crime would be displaced from the places where elite units are active to the neighborhoods where they are not. And so on. If I were to assert that lowering the crime rate in an entire city--even in New York City--is simply the process of repeating the success of the elite unit over and over again, many criminologists would be skeptical. They would be even more skeptical if I were to say that an entire police department--even the NYPD--could be geared to function like an elite unit, bringing to bear the same kind of timely intelligence, rapid deployment, effective tactics, and relentless followup that make elite units so effective. But that is exactly what I am going to argue because that is what the New York experience, both the Transit Police and the NYPD, demonstrates. The Transit Police When I became Transit Police Chief in 1990, subway robbery rates were rising steeply, disorder was rife in the system, and fare evasion was skyrocketing out of control. Robberies rose 21 percent in 1988, 26 percent in 1989, and about 25 percent in the first 2 months of 1990. Many of these robberies were what we called "multiple perpetrator" cases, involving five or more youths who would often attack and beat subway riders in order to rob them. A lot of the robberies seemed to be crimes of opportunity. The groups doing the robberies were not real gangs but loosely organized associations of youths who knew the subway was a good place to steal. They would meet after school or encounter each other in the system, look for a likely target, and strike. As more and more kids picked up the tricks of this nefarious trade, the subway robbery rate headed off the chart. The farebeating problem was just as severe. This is a petty crime that can collectively amount to a colossal theft. In 1990, at the peak of the problem, some 57 million fare evaders were costing the public about $65 million. The turnstile areas were overrun not only with farebeaters but with token thieves, who sometimes seized control of subway entrances and brazenly collected tokens from commuters as they shooed them through illegally opened exit gates. The public was appalled and frightened by the spectacle. The criminals were emboldened by it. In addition, we faced a huge disorder problem beyond the turnstiles. Some 5,000 homeless people--most of them drug abusers--were trying to live on trains, platforms, and in the restricted track areas. In fact, more than 80 homeless people died in the subway in 1989. In addition, aggressive panhandlers and illicit hawkers were everywhere, disrupting transit operations and lending an air of chaos and disorder to the entire subway environment. I drew on the collective wisdom of dozens of Transit cops--many of whom were frustrated because they had never been given a chance to try their ideas--to develop a Transit Police patrol strategy concentrating on robbery, fare evasion, and disorder. We all agreed there was a clear connection between felonious crimes of opportunity, i.e., robberies and petty crimes, and violations. Seeing an environment of apparent disorder, young multiple perpetrators reasonably concluded that they could get away with anything in the subway, including beatings and robbery. We had to change their perceptions in a hurry. We coupled a program of full enforcement of all subway rules and regulations with a targeted attack on repeat subway felons, especially youth gangs. Instead of closing multiple-perpetrator cases after one or two arrests--as we had been doing--detectives were instructed to pursue all of the participants in a robbery. Even if we failed to find them all, we reasoned, the effort of searching, bringing witnesses into schools, and the general ubiquity of Transit Police detectives in pursuit of subway robbers would start to alter criminals' perceptions of the chances of success in a subway robbery. We also greatly intensified the pursuit of people wanted on subway warrants. Using computers and faxes, we cut the time it takes for the police to act on a bench warrant from 30 days to 24 hours. Our warrant unit started work at 2 a.m. when the fugitives were still at home, and our apprehension rate rose sharply, eventually rising to more than 60 percent. We turned out to these locations in force, once again sending a message that subway criminals were being relentlessly pursued. In the fare evasion sweep, we developed a near-perfect tactic for the subway. Previous programs to attack farebeating had usually focused on deterrence by stationing uniformed officers in front of turnstiles. The cops hated this work, and the uniformed presence wasn't having any impact on the overall farebeating problem. We began intensive plainclothes fare evasion sweeps throughout the system. The sweeps not only caught farebeaters in the act, they also gave us an opportunity to intervene with robbers because every arrested farebeater could be searched for weapons and checked for warrants. Not surprisingly, most subway robbers weren't paying the fare, and a good number of them were caught in our sweeps. During the first 6 months of this operation, about one in seven people arrested for fare evasion was wanted on a warrant. The last piece of the puzzle was our attack on disorder. We mounted a huge outreach effort to the homeless, cutting the resident homeless population in the subway by about 80 percent over a couple of years by steadily enforcing the rules and offering round-the-clock transportation to shelters. We quelled disorder among school-age riders with a safe passage program on 80 key trains and intensive truancy sweeps. We began enforcing the rules and regulations of the subway system against panhandling, illicit merchants, smoking, drinking, lying down in the system, and many other antisocial acts. The message was sent by both our uniformed patrol force and anticrime plainclothes units: The subway system is under alert police control. It took about 6 months to put everything in place, but subway crime then began dropping, and it kept dropping for the next 5 years. Total subway felonies and robberies declined every month from October 1990 through October 1995, with the exception of March 1993, when there was a slight increase in both categories. If anything, the trend accelerated under my successor, Michael O'Connor, and has continued to accelerate since the merger of the Transit Police with the NYPD in April 1995. The bottom line? Subway felonies in the first 10 months of 1995 have fallen nearly 64 percent compared with the first 10 months of 1990. Subway robberies have fallen 74 percent. There are fewer than 20 felonies a day on a system that carries more riders daily that the population of most American cities. Even more surprising, given the proportions of the problem, was the Transit Police's success against fare evasion. By the end of 1994, it was cut more than in half. By the end of 1995, it will have dropped by two-thirds, for a total savings of about $40 million. It would be difficult to identify a demographic or social cause for the decline in subway crime. Subway ridership is poorer, younger, and more minority than the city as a whole. Yet, in the early 1990s, subway crime dropped far more steeply than New York City crime, of which it is a subset. Between 1990 and 1993, the drop in subway robberies was three times greater than the drop in citywide robberies. In 1991, subway robberies accounted for nearly two-thirds of the drop in the citywide robbery rate, even though subway robberies never represented more than 10 percent of the citywide robbery total. What, besides the work of the Transit Police, could possibly explain that? Yet, as a closed and contained system, the subway does present a special case. By intensifying police efforts in the subway, the Transit Police may have been driving crime to street level. It is possible to argue that subway crime was merely displaced to the rest of the city. The Transit Police experience in the early 1990s showed how a police department can swiftly and effectively redirect its efforts toward solving key problems and achieving key goals. It also showed that a redirected police department can prevent crime by changing criminals' perceptions of their chances of success. But it does not prove with any certainty that such a redirection can reduce an entire city's crime rate. For that kind of evidence, we will have to turn to the NYPD during the past 2 years. The NYPD When Mayor Rudolph Guiliani appointed me New York City Police Commissioner in 1994, we both believed the NYPD had vast untapped potential. But like the Transit Police, the New York City Police Department needed sharply focused strategies and a stronger direction to achieve its potential. With its array of skilled and experienced personnel, the department was like a race car that had never been driven more than 40 miles an hour. The mayor and I decided to experiment by putting the pedal to the floor. We challenged the NYPD to focus its full talents and resources on its core missions of driving down crime and controlling disorder. We set a public goal for the department of a 10-percent decrease in felony crimes in 1994. While many within and outside the department were skeptical that we could come anywhere near to achieving this goal, we ultimately exceeded it with a 12-percent decline in 1994, and we are exceeding it again with an expected 16- to 17-percent decline in 1995. It took some doing to propel the organization forward. Although the public believes that police departments spend all their time thinking about and combating crime, the truth is that these large organizations are rather easily distracted from their core mission by the political or social issue of the moment. In addition, the burden of emergency response leaves police leaders with the sense that there is always something urgent to do, and this day-to-day emergency footing cuts into the time spent on strategic planning. Work on crime is usually done on a case-by-case basis without any real strategic oversight. As a result, police organizations can be particularly subject to drift. Traveling further down the ranks, one finds many of the problems that plague any large bureaucracy. For years, the NYPD had been organized around avoiding risk and failure. Although the department is decentralized into 76 precincts, precinct commanders had been constrained on every side by regulations and procedures issued from headquarters. Many police operations, such as prostitution sweeps and execution of search warrants, could only be conducted by centralized units, reflecting an abiding distrust of precinct personnel and resources. Yet, despite the micromanagement, the department was providing little in the way of genuine strategic direction. It was clear what precinct commanders and personnel weren't allowed to do, but it was much less clear what they ought to be doing to combat crime, disorder, and fear. Beginning in 1994, there were major changes in the management philosophy of the NYPD. We established seven crime control strategies dealing with guns, drugs, youth violence, domestic violence, reclamation of public spaces, auto-related theft, and police corruption. In all these areas, we got the entire organization thinking about how to attack crime and disorder problems, best deploy police resources, disrupt criminal enterprises, and use each arrest to develop information that would lead to other criminals and arrests. Precinct commanders were granted far more latitude in initiating their own operations and running their own shops. Uniformed patrol cops were encouraged to make drug arrests and assertively enforce quality-of-life laws. At the same time, the central strategic direction of the department became far stronger and the lines of accountability far clearer. Today, avoiding failure is no longer a formula for success. Instead, the positive efforts of commanders and cops at reducing crime, disorder, and fear are being recognized and encouraged. For the first time in its history, the NYPD is using current crime statistics and regular meetings of key enforcement personnel to direct its enforcement efforts. In the past, crime statistics often lagged behind events by months, and so did the sense of whether crime control initiatives had succeeded or failed. Now there is a daily turnaround in the "Compstat" (computer comparison statistics) numbers, as the crime statistics are called, and NYPD commanders watch weekly crime trends with the same hawk-like attention private corporations pay to profit and loss. Crime statistics have become the department's bottom line, the best indicator of how the police are doing, precinct by precinct and citywide. At semiweekly Compstat meetings, the department's top executives meet in rotation with precinct and detective squad commanders from different areas of the city. During these tough, probing sessions, they review current crime trends, plan tactics, and allocate resources. Commanders are called back to present their results at the Compstat meetings at least once every 6 weeks, creating a sense of immediate accountability that has energized the NYPD's widely scattered local commands.[1] Four steps or principles now guide the department's patrol and investigative work: timely, accurate intelligence; rapid deployment; effective tactics; and relentless followup and assessment. Debriefing people taken into custody, even for minor crimes, is now standard practice, and it has greatly increased the department's timely, on-the-ground intelligence. Computer pin mapping and other contemporary crime analysis techniques are functioning as the NYPD's radar system, achieving early identification of crime patterns. The barriers that long separated the department's Patrol Bureau, Detective Bureau, and Organized Crime Control Bureau have been broken down, and a new spirit of cooperation is resulting in the rapid deployment of appropriate resources. Although overall strategic guidance flows down to the precincts, many of the tactics that are accomplishing the strategies flow up from precinct commanders, squad commanders, and rank-and-file police officers and detectives. In the 6-week Compstat cycle, the effectiveness of every new tactic or program is rapidly assessed. Failed tactics don't last long, and successful tactics are quickly replicated in other precincts. Gathering field intelligence, adapting tactics to changing field conditions, and closely reviewing field results are now continual, daily processes. The NYPD can make fundamental changes in its tactical approach in a few weeks rather than a few years. The new flexibility allows much quicker response to shooting and robbery patterns. Identified by computer pin mapping, shooting "hot spots" can be blanketed with uniformed and plainclothes quality-of-life enforcement. People carrying illegal guns begin to realize they risk facing gun charges after being arrested for a minor offense. The result is fewer guns carried, fewer guns drawn, and fewer guns used. We have seen a 40-percent drop in handgun homicides in New York City since 1993. The new strategic approach to crime problems has sharpened the focus on the criminal support system: on burglary fences, auto chop shops, stolen car exporters, and gun dealers who supply both drug dealers and armed robbers. In many instances, we have been able to dismantle key pieces of the criminal enterprise. Shutting down local fences, for instance, can have a dramatic effect on neighborhood burglary rates. It may take burglars some time to find another outlet for their stolen goods. The same is true of auto thieves, who need an immediate outlet--e.g., a chop shop or stolen auto exporter--because stolen cars are difficult to hide and easy to identify. We are also focusing on people wanted on warrants who we believe are likely committing additional crimes. Like the Transit Warrant Unit before it, the NYPD Warrant Unit has been revitalized. It has rearrested 10,103 wanted felons in the first 10 months of 1995, compared with 6,113 in all of 1993. Intensive quality-of-life enforcement has become the order of the day in the NYPD. Throughout the city, we are responding to problems such as public drinking, "boombox cars," street prostitution, and street-level drug dealing. Neighborhoods feel safer, and people see the police taking action against these highly visible problems. The NYPD's success against the "squeegee pests," who had begged for money by washing car windows at most highway entrances in Manhattan, is a prime example of what steady quality-of-life enforcement can accomplish. Continuing police pressure, backed by arrests when necessary, has all but eliminated what was once a constant urban annoyance. The NYPD Civil Enforcement Initiative has given us a powerful tool to combat petty crime and disorder. First developed by my predecessor, Commissioner Ray Kelly, and by Jeremy Travis, who then was NYPD's deputy commissioner for legal matters and now is the director of the National Institute for Justice (NIJ), civil enforcement sends NYPD attorneys into the field to assist precinct commanders in devising their enforcement strategies. Together, they use civil law--especially nuisance abatement law, police padlock law, and various forfeiture proceedings--to augment the traditional police sanctions of summons and arrest. They close brothels and drug and gambling locations and confiscate drug dealers' cars and cash. We have been able to have a significant impact on street prostitution by arresting johns and confiscating their cars, which we are authorized to do because the car would have been used in the intended crime. We have also had a powerful impact on boombox cars--using the threat of a temporary confiscation of the auto to be used as evidence. We have achieved a high level of compliance in neighborhoods that were once continuously assaulted by these drive-by noise polluters. All this focused, strategic police activity has translated into steep declines in crime. The seven major felonies were down 12 percent in 1994 and, according to preliminary data through November 12, are down 17 percent in 1995. The preliminary numbers through November 12 show a 2-year decline of 27.4 percent. Crime is down in every felony category, including 2-year drops of 39.7 percent in murder, 30.7 percent in robbery, 36.1 percent in auto theft, 24.4 percent in burglary, and 23.8 percent in grand larceny. Only the declines in felonious assault (12.9 percent) and rape (7.7 percent) have failed to reach 20 percent for the 2-year period. These relatively lower numbers probably reflect the department's domestic violence strategy, which is actively eliciting complaints of assault and sexual violence from battered spouses. In terms of human impact, the real numbers are even more impressive. After steep declines in 1994, there have been 51,728 fewer felonies in 1995 through November 12, including 373 fewer homicides, 47 fewer rapes, 11,949 fewer robberies, 3,103 fewer assaults, 12,520 fewer burglaries, 7,788 fewer grand larcenies, and 19,988 fewer auto thefts. There have been declines in every borough and precinct in the city. All five of the city's boroughs have registered 2-year declines of 23 percent or more. Keep in mind that Brooklyn and Queens would be the fourth and fifth largest cities in the country if they were independent municipalities. In effect, we have achieved crime declines of 23 percent or more in three of the five largest cities in the country. One clear benefit of the strategic policing approach has been the allocation of police resources where they are most needed and the consequent declines in crime in some of the most crime-prone neighborhoods in the city. As of November 12, for instance, the 75th and 77th precincts in Brooklyn, which are among the toughest in the city, were the leaders for real-number declines in homicides, shooting victims, and shooting incidents. The 75th precinct, covering East New York and Brownsville, has seen 45 fewer killings this year. The 67th precinct, another tough neighborhood in Brooklyn, leads the city in real-number decline with 544 fewer robberies. The 107th and 109th precincts in Queens, which had been the car-theft capitals of the world, saw real number declines of 1,186 and 1,063 auto thefts, respectively, through November 12. If the current trend continues through the end of this year, total Uniform Crime Report (UCR) index crimes in New York City will have fallen 26 percent between 1993 and 1995 and 38 percent since 1989. These decreases are even more impressive when compared with the percentage change in total UCR index crimes in other venues: Whereas crime fell 3.0 percent in the Nation as a whole and 9.0 percent in New York State during calendar year 1994, New York City's total UCR index crime fell 11.7 percent--our largest percentage decrease since 1972. New York City's ranking for total index crimes among the Nation's 25 largest cities moved from 18th in 1993 down to 21st in 1994. The reduction in New York City crime has effectively pulled the Nation's aggregate crime level down quite significantly. Based on the Federal Bureau of Investigation's (FBI's) preliminary 1994 UCR figures, crime reductions in New York City accounted for approximately 33 percent of the national homicide and robbery reductions and 70 percent of the national decrease in motor vehicle thefts. Although preliminary 1995 FBI UCR data are not yet available, we expect that New York City's decreases in crime will again contribute significantly to the Nation's overall reduction in crime. Why are the steep declines in crime happening at this time? I believe it is because of fundamental changes in the NYPD's management philosophy and operating principles. We have gone from a micromanaged organization with little strategic direction to a decentralized management style with strong strategic guidance at the top. Our four operating principles--timely, accurate intelligence; rapid deployment; effective tactics; and relentless followup and assessment--have made the NYPD a much more responsive, flexible, and effective force in the field. In the broadest sense, an effective police department can't keep people from becoming criminals or control the social and demographic forces that, according to many criminologists, engender criminal activity. But we can keep people from becoming successful criminals. We can turn the tables on the criminal element. Instead of reacting to them, we can create a sense of police presence and police effectiveness that makes criminals react to us. And then, in a narrower sense, we do keep people from becoming criminals or at least from committing criminal acts as they realize their chances of success are much smaller. This is certainly what the New York City Transit Police achieved in the subway to drive robbery rates down 74 percent. The young felons who committed most of the subway robberies quickly learned that their chances of success had been greatly reduced. Now the NYPD is sending the same message to New York City as a whole, and we are seeing comparable results. Criminology tends to view criminals as a kind of irresistible social force. Its prognosis for the future amounts to the cry of "Look out! Here comes a demographic bulge in the crime-prone age cohort of 15- to 19-year-olds, and we are all going to be swamped by it." I don't think so. Criminals are not an irresistible force. In fact, the criminal element responsible for most street crime is nothing but a bunch of disorganized individuals, many of whom are not very good at what they do. The police have all the advantages--in training, equipment, organization, and strategy. We can get the criminals on the run, and we can keep them on the run. It is possible. We are doing it in New York. Theory and practice One of the prevailing views in contemporary criminology as I understand it is the position that police have little impact on crime--that variations in the rate and prevalence of crime within a community are primarily or entirely attributable to variations in population demographics, the impact of social trends, and a number of economic factors. Criminologists, some of whom are quite fixed in their opinions, cite innumerable studies employing a variety of methodologies to show the relationships between these variables and the rate of reported crime or crime victimization. Specifically, they point to the relative size of a community's cohort of young males between 15 and 19 years of age as a primary determinant of crime rates, along with the availability of guns, the supply-and-demand economics of the illicit drug market, drug-abuse patterns in the community, and a host of other broad social and economic variables. These views are supported by empirical research showing statistically significant and highly positive correlations between the rate of crime and the various demographic, social, and economic variables over time as well as by intuitive arguments and anecdotal evidence. As a basic tenet of epistemology, however, we cannot conclude that a causal relationship exists between two variables unless the intuitive explanation for the relationship has face validity--it must make sense and conform to our objective observations of the world around us--and unless three necessary conditions occur: one variable must precede the other in time, an empirically measured relationship must be demonstrated between the variables, and the relationship must not be better explained by any third intervening variable. Although contemporary criminology's explanations for the crime decline in New York City meet the criteria of the first two conditions, they don't explain it better than a third intervening variable. That variable is assertive, strategic enforcement by police officers in a well-managed and highly directed police agency. When it comes into play, the causal equation is radically altered. As a corollary to the assertion that crime is primarily pulled by the engine of social and demographic trends, contemporary criminology maintains a longstanding belief that police activities have little or no appreciable effect on crime, despite the public ideology and political rhetoric periodically mustered to justify larger police budgets and staffing increases. In support of this belief, academicians proffer a number of empirical studies showing that the addition of police resources, including personnel, has rarely, if ever, had a sustained impact on crime rates. If increasing the number of police within a given jurisdiction has no discernible impact on crime, the reasoning goes, the institution of policing is powerless to influence crime. This logic incorrectly assumes that all police patrol activity is undertaken with the same intensity and that police officers in disparate agencies will be deployed, managed, and directed in the same or similar fashion. I do not take issue with the empirical validity of any of these studies or with the observation that police activity has historically had little impact on crime. I do question the basic premise that because no credible causal relationship has ever been shown to exist between police activity and reductions in crime, no causal relationship can exist. One of the earliest studies of this issue was conducted in the NYPD's 25th precinct in 1954, where the operational strength of the precinct was more than doubled for a 4-month period. At the project's conclusion, reported street robberies declined by an astounding 90 percent, and burglary and auto-theft reports--crimes that are typically visible to patrolling police officers--declined as well. Increased manpower had no impact on homicides and minimal impact on felony assaults, however, since many or most of these crimes took place indoors or in locations that patrolling police could not easily scrutinize. Despite the project's brevity and several flaws--it did not control for or measure the displacement of crime, and it did not account for reductions that might be attributable to factors other than manpower deployment--it was used to justify demands for an increase in police personnel and resources (Wilson, 1985: 62-63). In 1966, consistent results were obtained when this study was replicated through saturation patrol in the 20th precinct. Street crimes visible to patrol again declined in the target precinct, but no appreciable declines were noted in crimes occurring indoors or in other private places. As James Q. Wilson (1985) pointed out, the results of these two projects "were sufficiently striking and consistent to warrant entertaining the belief that very large increases in police patrols may reduce "outside" or "street" crimes significantly, at least for a short period of time" (p. 64). Neither study, though, used sufficient controls or measures to adequately determine how much of the crime-reduction effect was due to deterrence and how much was due to displacement. The main conclusion derived from these studies--that any impact the police may have on crime is due to a deterrent effect and is limited to the type of street crimes easily visible to patrolling officers--prevailed in criminology and police management circles for several decades. The accuracy of this conclusion is called into question by our contemporary experience in New York City, where we have achieved steep reductions in all categories of crime, irrespective of their visibility to patrolling officers. We have not found any significant variance in the relative proportion of reported "indoor" versus "outdoor" crimes in any offense category. Samuel Walker (1985) has argued that the addition of more police to an agency has historically had no demonstrable effect on crime. Although Walker acknowledges that police do deter crime to some unspecified and limited extent and arrests serve a specific deterrence purpose through incarceration of criminals, he says the impact of mere police presence as a crime deterrent can scarcely be measured in precise terms. Walker asserts that while police patrol since the time of Robert Peel has been designed to prevent crime, the "police are at best a last-resort, reactive mechanism" of social control, and he concludes quite validly that "even the most superficial evidence suggests no relationship between the number of cops and the crime rate" (p. 104). Walker's characterization of police patrol as a "last-resort, reactive mechanism" describes activities of agencies and officers cast in the traditional mold. Walker has argued elsewhere (1984) that this reactive model of police organization was in large part forged as the legacy of O.W. Wilson, whose classic Police Administration became the "bible" of an entire generation of police executives. These executives embraced Wilson's gospel of efficiency and were profoundly influenced by his ideology of crime suppression, which emphasized the deployment of resources to control "serious" crimes--the seven felonies comprising the UCR crime index (pp. 409-410). Indeed, for decades police executives were locked into a narrow mindset in which the UCR index was practically the sole benchmark for police performance. When index crimes declined, they took credit; when index crimes increased, they blamed either improved reporting rates or broad social factors beyond their control. The narrow mindset has its advantages. I can hardly dispute the empirical evidence cited by Walker (1985) or the overall validity of his argument, but I would emphasize that the state of contemporary policing in New York City differs enormously from the traditional reactive model on which criminologists have based their conclusions. In New York City, we have radically altered the face of policing by empowering the agency and its officers with policies and tactics that "capitalize on community crimefighting initiatives and take the bad guys off the streets," a strategic approach that John DiIulio has so graciously dubbed "Bratton's Law" (DiIulio, 1995: A19). Perhaps the best-known and most frequently referenced study of the effect of police patrol on crime is the Kansas City Experiment in 1974. This year-long study determined that changing the level of preventive patrol within demographically matched neighborhoods had virtually no impact on the number of reported crimes or the level of fear experienced by residents of the various neighborhoods. However, as James Q. Wilson (1985) observed, the experiment "did not show that police make no difference, and it did not show that adding more police is useless in controlling crime. All it showed was that changes in the amount of random preventive patrol in marked cars did not, by itself, seem to affect . . . how much crime occurred or how safe citizens felt" (p. 67, emphasis in original). He points out that the experiment might have yielded very different results if important changes were made in the way police were used, including assignment to plainclothes patrol, sustained attention to places identified as having been frequent sites of crimes, or more extensive followup investigation of past crimes (pp. 67-68). After examining the body of research on the impact of police on crime, Wilson (1985) concluded that "what the police do may be more important than how many there are, that patrol focused on particular persons or locations may be better than random patrol, and that speed may be less important than information" (p. 71, emphasis in original). There is much wisdom in Wilson's conclusions, and they certainly jibe with our experience in New York City. What we have done in New York is, in effect, to focus and coordinate police officers' activities, to free them from random patrol duties by providing coherent tactical directions and enforcement strategies to occupy their undevoted time, and to provide them and their commanders with accurate and timely crime intelligence necessary to make a difference. They relentlessly follow up their enforcement activities and identified crime problems, and we provide them with the discretion and authority to practice their considerable crimefighting skills and experiment with new methods and tactics in fighting crime. These policing skills were always present but usually underused. Street cops have always said they had the ability to reduce crime if the agency's executives would only relieve them of the constraints imposed by an unimaginative and timid management cadre. At the NYPD, we did remove many of these constraints without sacrificing discipline or our command and authority over police officers' behavior. In New York, random preventive patrol is a thing of the past because we've given our officers better and more productive things to do with their time. The time they once spent aimlessly driving or walking the streets is now devoted to tactical strategic enforcement activities. I would be remiss to leave you with the impression that the absolute number of officers deployed in the field is of little consequence. In fact, the number of officers deployed is an essential ingredient in this formula, but it is probably less important in terms of reducing crime than the manner in which officers are deployed. Certainly, we require a sufficient number or "critical mass" of officers to make our crime strategies effective and workable, but we could probably do with fewer officers if we could significantly reduce the amount of time they devote to purely reactive policing and increase the amount of time they spend in a proactive enforcement mode. At the same time, we cannot ignore the fact that visible police patrol leads to a heightened public sense of safety and security. Making people feel safer is an important police function, and a certain amount of police time and personnel will always be devoted to that purpose. In the traditionally managed, reactive agencies, police work often followed a set of contradictory, or at least conflicting, operating principles. Officers were deployed in reaction to crime trends and patterns that might, at best, be several weeks or months old. And yet, as part of O.W. Wilson's legacy, many police executives displayed a near obsession with shaving seconds off the response time to 911 calls about crimes that had already occurred. Although they were given a long list of rules intended to govern their behavior, police officers in reactive agencies operated virtually unsupervised, with little meaningful management oversight of their specific activities. These officers were, in effect, set loose on the streets without the benefit of coordinated and integrated tactical strategies. Police officers and executive alike shared a rhetoric and a sensibility that "real police work" involved fighting the "serious" crimes of robbery, burglary, larceny, assault, rape, and murder, to the exclusion of less important quality-of-life offenses. Yet few agencies developed strategies to deal with these crimes in their totality as opposed to dealing with them on a crime-by-crime and case-by-case basis. And few recognized that the failure to enforce quality-of-life laws was sending a message of lax police enforcement and encouraging the commission of more serious crimes. As described earlier, the NYPD now has the technological capacity to identify crime patterns almost immediately, and our response can be virtually contemporaneous with evolving patterns. We also have significantly tightened our management controls over police activities, empowering officers and commanders at the local level while holding them accountable for their crimefighting results. Officers and com-manders are now guided by comprehensive and coordinated strategies and tactical plans that provide enough flexibility to permit the crafting of appropriate site-specific responses. We relentlessly follow up on their activities to ensure that problems are solved rather than displaced. We have also recognized and embraced the wisdom of Wilson and Kelling's "broken windows" theory and its emphasis on the criminogenic nature of quality-of-life offenses (1982). We have convinced officers and commanders that serious crime as well as public fear of crime can be reduced by tending to these "minor" offenses and annoyances of urban life. The NYPD circa 1995 is a very different agency than the reactive organizations that previously characterized American policing, and it is achieving very different results. The assumption that all police departments can provide only a "last-resort, reactive mechanism" is in need of thorough study and evaluation. A new kind of police department is emerging--a flexible, responsive, focused organization that can swiftly identify new crime patterns and just as swiftly counter them. It is time for the discipline of criminology to recognize the change. To compare the old reactive agencies to the NYPD circa 1995 is to compare apples and oranges. I turn now to the main hypotheses, inferences, and research data that make up the view that crime is primarily pulled by social and demographic engines. Let's look at how these theories are challenged by empirical facts in New York City's contemporary crime picture. Age, demographics, and crime The relative size of the cohort between 15 and 21 years of age historically has been shown to have enormous influence on the rate of reported crimes. Criminologists have clearly demonstrated that adolescents commit a disproportionate number and percentage of total crimes, criminality peaks between the ages of 16 and 20 for the majority of specific offenses, and the rate of offenses attributable to a particular age cohort declines as the cohort ages (Hirschi and Gottfredson, 1983; Wolfgang et al., 1972; Tracy et al., 1990). These conclusions are supported over time by the UCR data as well as by victimization studies. It should also be noted that individual criminologists define such important variables as "youth" and "youth crime" differently, which complicates the comparability of their research. By slightly altering the operational definitions used to collect data sets or altering the upper and lower limits used to categorize an age group, for example, substantially different results might be obtained. Despite these caveats, official data and criminological research do reveal that the rate at which adolescents and young adults commit crimes is three to five times higher than their proportional representation in the general population. They account for a disproportionate number of arrests as well. In particular, the highly credible cohort research conducted by Marvin Wolfgang and his colleagues ( Wolfgang et al., 1972; Tracy et al., 1990) found that about one-third of both Philadelphia birth cohorts they studied had been arrested by age 18 and one-half had been arrested by age 30. These results support the general observation that the number of male adolescents in a population will have considerable impact on levels of crime. Between 40 and 50 percent of the increase in crime index offenses during the 1960s, for example, is attributed to the "baby boom" generation. Arrest data from New York City also show the heightened criminality of adolescents aged 15 to 19. Between 1980 and 1994, for example, the average annual robbery arrest rate for young people between 15 and 19 (17.38 per 100,000 population) was more than five times higher than for the population as a whole (3.29 per 100,000) and nearly double that of the next closest age group (20 to 24, 9.20 per 100,000). In 1994, this cohort accounted for more than 37 percent of all robbery arrests in New York City, almost four times the percentage for the population as a whole (9.47) and almost two-and-one-half times the percentage for the cohort aged 20 to 24 (15.7). The age 15 to 19 cohort clearly accounts for a disproportionate number and percentage of robberies, and generally similar relationships can be discerned by examining complaint and arrest data for other specific offenses. When robbery arrest trend data from 1980 through 1994 are examined, however, a somewhat different picture emerges. Although the age 15 to 19 cohort has consistently accounted for the greatest proportion of robbery arrests, that proportion in New York City has declined over time--from 47 percent in 1980 to 37 percent in 1994. This cohort's share of the total robbery arrests declined steadily between 1980 (47.0 percent) and 1987 (30.8 percent), when it began to climb upward by one or two percentage increments per year. Criminology's conclusions about the influence of the age 15 to 19 cohort on overall crime may have been historically accurate, but they no longer seem to apply in New York City. The city's youthful population declined during the two decades from 1970 to 1990 when crime rates soared in New York City and across the Nation. The group between 15 and 19 declined by almost 22 percent in New York City during this period, but the proportion of the cohort involved in crimes increased enormously. Per capita arrests for youths between 15 and 19 increased almost 60 percent between 1970 and the early 1990s. During this period of significant decline in the city's high-risk youth population (between 1970 and 1990), total index crimes increased by 22.8 percent--from 578,149 index crimes in 1970 to 710,221 in 1990. Both homicide and motor vehicle theft hit 20-year peaks in 1990. But as New York City crime started to decline in the 1990s, the decline in youth population reversed itself. Based on its analysis of the 1990 U.S. census, the Department of City Planning estimates that the city's population of youths between 15 and 19 years of age has increased slightly between 1990 and 1995. Most significant, especially for criminologists who consider race as a variable, the number of black males between 15 and 19 is estimated to have increased by nearly 2 percent and the number of male Hispanic youths by 5.7 percent. Asian and Pacific Islander males between 15 and 19 also increased an estimated 2.36 percent. Pulling the average for the entire cohort down were the white males whose numbers decreased 8.4 percent. These data are confirmed by New York State Department of Education school enrollment figures for the City of New York, which show that total public school enrollment increased 4.4 percent between the 1989-90 and 1994-95 school years. The number of public school students in grades 9 through 12, comprising a significant portion of the high-risk group aged 15 to 19, increased by 12 percent. The demographic rationales for crime and their emphasis on criminality among the cohort of males between the ages of 15 and 19 cannot explain the significant crime reductions in New York City over the past several years. These rationales would, in fact, predict the opposite effect. The demographic data provided here point to the indisputable, if theoretically inconvenient, reality that the number of individuals who have historically been shown to account for a disproportionate amount of crime relative to their percentage representation in the overall population was relatively low during the late 1980s when New York experienced a rise in crime, and that that number has actually increased between 1990 and 1995, when New York City began to realize a notable decrease in crime. Drugs and crime A great deal of recent discourse and research in contemporary criminology has focused on the nexus between drug abuse and crime, particularly violent crime. Hypotheses typically establish a causal link between drugs and crime in two ways: (1) The physiological effects of a particular drug are said to induce violent crime through the removal of inhibitions or other pharmacological effect. (2) The prohibitive cost of some drugs is said to cause users to commit crimes, particularly property crimes, to generate sufficient income to satisfy their addiction. Of central concern to the "drugs cause crime" hypothesis is the question of which variable comes first--do individuals become addicted and then commit crimes, or do criminals begin to use drugs after their criminal careers have begun? It is my understanding that this empirical question remains unresolved despite a quantity of research. Nevertheless, positive correlations between drug use and criminality have been demonstrated, despite the fact that many of the studies are based on convenient samples of prison and jail inmates and therefore present the problem of sample bias (Bureau of Justice Statistics, 1988, 1991). Another empirical issue is the difficulty in determining what portion of overall crime is committed by drug abusers. As Wilson and Herrnstein (1985: 366) pointed out, it is virtually impossible to calculate how much crime heroin addicts commit even if there are accurate data about the number of addicts and the monetary costs of their addiction. Criminologists seek to explain fluctuations in crime rates by pointing out how variations in drug markets and drug-abuse patterns have historically correlated with crime trends. Specifically, some have argued that the precipitous increases in robbery complaints experienced nationwide during the late 1980s were attributable to the emergence of crack cocaine, a drug that has been intuitively and anecdotally linked to higher rates of crime. Crack cocaine exploded onto the drug scene in New York City in 1985 and 1986, a period in which robbery complaints did in fact increase dramatically. Based on the concurrence of these historic trends and a general tendency to infer causation from mere correlation, many criminologists would conclude that New York City's increase in robberies during the late 1980s was driven by the advent of crack. Conversely, those criminologists would tend to conclude that New York City's recent decline in robberies signals a dramatic reduction in crack addiction and use. Some would argue, in a similar vein, that the supposed reemergence of heroin as the drug of choice among street criminals might translate into an increase in burglary complaints because burglary rates have long been associated with or attributed to the extent of heroin addiction. Unfortunately for these criminologists, however, neither of the hypotheses is supported by the current empirical evidence in New York City. In 1984, just prior to the crack explosion, the first NIJ-sponsored Drug Use Forecasting (DUF) urinalysis study at the NYPD Manhattan Central Booking facility revealed a 42-percent positive rate for cocaine among all arrestees sampled, irrespective of charge. By 1988--perhaps the height of the crack epidemic--the prevalence of cocaine use among all arrestees had nearly doubled to 83 percent, lending credibility to the hypothesized relationship between crack cocaine and crime. Although a decline has been recently noted in cocaine use among all arrestees, it has been fairly modest. In February 1995, 78 percent of arrestees tested positive for cocaine, and in May 1995 (the most recent quarterly data available), 68 percent tested positive for cocaine. These quarterly data fall within the typical range of variance for positive cocaine tests. Since 1988, the proportion of arrestees testing positive for cocaine in each quarterly sample varied from 59 percent to 83 percent, and since 1993, the proportion of positive cocaine tests varied from 63 percent to 78 percent. Cocaine use among those arrested in New York City has not declined substantially, certainly not to the extent that declining cocaine use could account for the enormous decline in the crime, particularly violent crime, that cocaine supposedly engenders. The hypothesized increase in heroin abuse has not been evident in the quarterly DUF data either. In 1984, 21 percent of arrestees tested positive for opiates; positive tests peaked at 27 percent in June 1988 and 25 percent in October 1988. In the most recent DUF testing quarters, February and May 1995, 22 percent and 20 percent of arrestees, respectively, tested positive for opiates. Narcotics enforcement activity data also provide indirect evidence that drug abuse has not diminished significantly. In 1994, total arrests for narcotics offenses in New York City increased 28.9 percent, reaching their highest point since 1989. Felony drug arrests rose 11.4 percent in 1994, and misdemeanor drug arrests rose 54.2 percent. Through November 12, 1995, total NYPD narcotics arrests increased 10.14 percent over the comparable 1994 period and 39.06 percent over the comparable 1993 period. Although this increase is clearly due to our heightened enforcement and the strategic approach we are taking to address the city's narcotics problem, and although arrest data cannot be taken as conclusive evidence of the prevalence of drug abuse, these numbers provide a rough indicator that drug abuse remains pervasive. Firearms use Without engaging in the contentious and ongoing debate about gun control and the right of citizens to possess firearms, one can intuitively grasp a connection between the availability of guns, particularly handguns, and violent crime. Guns are certainly more lethal than other weapons used in the commission of crimes, and it is a reasonable assumption that gun availability facilitates the commission of many crimes. Roughly half of the Nation's homicides are committed with guns, and guns are used in about one-third of all robberies and one-third of all rapes. I won't address the question here of whether guns cause crime in the sense of serving as a catalyst for the escalation of violence or if they deter crime when they are in the hands of law-abiding citizens. It is scarcely debatable, however, that a large number of criminals have carried and used guns in the commission of their crimes or that, in the case of New York City at least, the vast majority of these guns are illegally possessed. The number of firearms, especially handguns, used in criminal activity has declined substantially in New York City during the past 2 years. The data supporting this conclusion are derived from several sources, each of which confirms the observation that fewer criminals are carrying and using guns. The percentage of robberies in which firearms were used, for example, fell from 36.3 percent in 1993, to 33.05 percent in 1994, to 28.7 percent for the first 6 months of 1995. The total citywide number of shooting incidents between January 1 and November 12 fell 39.67 percent between 1993 and 1995, and the number of shooting victims injured in these incidents fell 37.62 percent. The decline in firearms use can also be inferred from the declining number of calls reporting "shots fired" to our 911 system. The department received 23 percent fewer shots-fired calls from citizens and dispatched 12,353 fewer radio cars for these calls in the first 9 months of 1995 than it did for the comparable 1994 period. The declining number of shooting incidents and shooting victims reflects a general decline in the number of firearms being carried and used by criminals, which we attribute to the effectiveness of our strategic gun enforcement efforts. We are hard pressed to conceive of any demographic or social variable that might induce street criminals to refrain from carrying or using their guns. Although the total number of gun arrests for the year-to-date period through November 12 declined 34.8 percent from comparable 1993 levels, we do not claim to have taken all of these guns off the streets or away from criminals. We merely assert that criminals have considered the wisdom of leaving their guns at home. Indeed, our gun arrests increased fairly rapidly subsequent to the introduction of our gun strategy and then began to decline as a function of the aggressive enforcement. It should also be noted that implementation of our strategy seems to have had the unanticipated consequence of promoting the use of other, but fortunately less lethal, weapons. The number of arrests for nonfirearm dangerous weapons increased more than 6 percent during the 1993 to 1995 year-to-date period. The following example illustrates one creative way of approaching the problem of illegal guns. Our research and investigations showed that unscrupulous private gun dealers holding Federal firearms licenses (FFLs) were a major source of illicit guns on New York City's streets. In March 1993, we began to jointly review FFL applications from New York City residents with the Bureau of Alcohol, Tobacco and Firearms. Of the 238 new applications received through December 1994, 97.4 percent were disapproved. In addition, 71 percent of the renewal applications between August 1993 and December 1994 were abandoned, surrendered, or disapproved in the face of increased police scrutiny. Although we cannot quantify the extent to which this policy actually reduced the availability of illegal firearms and handguns, we believe that it is certainly a contributing factor. Social and economic factors Whether or not poverty causes crime has been one of the most controversial and enduring issues in criminology and the political arena. Academic research efforts have failed to provide conclusive data to support or reject any of the common economic theories of crime causation. Arguments over the role of poverty and other economic factors tend to follow the lines of political ideology and are largely based on rhetoric and intuitive reasoning. Wilson and Herrnstein (1985) pointed out that the presumed connection between unemployment and crime is rather tenuous. They said the empirical research in this area is inconclusive and noted several logical faults within the competing theoretical models that seek to link unemployment and crime. In any case, none of the common social or economic factors that criminologists typically cite to explain fluctuations in crime has registered changes of sufficient magnitude in New York City to suggest they are responsible for any appreciable decline in crime. New York City's economic picture has improved slightly over the past several years, but those years cannot be accurately characterized as a boom period or even as a period of significant growth. Monthly data from the U.S. Department of Labor show New York City's unemployment rate at 10.8 percent in January 1994, 7.2 percent in September 1994, 9 percent in February 1995, and 8 percent in September 1995. Throughout the 2-year period, the city had a higher unemployment rate than the Nation. A comparison of the New York City Human Resources Administration's July 1994 and July 1995 public assistance rolls reveals that the number of city residents receiving public assistance benefits declined by 45,354, or fully 4 percent. A comparison of the number of city residents receiving food stamps in August 1994 and August 1995 reveals a very modest decrease of 0.4 percent. Certain other indicators, however, seem to show a return of confidence in the safety of the city. In time, we might see an improvement in the city's economy following a decline in crime rather than the other way around. The New York City Convention and Visitors Bureau estimates that the city will welcome more than 25 million visitors in 1996, a 14-percent increase over 1995 levels. This translates into 3,500,000 more visitors who contribute to the local economy. New York City's hotel occupancy rate rose from 71.7 percent during the first 6 months of 1994 to 74.2 percent during the comparable 1995 period. Overall airport arrivals rose 2 percent, and international arrivals rose 7.4 percent. Attendance at Broadway shows rose 14.1 percent, and the number of visitors served by the Convention and Visitors Bureau increased by 5.1 percent. Similarly, subway ridership has mirrored the decline in subway crime. Daily subway ridership fell 3.5 percent between 1990 and 1991, but it increased 0.2 percent between 1991 and 1992 when subway crime fell 15 percent. In 1992 and 1993, when subway crime fell an additional 24.3 percent, daily ridership rose 5.1 percent. In 1994, with subway crime falling another 21.7 percent, ridership increased an additional 5.2 percent. From these data we can infer that public fears associated with riding the city's rapid transit system have declined and residents and commuters are increasingly willing to travel freely throughout the city using public transportation. Prison and jail populations, arrests, and incapacitation Even the best-managed, most effective, and most highly directed police agency cannot reduce crime solely through arrest and enforcement. Other spheres of the criminal justice system--the courts and corrections, probation, and parole functions--take responsibility for an offender once he or she is in custody, and each plays a salient role in reducing crime and enhancing public safety. Corrections agencies in particular are instrumental in reducing crime through incapacitation and perhaps to some extent through deterrence, although the importance of the correctional role rarely receives much attention in the public discourse on crime. Like each of the other spheres of the criminal justice system, the view of correctional agencies is subject to prevailing political and organizational ideologies. During the 1960s when national crime rates tripled, correctional policies and practices were driven to a large extent by the rehabilitative ideal. We did not conclude until the 1970s that, in terms of rehabilitation, "nothing works" (Lipton et al., 1974; Martinson, 1974). In the 1980s and 1990s, the ideology of incapacitation has come to the fore. Although it may be difficult to accurately estimate the relative effectiveness of incapacitation strategies, the rationale for incapacitation is fairly simple. We know that some criminals, particularly "career criminals," commit a highly disproportionate number of criminal offenses. Blumstein and his colleagues have noted that the most active 10 percent of offenders each commit in excess of 100 crimes per year (Blumstein et al., 1986: 94). The clear implication is that drastic reductions can be made in the overall crime rate if this group of high-rate chronic offenders is incapacitated. As discussed above, cohort research on youth crime (Wolfgang et al., 1972; Tracy et al., 1990) also reveals that a relatively small percentage of young people are responsible for a vastly disproportionate share of offenses. Statute law and the ideology of the juvenile justice system preclude sentencing youthful offenders with the same severity directed toward adult criminals. But it also stands to reason that significant inroads can be made in the overall crime picture if we implement some sort of realistic intervention to discourage criminals at the early stages of an evolving criminal career. Too often in the past, police and juvenile courts have not treated youth crime seriously enough. Both police and courts have operated on the assumption that it is not in children's best interest to burden them with criminal records. Many police officers have failed to take appropriate discretionary action in cases involving young people, possibly in the cynical belief that juvenile court authorities would, at best, merely give the juvenile offender a "slap on the wrist." It should be no surprise, then, that many young people who have had contact with the juvenile justice system learn that their offenses will not be taken seriously. For the small percentage of feral youth whose contacts with police and courts are frequent, this perception is repeatedly reinforced. Some are genuinely surprised when the criminal court system finally imposes a real sentence. An article in the Detroit News described New York City's tremendous drop in crime and speculated whether the strategies and tactics the New York City Police Department pursued would have a beneficial effect in Detroit. The article also noted that criminologists were skeptical about the role of the NYPD's strategic approach in achieving these reductions as well as the credit police deserve for them. One criminologist was quoted as saying that police do not control any of the things that generate crimes: "[Cops] don't control the demand for drugs. They don't control who's on welfare and who's not. They don't control who has a job and who doesn't. They don't control what Republicans like to call Ôfamily values'" (Tobin, 1995: A3). This is a fair and accurate assessment. The police do not control these broad social and economic factors. But the same criminologist went on to explain why, in his opinion, crime had declined so precipitously in New York City: "The bad guys are in jail," he said. "Even a small number of crooks taken off the street can make a big difference in crime statistics." Who, if not the police, put them there? For the year-to-date period ending November 12, 1995, the total number of arrests for all criminal offenses in New York City--felonies and misdemeanors--increased 26.73 percent over 1993 levels for the comparable period. Arrest for combined index crimes--all felonies--increased 4.27 percent. The disparity in these data demonstrates the effectiveness of the department's shift away from limiting emphasis on the traditionally "serious" index offenses committed by adults toward strategic enforcement of appropriate and applicable laws, and it provides evidence of the efficacy of the "broken windows" theory. By increasing enforcement--as measured through arrests--for misdemeanor quality-of-life offenses among adults and young people, we were able to achieve enormous reductions in felonies, particularly index crimes. Not all of those arrestees were incapacitated through incarceration. Although a large percentage of the 3.4-percent increase in New York State's prison population between 1993 and 1994 is attributable to arrests from New York City, it must also be noted that both admissions to and releases from State prisons declined in 1994. Admissions fell by 3.4 percent and releases by 1.8 percent. Fewer criminals are being incarcerated, but they are being incapacitated for longer periods. The increase in arrests, especially misdemeanor and juvenile arrests, did not impose an untenable burden on our jail system. In fact, the city's average daily jail population actually fell 1.2 percent between 1993 and 1994, after rising in both 1991 and 1992. For the first 9 months of 1995 versus the comparable 1994 period, the average daily jail population fell by 5.9 percent, from 19,558 inmates to 18,397 inmates. The inference to be drawn from these data is that dramatic crime reductions can be achieved through the sustained and tactical enforcement of quality-of-life misdemeanor offenses, coupled with vigorous enforcement of "serious" felony crimes and the concomitant incapacitation of "career criminals." Summary The magnitude and direction of change among the various socioeconomic and demographic variables reviewed here lends little credibility to traditional criminological conceptions about the causes of crime and crime reduction. Indeed, given the direction and magnitude of change evident in many of these variables, traditional criminological thought might have predicted increases in crime in New York City rather than the significant declines we have actually experienced. A third intervening variable--a well-managed and highly directed police agency--provides a better explanation for the decline in New York City crime than any of the traditional explanations cited by criminologists. Note 1. For a good account of Compstat meetings, see Kelling, George, "How to Run a Police Department," City Journal, Autumn 1995. References Blumstein, Alfred, Jacqueline Cohen, Jeffrey A. Roth, and Christy A. Visher, eds. Criminal Careers and "Career Criminals." Washington, DC: National Academy Press, 1986. Bureau of Justice Statistics. "Drug and Crime Facts, 1991." Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics, 1991, NCJ 134371. DiIulio, John J. "Why Violent Crime Rates Have Dropped." The Wall Street Journal, September 6, 1995, A19. Hirschi, Travis, and Michael Gottfredson. "Age and the Explanation of Crime." American Journal of Sociology 89 (November 1983): 552. Innes, Christopher A. "Drug Use and Crime." Special Report. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics, 1988, NCJ 111940. Lipton, Douglas, Robert Martinson, and Judith Wilks. The Effectiveness of Correctional Treatment: An Empirical Assessment. New York: John Wiley, 1974. Martinson, Robert. "What Works? Questions and Answers About Prison Reform." The Public Interest 35 (1974): 22-54. Tobin, Jim. "Can Detroit Bite Into Crime Like Big Apple?" The Detroit News, September 18, 1995, A1, A3. Tracy, Paul E., Marvin Wolfgang, and Robert M. Figlio. Delinquency Careers in Two Birth Cohorts. New York: Plenum, 1990. Walker, Samuel. "ÔBroken Windows' and Fractured History: The Use and Misuse of History in Recent Police Patrol Analysis." Justice Quarterly 1 (1984): 57-90. Walker, Samuel. Sense and Nonsense About Crime: A Policy Guide. Monterey, CA: Brooks/Cole, 1985. Wilson, James Q. Thinking About Crime, revised edition. New York: Random House, 1985. Wilson, James Q., and Richard J. Herrnstein. Crime and Human Nature. New York: Simon and Schuster, 1985. Wilson, James Q., and George Kelling. "Broken Windows." Atlantic Monthly (March 1982): 29-38. Wilson, O.W. Police Administration, 2d ed. New York: McGraw-Hill Book Company, Inc., 1963. Wolfgang, Marvin E., Robert M. Figlio, and Thorsten Sellin. Delinquency in a Birth Cohort. Chicago: University of Chicago Press, 1972. ------------------------------- Measuring What Matters: A New Way of Thinking About Crime and Public Order George Kelling [This article is reprinted with permission from City Journal.] Here is a public policy paradox: New Yorkers are frantic over what seems to them the increasing lawlessness of the city. Crime and fear are consistently among the top two or three reasons cited by New Yorkers who say they want to leave town. Yet according to professional standards and the most common statistical measurements, the New York City police departments are among the best in the country, especially after taking into account their size and the problems they face. For generations, police have tried to develop a model of policing that is equitable, accountable, efficient, lawful, and honest. They have largely succeeded: In the quest for equity, police are distributed across cities on the basis of crime rates and calls for service--seemingly objective criteria. To be unobtrusive, police have relied on responding to citizens' calls for help, rather than initiating action on their own. To ensure lawfulness, police have focused their resources on serious crimes--murder, rape, assault, robbery, and burglary--acts prohibited by unambiguous laws and about which a broad consensus exists that police should take strong action. To ensure honesty, police have limited contacts with possible sources of corruption, including citizens. By these measures, New York City is excellently policed: Its departments, especially the New York City Police Department, distribute police equitably throughout the city, respond quickly to 911 calls (especially considering the enormous volume here), are unobtrusive (despite rare and highly publicized exceptions), have concentrated on serious crime, and maintain high levels of integrity. Among professionals, the NYPD is widely believed to be one of the "cleanest" very large departments in the country. Even by more widely touted measurements, New York police do relatively well; so many people have been arrested that neither jails nor prisons can hold them. If the number of cells was expanded, few doubt that New York City police could fill almost any added capacity as well. Crime rates are also encouraging, at least compared to other large cities. In 1989, eight large American cities had higher homicide rates than New York City, 21 had higher rape rates, 17 had higher burglary rates, and eight had higher automotive theft rates. The differences were not trivial: Washington's murder rate was almost 2.8 times as high as New York's; Cleveland's rape rate 3.5 times higher; Dallas's burglary rate twice as high. Only in robbery did New York lead the nation, and not by much. But New Yorkers are not the least bit reassured by these statistical and relative achievements. One prominent local political leader eager to discover his constituents' concerns recently gathered some New Yorkers in "focus groups" to discuss major issues. When he asked them to react to the statement "New York City is tough on crime," their response was incredulous laughter. The citizens are right. These formal measures of police work have little to do with community needs. After all, even after decades of increase, individual serious crimes remain relatively rare. But if a typical annual increase in the mugging rate does not materially increase the chances that one will be mugged, neither does a similar decrease reduce the real harm done to those who are not mugged--which is to make them afraid and cheat them out of a little bit more of their lives. Lawlessness consists not just in the relatively rare "index" crimes counted by the FBI, but can also refer to an atmosphere of disorder in which it seems like these and less serious crimes and harassments might occur at any time. Lawlessness locks neighbors behind doors, chases storeowners off streets, shuts down business, and spreads poverty and despair. Still, twice a year when the official FBI crime statistics are released and the Times announces, "New York Leads Big Cities in Robbery Rate, but Drops in Murders," and the Post and the News chip in with their more-colorful versions, police officials frantically counter with their own numbers that show how well they are doing. Even now, when "community policing" (which is supposed to deemphasize statistics) is all the fashion, police chiefs know that every time the ritual is repeated, the political powers-that-be will call them on the carpet and the powers-that-would-be will call press conferences. Police strategy, tactics, and even police mythology and esprit de corps are driven by statistical and bureaucratic measures of performance. The result is disastrous for the community. Ironically, the statistics police find most nettlesome, the FBI's Uniform Crime Reports, were invented by The International Association of Chiefs of Police in the 1920s. The original UCR index consisted of seven crimes: murder, rape, robbery, aggravated assault, burglary, larceny, and motor vehicle theft. In 1979, arson was added to the list. The UCR also include data on crimes cleared (someone was arrested), on the people who were arrested, and on law enforcement personnel. Victimization surveys supplement the UCR by providing additional information about victims and offenders in crimes which may never have been reported. Once chiefs had high hopes for the UCR, believing that reported crime and clearance rates would provide "scientific" measures of the nature and extent of serious crime and of the relative effectiveness of police departments. And during the comparatively quiet years of the Forties and the Fifties, police were quick to claim credit for the relatively low reported crime rates. In the Sixties, this honeymoon ended. Crime levels, in the statistics and in the minds of citizens, became intolerable. As the crisis worsened and became a bigger national story, the UCR framed the problem for the media, the general public, and therefore for politicians and police as well. The crime problem was reduced to the seven crimes on the index; important crime-control activities were clearances and arrests for index crimes. Police departments, broadsided biannually with bad news, became obsessed not only with statistics, but also with statistical responses. They pointed with pride to figures showing that arrests were up, response times were faster, police were working hard, and criminals were going to jail. And by all these quantifiable standards, their departments were indeed going well. If crime still raged after such prodigious efforts, it could hardly be the fault of the police. Better to blame lazy prosecutors, lenient judges, push-over probation officers. And don't forget the liberals. Got a problem, buddy? Tell it to Earl Warren. If it had only been a dodge for the press and the pols, it would not have been so bad. Unfortunately it is hard to say things too often without coming to believe them, and in any event bureaucracies of all sorts love numbers, which hold out the promise of order and accountability, a way of toting up the score at the end of the game. Unfortunately crime, arrest, and response reports not only fail to keep an accurate score, they also confuse everybody about the object of the game. While low levels of recorded crime may conceivably reflect low crime rates, they can also reflect a lack of confidence in police. It is well known, for instance, that about half of all rapes are ever reported to police. Women fail to report rapes because of embarrassment, fear, and guilt--emotions that depend in part on how police agencies handle rape victims and their cases. So what does the difference between Cleveland's and New York's rate mean? Is it true that there are more rapes in Cleveland than in New York? Are New York police to be credited with being more efficient? Or are women in Cleveland more confident that they will be treated sensitively by police and other criminal justice agencies in Cleveland? What about burglary? Does Dallas have more burglaries than New York? Perhaps. But another explanation is that burglary victims in New York City have simply come to expect so little from police that they often do not report the crime. The UCR's stiff legal categories say little about the crime problem as citizens actually experience it. The popular conception is that serious crimes are acts committed by ruthless predators against innocent strangers. In 1989, however, more than 40 percent of violent crimes, including one-third of all rapes, were committed not by strangers, but by friends, lovers, spouses, and colleagues. Within families and relationships, abuse can be repeated over and over with increasing ferocity and suffering. Society has an enormous investment in the institutions in which these victimizations occur: family, schools, the workplace, just to mention three. For communities, the intent of crimes often is more important than the actual crime itself. Generally, we consider vandalism a relatively minor crime, often committed by obstreperous youth. It does not show up on the UCR. Yet a swastika painted on the door of a Jewish home or a cross burned in front of a black family's home often has more serious consequences than a random robbery or burglary. Such vandalism demoralizes communities, destabilizes neighborhoods, and terrorizes families. Arrest counts are no more reliable than the UCR. Consider the following: An officer sees a dispute between a Korean merchant and a black citizen. The officer stays at a distance observing the dispute. It flares into violence. The officer moves in to stop the violence and proceeds to arrest both of the citizens. Tensions increase in the neighborhood, but two arrests are chalked up for the officer. Is this a success? Should the officer and department be credited for this performance? Or were the arrests really indications of failure? Would it not have been better to intercede earlier and prevent the violence that not only threatened the individuals' well-being, but the community's peace? Obviously. And in such a situation most New York City police officers almost certainly would have done the right thing. Yet it is important to note that if the officer had stepped in to defuse the incident, perhaps sparing the community months of anguish, his action would never have been recorded. That suggests a serious problem, not only in providing recognition for officers, but also in keeping the department accountable to the community and focused on its real needs. Likewise, consider the much-studied problem of graffiti on subway trains. For over a decade, while police had been unable to reduce subway graffiti, arrests for graffiti increased year by year and were touted by the Transit Police Department whenever it was queried about the problem. Then Transit Authority President David Gunn instituted a successful program to eliminate graffiti--a program based not on arrests but on quickly cleaning cars and painting over graffiti so as to frustrate the "artists" and create the impression that the TA [Transit Authority] took the antigraffiti rules seriously. Arrests immediately dropped and stayed at a low level throughout the five-year effort. The earlier volume of arrests had indicated failing policy, not success. If the volume of arrests says little about the effectiveness of police performance, another favorite set of police statistics, the number and speed of responses to emergency calls, are equally uninformative. The anticrime potential of 911 was once thought to be quite high. Research and experience, however, have suggested that though rapid responses to calls for service have very limited impact on crime, they consume enormous amounts of police time. This view is now widely shared by police and police scholars, although less so by city policymakers and politicians, for whom 911 has become a symbol of being "tough on crime." Former Police Commissioner Ben Ward put the trade-offs starkly at a meeting of community leaders, one of whom complained, "We have our neighborhood foot patrol officer, we now want rapid response to calls for service." Ward's response was refreshingly frank: "You can't have both." As I have previously noted, since the 1960s, research has confirmed that crime, as well as the fear of crime, is closely associated with disorder. Disorder includes petty crime and inappropriate behavior such as public drunkenness, panhandling, and loitering; its physical manifestations include graffiti, abandoned cars, broken windows, and abandoned buildings. For most people, New York's crime problem comes down to the fear they endure as a consequence of disorder--the well-founded belief that in disorderly places society has ceded control to those who are on the margin of or outside the law, and therefore that anything might happen in such places. I say this belief is well-founded because both experience and substantial formal research demonstrate that disorder left untended ultimately leads to serious crime. Citizens' fear of disorder is entirely rational. Fighting disorder, by solving the problems that cause it, is clearly one of the best ways to fight serious crimes, reduce fear, and give citizens what they actually want from the police force. Yet disorder and police efforts (or lack thereof) to eliminate it have recently been largely ignored by official police doctrine. The reasons for that are many and complex, ranging from the belief that uncivil, threatening, and bizarre behavior is a constitutional right, to fears created by past police abuse of statutes prohibiting disorderly behavior. But a significant reason disorder has been ignored is that professional criminal justice ideology narrowly defines the appropriate business of police and criminal justice agencies as dealing with serious crime--that is, index crimes. Crime, response, and arrest statistics form a pillar of that ideology. Disorder does not appear on any FBI index; therefore, it has not been a priority. Community policing, which is being put into place in this city [New York] slowly and with considerable difficulty, is supposed to take disorder seriously. But community policing itself is hampered by the tools police use to measure the crime problem and police performance. There is a great gap between the current bureaucratically defined measures of productivity and the kinds of help communities really want from their police. Levels of fear and disorder, evidence of mounting community tension, and, most importantly, information about the specific sources of such difficulties and police response to such problems, go officially uncounted. Can we develop new measures of performance, measures more in line with what communities really need and want? Can we quantify the "soft" indicators that really matter to communities? Or are we doomed, like the man who lost his keys in the alley but searches for them under the street light, to keep looking in the wrong place because it is too hard to turn our attention where it belongs? During the 1980s and into the 1990s, a series of independent studies tried to define New York's real crime problem. Citizens, neighborhood groups, business associations, and others examined community problems, at times in collaboration with police and criminal justice officials, but often without any official support. With remarkable consistency, the studies tell us what citizens want government to do. Implicitly, and in at least one case explicitly, they tell us how to measure community crime problems and police response. One of these studies, "Downtown Safety, Security, and Economic Development," was published by the Citizen's Crime Commission of New York City and the Regional Plan Association in July 1985. As Laurence A. Alexander wrote in the preface: Working with both city officials and with developers, it was clear that many private and public downtown investment decisions were being killed by underlying nagging worries over the safety and security of people and of investments. At the same time, I saw many studies that showed downtowns were not necessarily high-crime areas (especially not with respect to so-called serious crimes). But, nevertheless, shoppers, workers, bosses, and bankers were all convinced that crime was rampant downtown. It was very clear that this problem--to some degree real and to some degree a matter of perception only--was a major deterrent to rational downtown planning, development, marketing, and management. The report went on to document fear of crime in downtown Brooklyn, Fordham Road in the Bronx, and Jamaica Center in Queens. The results were stark: Almost 60 percent of those surveyed believed that if they went to these areas their car would probably be stolen or broken into; 40 percent believed that they would be attacked, beaten, or raped; and 75 percent believed that they would have their money, wallets, or purses stolen. Confirming earlier research, the study found strong correlations between levels of fear in the area and the amount of drug use and sale, public drinking, street gangs, loitering teenagers, and graffiti. The consequences of fear were considerable: People stayed off the streets and avoided public transportation and "multi-purpose visits" (that is, shopping). While "Downtown Safety" documented citizens' fears about shopping in commercial centers, a report called "Small Business, Big Problem," published in May 1989 by the New York think tank Interface, focused on the impact of the crime problem on commercial establishments. The organization surveyed 353 small businesses--retailers, service companies, manufacturers, and wholesalers with an average of 27 employees--in the Bronx, Brooklyn, Manhattan, and Queens. Direct losses from crime, especially from break-ins, vandalism, shoplifting, and auto thefts, were high. More than 80 percent of the firms reported being victimized during the previous three years. Crime, and the fear of crime, also took an indirect financial toll on those firms in the form of increased labor costs from high employee turnover, reduced sales, and curbed expansion plans. The neighborhood conditions most often cited as causes of these difficulties included loitering, drug dealing, panhandling, illegal peddling, and in manufacturing and wholesale areas, prostitution. Thus, even in an area where indexed crimes were a serious part of the problem, merchants specifically cited disorderly conditions as a major difficulty and were able to point to consequences. The section of the report devoted to solutions specifically recommended measures usually associated with maintaining order and reducing fear--foot patrols, community policing and neighborhood watches. Another study, "CPOP: the Research--An Evaluative Study of the New York City Community Patrol Officer Program," published by the Vera Institute of Justice in 1990, offers insights into the problems of primarily residential neighborhoods. Their analysis of a set of reports by CPOP (community policing) officers and a survey of community leaders is particularly interesting. CPOP officers used "Beat Books" to record the types of problems with which they dealt. The problems that citizens complained about most often were drugs (29 percent), parking and traffic (16 percent), disorderly groups (14 percent), auto larceny (10 percent), and prostitution or gambling (6 percent). Burglary and robbery followed at 5 percent each. Explaining "drugs" as a priority, the authors indicate: "These were typically problems of fairly low-level street dealing, rather than large volume trafficking." None of the top five problems was an index crime. Yet all five contribute to perceptions that one's neighborhood is out of control, that one's turf is not secure. Even parking and traffic problems can add to such fears, particularly if residents believe the source of the problem is "outsiders" --fast drivers using residential streets as throughways; unfamiliar cars parking on residential streets; increasing the number of strangers and making it difficult to tell who has a good reason for being there. Turning to the survey of neighborhood leaders, the report states: "Very few respondents who lived in predominantly white, middle-class, residential areas identified robbery or burglary as problems." Or as the president of a merchants' association reported: "Is there a crime problem now?" Yes. We have eggs splattered on our store windows, but we don't have stick-ups. Commercial crime involves shoplifting and pickpocketing in the larger stores. There is also residential crime, which involves burglaries. But no, we don't have a crime problem of any grave consequence. Certainly neither the authors of the report nor I would want to give the impression that these responses are typical of all of New York's neighborhoods. Violence among youths is endemic in many areas and should be the highest priority for community leaders, public health officials, police and criminal justice officials, and political leaders. Nonetheless, the experience of community organizers, confirmed by my own research, is that disorder is as much or more of a problem in middle- and working-class neighborhoods, even in neighborhoods that are seriously marred by violence. Like other purveyors of goods or services, the Metropolitan Transportation Authority regularly conducts market research to learn about user satisfaction, market potential, and problems in service delivery. My own research as a consultant to the TA, using surveys, focus groups, and other data, confirms that fear has seriously hindered the public's use of subways. Ninety-seven percent of passengers report taking some form of defensive action when riding the subway: They stay away from certain types of people, locations, cars, and exits. Forty percent of New Yorkers believe that reducing crime is the top priority for improving the subway. Only 9 percent believe the subway is safe after 8 p.m.; 76 percent disagree with the statement that there is very little chance you will be a crime victim if you ride the subway after that time; and 62 percent say that fear of crime keeps them from riding the subway at night. Overall, those questioned estimated that about 25 percent of the city's serious crime occurred in the subway. These perceptions are important. But they are not accurate. In reality, only 3 percent of New York City's recorded felonies occur in the subway. By some estimates, only one in 200,000 subway trips is marred by a confrontation felony, which means most New Yorkers could ride the subway regularly for hundreds of years without being part of such an incident. So why are people afraid? Though they rarely experience serious crime, they are constantly exposed to disorder and left with the impression that no one is in charge. Broken turnstiles, litter, graffiti, the homeless, and panhandlers threaten riders and lead New Yorkers to believe that serious crime is more frequent. Fare-beating and other turnstile scams not only amplify this message, but also cost the system as much as $120 million annually. What do subway riders want? First, more police. Second, order: 84 percent of survey respondents agree that it is important for police to stop fare-beaters and 65 percent believe that the homeless should be removed from the subway. In sum, studies of commercial centers, neighborhoods, and subways all call for increased attention to quality-of-life offenses including disorder and drug dealing and for new partnerships between police, citizens, neighborhoods, and businesses. They ask for community policing, often endorsing CPOP by name, and for foot patrols. These studies are hardly exhaustive, but they tend to confirm what common sense and experience suggest: The professionalized, bureaucratized preoccupations of police organizations do not reflect the concerns of most citizens. Police and policymakers must undertake a systematic effort to discover what citizens want from police, what problems are really undermining communities, and how effective police are in fighting them. What these studies have done in fragmentary and informal ways is what formal law enforcement evaluations ought to be doing. We need a new sort of database that will shift the attention of press and politicians alike away from the UCR and arrest and response reports and toward citizens' real problems. Ironically, in the late 1970s the New York City Police Department performed an experiment called "Operation Crossroads" that nearly did just that, although without actually meaning to. Unfortunately, the experimental program was allowed to die and the NYPD never capitalized on what it learned. As described in an unpublished study by the Fund for the City of New York, one of the program's funders, the program's goal was to clean up Times Square, which suffered from the same problems it does today: prostitution, hustling, gambling, scams, and drug dealing. Even then, the consequences of disorderly conditions were intuitively understood: Police and other enforcement officials believe that certain types of street conditions such as the number, type, and frequency of street solicitations, the number of individuals loitering in doorways, and storefront uses and their hours of operations do contribute to . . . serious crime. At the very least, offensive street conditions are perceived as dangerous and threatening to the public . . . . They are a primary contributor to the negative image of Times Square, part of a self-perpetuating cycle of decay. Before Operation Crossroads, police in the area relied on repeated aggressive "sweeps" as their main cleanup tactic. They would identify a problem area, mobilize a squad of officers, and arrest all those who were loitering. Little was accomplished. The troublemakers were often back on the streets sooner than the officers who arrested them. Sweeps consumed enormous amount of police time and were eventually declared unconstitutional. Operation Crossroads addressed three separate but linked questions. First, could counts of disorder be useful in assigning police officers to particular beats or neighborhoods? Second, were alternative tactics available that were both legal and successful in reducing disorderly conditions? Third, could the same counts of disorderly conditions be used to evaluate police tactics for reducing disorder? Researchers established a procedure for documenting disorder. Trained observers counted incidents of disorderly behavior in specific areas. Disorderly behavior was defined to include solicitation or sale of sex or drugs, use of drugs or alcohol by loitering people, all non-food vendors, and several categories of loiterers including vagrants, troubled persons, three-card monte dealers, other gamblers, handbillers, and hawkers. It developed that although the entire Times Square area was viewed as disorderly, the problems tended to concentrate on a few blocks. And while disorder continued throughout the day, the ratio of disorderly persons to other street users changed as evening approached, thus making the area seem more threatening. But perhaps most important was the discovery that disorderly conditions could actually be quantified in this manner. Armed with these new data on disorder, the police decided on a markedly different approach: a high-visibility but low-arrest strategy that explicitly rejected mass arrests in favor of direct action to interrupt and deter disorderly behavior. Thus police would order, counsel, educate, cajole, and use other noncoercive methods to discourage offenders, and would arrest them only as a last resort. The researchers hoped that the disorder counts could be used to allocate officers. Police managers, however, continued to rely on traditional measures to assign police--reported crimes and calls for service. A crisis, however, made it clear that the street condition reports (as they were called) could be useful. Parks commissioner Gordon Davis threatened to close Bryant Park (adjacent to the main branch of the New York Public Library). Drug dealing had reached epidemic levels. Police could not or would not control it. Police managers responded to Davis's threat and the publicity that followed with an aggressive effort that relied on the low-arrest tactics of Operation Crossroads. Instead of using such traditional means as arrest counts to evaluate their own efforts, they used the condition reports. The results were not only interesting but of great practical value: o The number of people engaged in positive activities increased by 79 percent; the number of drug sellers, buyers, and users decreased by 85 percent. o The percentage of loitering and drug-related use as a function of total use declined from 67 percent to 49 percent. o Drug selling was not displaced en masse to any single location outside the park. o While the decrease in the number of dealers was not as dramatic as police had hoped, dealers behaved more discretely. o The aggressiveness of the uniformed officers, not just the fact that they were in the park, appeared to be the key factor in changing the dealers' mode of operation. o Supervised, directed patrol, rather than the absolute number of officers assigned, seemed critical to affecting conditions in the park. o Stationing a uniformed officer in front of the library during lunchtime and early afternoon virtually eliminated the clustering of drug activity. Nevertheless, the project was aborted. Once the crisis was over, police simply were not interested in using the information. As time went on, key personnel were transferred, not to frustrate the project, but as a matter of routine police practice. Soon the funders had little choice but to drop the project altogether. It does not take much reading between the lines to know what was going on: the police were not about to abandon their traditional ways of evaluating their performance and assigning officers in favor of the low-arrest strategy. Operation Crossroads and the Bryant Park crisis had forced police back into a problem area--disorder--that violated the dominant police paradigm. However police managers might phrase their reluctance, in effect they were unwilling to shift to a system that would measure actual results as citizens might experience them, rather than such apparent efforts as arrests. For the police, the goal was still to demonstrate that "we held up our end," rather than "we solved the problem." Distinguishing between what citizens experience in their neighborhoods, shopping centers, and subways and the official crime problem as defined in crime, response, and arrest statistics is not an academic quibble. For generations, public policy has been built around priorities established in response to these data, satisfying the eternal bureaucratic yen to be evaluated by numbers and process rather than by results. Yet whenever citizens are queried--whether systematically, as in many of the reports noted above, or informally--their greatest complaints always include disorder and an accompanying fear. Statistics which indicate that people are hardly ever raped or murdered in their neighborhood or that help is just a 911 call away offer little comfort. I am certain that if systematic studies were available about the "crime problem" in schools, parks, and public housing, the results would be similar. Official police doctrine is changing, especially in New York City. The Mayor, the MTA, the Transit Police Department, and the NYPD all strongly endorse the notion that police must focus on solving the problems that really upset New Yorkers. By controlling disorder and stemming fear, they will keep citizens on the street and thereby discourage serious crime. Serious programmatic reform plans are already underway, with the most well-known being the Mayor's Safe Streets, Safe City plan. At the level of theory, the corner has been turned. But the real change will be much harder than is imagined by those who glibly drop phrases like "community policing" and then stand back and wait for miracles. Despite the city's enormous official commitment to community policing, the issue is still very much in doubt. The dominant criminal justice model has been in place a long time and is supported by powerful traditions and mythologies. The task facing police forces here, and across the country, is to turn away from several decades of accumulated, preconceived, and self-regarding notions about their mission, and to discover instead the real needs of the communities they seek to protect. It is not easy to change an entire subculture. First and foremost, police need to change their own minds about their mission, and give up the view that police work consists of racing around in patrol cars, apprehending criminals after the fact, and feeding them into a "criminal justice system." That "cowboy" version of policing has considerable allure for most of the young people who become police officers, attractions that "problem solving" and community work (often with civilians) do not necessarily have. Former Chief Robert Igleburger of the Dayton Police Department, one of the country's most innovative police chiefs during the 1960s, has likened police departments to rubber bands. They can be stretched, pulled, and twisted into a variety of shapes, yet whenever pressure is relieved, they snap back into their previous shape. Many forces bridle public organizations: traditions, habits, vested interests of groups both within and outside the organization, political chicanery, public myths, and so forth. As we know from the current experience of the auto industry, which had to be brought to the brink of bankruptcy before it began to reform itself, repositioning organizations is difficult, and keeping them repositioned is harder. One way to start--one way that has been overlooked so far--is for New York's Police Department to begin a revolution in American crime statistics. They should move American police (and the American media) away from their unproductive preoccupation with current official data. Taking a cue from Operation Crossroads, the city's police should build new citywide databases that measure the problems that citizens really care about, the ones that spread crime and fear, disrupting the trust of neighbor and community cooperation that is essential to preventing crime. They should develop databases that measure whether police are responding to these problems and databases that measure whether the problems are getting better. Collaborating with citizens to prevent crime and disorder requires knowing what citizens think about crime and disorder. It is useless to demand that police respond to community needs rather than self-serving bureaucratic standards, unless we know what those needs are. It would be unjust and demoralizing to criticize police for not helping to maintain order (which they have been doing to some extent, albeit fitfully, and without commendation or encouragement throughout the 911, UCR-dominated decades) without the data to prove the case, or to commend them when deserved. Creating such databases is one thing, maintaining and updating them will require a real commitment of resources and managerial will. For if they are to be useful, the surveys must measure New York's many neighborhoods separately and in detail. To assume that all communities have the same priorities would be fatal to the effort described here. Yet despite all the work, will, and widgets this effort would consume, it would be very efficient even in the medium term. Such data would be crucial in helping transform police culture and make community policing self-sustaining. By providing police with a new way of thinking about their jobs, they would overcome the entrenched traditions that have impeded past reforms. Even police who initially regard such community policing tactics as foot patrols with distaste almost always learn to like them as soon as the programs get underway. But liking a duty does not go very far unless it is linked to career advancement. Currently, officers move up in the force by leaving patrol work for a job with a specialized unit. And they are promoted out of patrol by doing things that can be added up statistically, like making lots of arrests, rather than by solving community problems. In order to truly change the culture of the police department, the department must tie career advancement to the tasks that make community policing work, especially being a good patrol officer. The department will not be able to do this without data. It is, after all, a bureaucracy, and a bureaucracy it will remain until its dying day. As such, it will always want to play by the numbers. So we must find a way to change the numbers and show police officers that the new way to get ahead is to rack up good numbers of a different sort. For the same reason, the New York Police Department, and all the other departments that follow in its wake, should make an enormous annual or biannual public fuss about the new numbers, crowing shamelessly about every bit of good news, and cheerfully expending the great portions of patience and fortitude it will take to explain them to the press. For to really ensure the future of community policing, we have to change not only the internal culture, but also the public mythology of policing. As one prominent New York police official has put it, "It's not just what these guys learn on the force, most of them are cowboys or Ôbuffs' [lovers of police tradition and lore] before they sign up." And while chiefs battered by the UCR twice a year may no longer be cowboys, there is no doubt that the enormous publicity that accompanies the current statistical measures of performance affects the way police forces behave. Powerful images sustain the "crime fighting" view of policing: the "thin blue line" and the "wars" on crime, drugs, and violence waged by arresting and incarcerating offenders. The statistical parallels of those images, broadly accepted by the media as a scorecard for police performance, now come back to haunt police. Tragic events, such as killings in schools, get wide publicity and fuel demands that police "do something," regardless of what it is. Tough measures must be taken against those who are violent. But we must also take tough measures against myths that deflect press, public, and police alike from the real problems of the community. Not much more than a generation ago, there were other police myths that were powerful and emotionally rewarding: myths of the cop on the beat who knew his block, his people, and what they needed. Officer Murphy--and his nightstick--would not be popular in most New York neighborhoods today. But we can create new heroes of public service in his place, citizen soldiers who know how much their fellow citizens suffer from the grinding, day-to-day incivilities and minor street offenses that erode the quality of urban life, make people afraid, and create the milieu within which serious crime flourishes. Images as powerful as the war metaphors of the 911 era can support them in their struggle. But all this would be made far easier with, and may be impossible without, concrete measures of achievement that redefine successful policing as policing that actually makes people want to live here. ------------------------------- Measuring What Matters: Crime, Disorder, and Fear Wesley G. Skogan This chapter considers two issues: (1) measuring the possible effects of an innovative policing program, and (2) doing so in a framework that could support the inference that the program caused variations that the measurements might reveal. Measurement involves (among other things) the collection of data that represent--sometimes only indirectly--the problems that programs target. These are "outcome" measures, and it is vital that they represent the scope of a program's intentions as accurately as possible. The framework within which these data are collected is evaluation's research design, and it is crucial that the design account for as many alternative explanations for what is measured as is possible under the circumstances. Arguing that "the program made a difference" over the past month or year involves systematically discounting the potential influence of other factors that might account for changes in the measures through the use of randomization, matched control groups or time series, and other design strategies. Measurement issues are a bit more closely related to analytic issues than this distinction suggests. One cannot divorce what is measured from how the measures can be linked causally to programs. What evaluators call the "logic model" of a program--how, exactly, it is supposed to have its desired effect--needs to be specified clearly enough that appropriate outcomes can be identified and their measures specified. For instance, if evaluating a crime prevention program, exactly what kinds of crimes involving what kinds of victims during what periods of the day or night should we examine for evidence of impact? This essay focuses on measurement issues, but it addresses issues through concrete examples of how measures have been used to make judgments about the impact of programs. It examines some of the experiences the evaluation community has had in taking the vital signs of a community by measuring crime, disorder, and fear. This is far from a complete list of what matters in policing, as other articles in this volume attest. However, in Kelling's original plea for a focus on "what matters" in policing, he concluded with a call for a renewed focus on "the grinding, day-to-day incivilities and minor street offenses that erode the quality of urban life, make people afraid, and create the milieu within which serious crime flourishes" (1992: 33). In recompense for the brevity of the list of issues considered in detail in this chapter, I conclude with an inventory of other issues that need to be considered--and appropriately measured--in any thoroughgoing evaluation. Measuring crime The development of a new research technology--survey-based measures of victimization--has enabled evaluators to dig deeper into claims about the effects of policing on crime. Although not without their problems (which will be examined below), survey measures of crime bypass two enormous sieves that strain out so many offenses that it can be difficult to interpret official crime statistics. These sieves are citizen reporting and police recording practices. Together, they work to the disadvantage of the poor and residents of higher crime areas, and they can disguise the effects of programs that might otherwise appear promising. Citizen reporting Interviews with victims indicate that many incidents are not reported to the police, either by themselves or (as far as they know) anyone else. Among crimes measured by the National Crime Victimization Survey, about 40 percent of all personal crimes and 33 percent of property offenses are reported. Reporting is high for auto thefts (93 percent of successful thefts) but much lower for simple assaults (43 percent), attempted rapes (33 percent) and robberies (36 percent), and pocket pickings (22 percent). Only 52 percent of successful residential burglaries and less than 12 percent of thefts of less than $50 are reported (Bureau of Justice Statistics, 1996, table 91). Crime reporting by witnesses rather than victims is even lower. In Britain, only about 12 percent of the instances of shoplifting, 8 percent of serious fights, and 29 percent of thefts from cars observed by the public are reported to anyone (Skogan, 1990b). Furthermore, the National Crime Victimization Survey reveals that reporting differs by population group. Generally, lower income people, younger victims, and men report victimizations at a low rate, while homeowners report at a high rate. Incidents away from home, those with smaller financial consequences or for which victims had no insurance, and crimes in which victims and offenders know one another well are reported less frequently. Black on white crimes are also more likely to be reported. In some crime categories, fear of retaliation discourages reporting; in others, people do not report because they plan to take action on their own. The belief that police would not want to be bothered or that they are ineffective or biased is responsible for about 10 to 15 percent of nonreporting, depending on the category of crime. In addition, programs and practices that involve people more intimately with policing also encourage crime reporting when these people are victimized. That is, crime prevention and other programs that ask citizens to "be the eyes and ears" of police, hopefully do increase reporting, but the higher crime figures could make those efforts look counterproductive even if the actual crime rate has not changed or has decreased. It appears this effect has only been documented once--by Anne Schneider (1976) in an evaluation of a residential burglary prevention program in Portland--but the threat of looking worse as a result of doing better has made almost all evaluators aware of the difficulties of using reported crime figures to evaluate programs. Police recording practices In addition to the fairly systematic bias introduced by citizen nonreporting, official figures are further confounded by the vagaries of police recording practices. Founded incidents are not the same thing as reported incidents, often for good reasons, but the gap between the two can disguise deceptive recording practices. At several levels, police may act to avoid unpleasant or seemingly unproductive work, forestall complaints about their behavior, or respond to pressure from their supervisors to keep the crime count down. Bona fide reported offenses may be shifted from one category to another, mostly to downgrade them or so they can be ignored. In numerous well-documented cases, there have been sharp changes in crime rates associated with reform movements, changes in political administration, turnover among district commanders, and the like. In Chicago, detectives were caught "killing crime" at an enormous rate by unfounding (determining that a case is unverifiable) rape, robbery, and assault incidents without investigation. The practice was widely understood within the department, which kept two sets of books--one public and one private--on reported offenses (Skogan and Gordon, 1983). Administrators who want honest accounting have few choices. One is to examine the ratio of recorded crimes to arrests in hope of spotting districts where the two figures are too close together; they can also monitor the crime clearance rates reported by their detectives. Another strategy for encouraging honesty in bookkeeping is to conduct expensive field audits that track the course of 911 calls, beginning with the communication center's running tape; Chicago's department did this for a decade in response to the "killing crime" scandal. However, changing technology is undermining the apparent control that centralized complaint-taking and dispatch gave downtown managers over police operations. Police and the public are increasingly communicating with each other directly--using beepers, cell phones, and voice mail--rather than through 911. In addition, community policing strategies almost always involve increasing the frequency of face-to-face meetings and informal encounters between the police and the public for the purpose of exchanging information. The old systems for command and control within police agencies produced a torrent of data on crime and disorderly conditions; these data were sometimes of dubious quality, and now they are becoming increasingly unreliable. Survey measures of crime There are alternative measures of crime, however. The most well known are victimization rates based on surveys that quiz respondents about their recent experiences with crime. These measures bypass citizen reporting and police recording practices and typically produce estimates of the crime rate that are two to three times those based on official sources. In the aggregate, they sometimes trend in the same direction as official figures. This is particularly true at the national level when expansive categories of crime are considered over a period of years and after adjustments are made to account for some of the differences between the two series (Biderman and Lynch, 1991; Mirrlees-Black et al., 1996). However, for small areas, tight timeframes, and detailed categories of crime, it is unwise to expect survey and official figures to point to the same conclusions. Exhibit 1 presents a fragment of a typical victimization screening questionnaire designed for telephone administration. The original questionnaire (Skogan, 1995) included 18 screening questions that probed for both personal and property victimizations. The questioning strategy was to first elicit yes-no responses about each scenario on the list, and then return to followup questions like those employed in this study ("Was it reported to the police?" "Did this happen in your neighborhood?"). For the respondent, this breaks any apparent link between giving a "yes" response and the burden of answering additional questions, a link that suppresses the victimization count (Biderman et al., 1967). Information about the location of incidents is frequently required to identify those that took place in the targeted area and those that occurred elsewhere. In personal interviews it is possible to show respondents a map and ask them to identify where specific incidents took place. This is particularly useful if the area under consideration is a police district or administrative unit that does not closely correspond to popular conceptions of local neighborhood boundaries. Problems with survey figures Coverage. Not everyone will be included. Interview refusal rates can be high, and they are growing. The problem is compounded in multiwave studies in which respondents are reinterviewed over time. In a mobile society, recontact rates can be low if more than a few months pass between the waves of a survey, and that loss differs from group to group. In particular, young people, renters, and short-term residents of the community are difficult to reinterview, while women, family members, and homeowners can be found again more easily. Young people (who are at greatest risk) are hard to find at home at any time. Also, many crimes are reported by organizations (such as schools), merchants (Shapland, 1995), and others who will be left out if only households are included in the survey. These groups experience a considerable number of victimizations. The last national commercial victimization survey revealed a burglary rate of 217 per 1,000 establishments, as contrasted to a household rate of 89 burglaries per 1,000 dwellings (National Criminal Justice Information and Statistics Service, 1976). Among crimes reported to the police, one-third of burglaries involve "nonresidential" (largely commercial) targets (Federal Bureau of Investigation, 1995). However, it is common practice to survey only households. There is a great deal of debate about the relative merits of telephone versus in-person surveys. The latter cost more, but many inner-city homes have no telephones. In Chicago, there are strong links between race, poverty, crime, and the accessibility of people for telephone surveys. At the census-tract level, the correlation between telephone access and the gun crime rate is (-.44). It is (-.67) for families on public aid and (+.50) for homeowners. Among the city's prototype community policing districts, 10 to 19 percent of households in the two poorest areas did not have a telephone, and more than 20 percent of households in the northern end of another district did not have a phone (Skogan, 1995). On the other hand, survey refusal rates in big cities are lower for telephone than in-person surveys, partly because respondents are unwilling to let strangers into their homes. The difficulties involved in managing and protecting the safety of interviewers in higher crime neighborhoods are considerable because it is important to conduct interviews during evening hours (Groves and Kahn, 1979). It is not clear what the bottom line is on this issue, and in the end it is usually decided by cost. Expense. Surveys typically use samples to represent the populations of neighborhoods, districts, or cities. This introduces error in the findings; if that error is going to be acceptably small, the surveys have to involve fairly large numbers of respondents. The issue of how many respondents are needed is determined by the subject. For example, documenting an anticipated drop in the prevalence of burglary victimization from 15 percent to 10 percent of households (a 33-percent decline) requires interviews with about 340 respondents each time (cf., Kraemer and Thiemann, 1987). Getting the count straight. One of the most interesting developments in studies of victimization is the analysis of what makes high-crime neighborhoods "high crime." Research in Great Britain suggests that the key fact is not that more people are victimized in these areas; while that percentage is higher in high-crime areas, what distinguishes the worst areas is that some residents are repeatedly victimized. Repeat or multiple victims contribute disproportionately to the overall crime count in high-crime areas (Farrell, 1995; Trickett et al., 1992). This is both good news and bad news. It is good news because it gives us more leverage on the crime rate. It suggests that programs that target first-time victims could have "more bang for the buck" than scatter-shot prevention efforts because once-victims are much more likely than nonvictims to be targeted. This phenomenon presents a cheap and apparently effective way of targeting criminal justice resources and suggests that cities that have invested in security surveys, hardware upgrades, and other support services for victims were on the right track (Anderson et al., 1995; see Spelman, 1995, for another view). It is bad news because even the best surveys are not very good at measuring repeat victimization. The reasons victim surveys are poor at measuring repeat victimization are complex: A combination of general bounding, telescoping, temporal ordering, forgetting, differential recall, series victimization, estimation, design-effect, and confidence-interval problems pile up on this particular issue (Skogan, 1981). One way of ignoring some of these problems has been to avoid trying to measure victimization rates, that is, the number of crimes occurring in an area divided by the number of residents or households. Rates are severely affected by most of the problems listed above because rates involve estimating the number of crimes that have occurred. Instead, almost every published evaluation in the police field has examined survey measures of the prevalence of victimization, or the percentage of persons or households who have been victimized once or more. This figure is resistant to some of the problems outlined above: We only need to know that something happened to someone to categorize that person as a "victim." Prevalence measures are also easier to analyze using multivariate statistics, because whether or not a person was a victim is an experience that easily can be related to the individual's background, household, and lifestyle factors. Finally, prevalence measures require less questionnaire space and interviewer time because fewer details are required to get a yes-no answer. But we now know that this approach is remarkably insensitive to one of the forces that drives up neighborhood crime rates, and it is not well-suited for evaluating what appears to be a promising crime prevention strategy. An example The situation is not as hopeless as the discussion above might suggest. Because they are so difficult to assess when many issues and potential program outcomes compete for evaluation resources, I have found triangulation a useful strategy for analyzing multiple, flawed measures of crime rates. Exhibit 2 illustrates the findings of a recent evaluation of community policing in two of Chicago's police districts (Skogan et al., 1995). It compares the findings of household surveys and an analysis of 34 months of founded crime incidents. Exhibit 2 reports (1) perceptual measures asking "how big a problem" specific crimes were in the community (see the next section about this); (2) officially recorded crime counts; and (3) survey measures of the prevalence of victimization. These two crimes were selected for close examination because they were among the four top-rated problems in these two districts. The probability figures presented below each of the survey-based figures indicate how likely the changes described were to have arisen by chance. The percentage change is presented for officially recorded crimes. In this example, all of the measures pointed in the same direction, lending more confidence to the conclusion that crime went down substantially in these districts. In Morgan Park, auto theft as measured in the survey was down significantly, as were reports that it was a "big problem" in the area. In Austin, robbery was down in both survey measures. Both districts saw a decline in officially recorded crimes in these categories, especially Morgan Park. In the comparison areas matched to these districts, robbery and auto theft also declined, but only slightly. Measuring disorder Important as it is, there is reason to doubt that crime reduction is the sole "bottom line" for evaluating policing. Narrowing their traditionally wider scope of responsibility was one of the strategies reformers used to capture control of police organizations (Kelling and Coles, 1997), but the profession has paid a price for the consequences. To "police" society implies a wider mission, and expanding the police mandate is a fundamental feature of modern problem-oriented policing. Police are the only servants of the people who are available 24 hours a day and continue to make house calls. They also have taken on a wider range of problems because, when given the opportunity, their "customers" demand it. In Chicago, observational studies of small public meetings that are an integral part of the city's community policing program reveal that neighborhood residents are concerned about a broad range of problems, including traffic enforcement, illegal dumping, building abandonment, and teenage loitering (Skogan and Hartnett, 1997). One aspect of this new and larger police agenda is an untidy bundle of problems that I have labeled "disorder" (Skogan, 1990a). Disorder is apparent in widespread junk and trash in vacant lots, decaying and boarded-up buildings, vandalism and graffiti, and stripped and abandoned cars in the streets and alleys. It is also signaled by bands of teenagers congregating on street corners, prostitutes and panhandlers, public drinking, verbal harassment of women on the street, and open gambling and drug use. For many purposes, it is useful to think of these problems as falling into two general classes: social and physical. Social disorder is a matter of behavior: You can see it happen or observe direct and tangible evidence that it is a problem. Physical disorder involves visual signs of negligence and unchecked decay: abandoned or ill-kept buildings, broken street lights, trash-filled lots, and alleys strewn with garbage and alive with rats. By and large, physical disorder refers to ongoing conditions, while social disorder appears as a series of more-or-less episodic events. What these conditions have in common is that they signal a breakdown of the local social order. They are violations of what James Q. Wilson (1968) called "standards of right and seemly conduct." Of course, to be useful, a concept must also be bounded. It cannot encompass every nuance of behavior. Disorder violates widely shared norms about public behavior; these norms prescribe how people should behave in relation to their neighbors or while passing through the community. They are not a neat bundle of rules, because legislatures have not set some of them in cold type even though they are widely agreed upon. Some activities in the bundle are unlawful, but it has been difficult to get police to take most of those very seriously. Because many norms about public behavior are uncodified and others are not traditionally defined as "serious," evaluators need to work through the untidiness of disorder to identify its dimensions in a particular context. They usually need to develop new measures of their prevalence because the uncodified status of many disorders means there are few official reports or indicators of the extent to which they plague particular neighborhoods. The importance of disorder to policing's customers can be illustrated by what happens during beat meetings in Chicago. These meetings are a central aspect of the city's program, for they are the principal arena in which joint problem identification and problem solving takes place. Attending 146 of these meetings, we noted a total of 113 different problems that were discussed, as well as 36 types of solutions to them. Of the problems recorded in our observations, 21 were mentioned in at least 10 percent of all beat meetings. These are depicted in exhibit 3. About half of these problems are related to social disorder issues; note the high rating given to "youth problems." Complaints about police procedures made up another quarter of these issues, including two of the top four problems. Another fifth of the top issues involved the decay of the physical environment, in the form of graffiti, litter, and abandoned cars and buildings. The kinds of core problems around which reactive policing was organized--represented here by complaints about either burglary or robbery--ranked only 17th on the list and were brought up in only 12 percent of all meetings (Skogan et al., 1995). There are at least three approaches to measuring the extent of disorder: analysis of archival records, direct observation by trained observers, and sample surveys. Each has strengths and weaknesses, and these are reviewed in detail by Ralph Taylor in his essay "The Incivilities Thesis: Theory, Measurement, and Policy" in this volume. I focus here on survey-based measures of disorder. Survey measures of disorder The importance of disorder in the eyes of the general public can be seen in surveys. Boston's 1995 public safety survey asked respondents about 16 different kinds of incidents or conditions in their neighborhood, asking them to rank "how big a problem" each was. The top rankings belonged to auto theft and drugs, but next were noise, public drinking, and vandalism; then, after burglaries, came kids hanging around, graffiti, and panhandling (Boston Police Department, 1995). A survey of the most dangerous district involved in Chicago's community policing project found that two of the most highly rated local problems were gang violence and drug dealing, but between them came abandoned buildings; the fourth-biggest problem was junk and trash in the streets and sidewalks. Respondents in that survey also thought that public drinking was a bigger problem than burglary, assault, or rape (Skogan et al., 1995). While many surveys ask "how big a problem" specific disorders are, other formulations of the question include "how worried are you about . . ." (Maxfield, 1984) and "how concerned are you about . . ." (Mayhew et al., 1989). These approaches confound the prevalence of problems in their environment with their perceived impact on the respondent, which are not necessarily the same issue, and I would not recommend them. Determining what disorders to include in an evaluation is, of course, driven by the problems facing the communities involved and the nature of the programs being developed. For example, some circumstances might call for targeting alcohol-related problems. In Chicago, we asked residents of program and comparison areas about "things that you may think are problems in your neighborhood." They were read short lists of problem descriptions and asked each time if they thought it was "a big problem, some problem, or no problem in your neighborhood." The following alcohol-related problems were addressed: o Public drinking--27 percent thought it was some problem; 20 percent, a big problem. o Taverns or liquor stores selling alcohol to minors--21 percent thought it was some problem; 15 percent, a big problem. o Taverns or liquor stores attracting troublemakers--23 percent thought it was some problem; 19 percent, a big problem. In other studies, I have examined survey reports of the extent of a variety of disorder problems: o loitering o vandalism o street harassment o fly dumping o massage parlors o abandoned buildings o noise o abandoned cars o junk-filled vacant lots o truancy o panhandling o litter and trash o graffiti o public drinking o broken windows o public gambling o loud parties o school disruption o public insults o spray painting o dilapidated buildings o taverns o topless bars o dirty streets and sidewalks o pornographic theaters In each case, it was necessary to tailor the specific wording of the question to local conditions. For example, questions about topless bars were included in surveys in Houston because I could not help but notice beer halls with flashing neon signs announcing "Naked Girls Dance" in several of the targeted residential areas (Skogan, 1990a). Are these perceptual measures valid indicators of the true extent of disorder in a community? Unlike survey measures of victimization, relatively little research has addressed the matter, and much of it is reviewed in Ralph Taylor's "The Incivilities Thesis: Theory, Measurement, and Policy" in this volume. The question is whether responses to these kinds of survey questions can be accepted as useful reports on neighborhood conditions and whether we can treat respondents as informants. Responses to questions about disorderly conditions might reflect respondents' biases or personal preferences, or they might be random answers made up on the spot to satisfy interviewers. The middle choice (respondent bias) implies that disorder largely rests in the eye of the beholder and that surveys are not a very useful way of gathering intelligence about the distribution of neighborhood problems. However, statistical analyses suggest that the surveys are not just measuring intolerance for all but conventional middle-class views of how people ought to behave. Rather, there is evidence that major economic, social, and lifestyle groups within neighborhoods are in a great deal of agreement about the problems they face and that the surveys actually represent neighborhood differences in conditions, not just individuals' views. Another approach to validating survey results is to compare them with the extent of specific disorders measured by observing the same area. This is easiest to do for such observable neighborhood conditions as litter, graffiti, and building abandonment. Ralph Taylor and his colleagues made carefully controlled observations of those factors in 66 neighborhoods. The results were correlated with perceptual measures gathered in surveys of the same areas. The correlations were not always very high. They were highest when the survey and observational data were combined to form general indices and when they were compared for small areas. However, at the single-measure, problem-specific level, the extent to which the low correlation could be attributed to measurement errors on both the survey and observational sides of the comparison is unclear. Observational measures of disorder As this hints, there are great possibilities for observational measurements of the targets of some policing programs. This work was pioneered by Ralph Taylor, who has conducted block-by-block physical surveys of neighborhoods in Baltimore. His observers assessed and scored the physical dilapidation of individual buildings as well as the deterioration of streets, alleys, and sidewalks. They noted the presence of abandoned buildings and storefronts, graffiti, and litter. These factors were then correlated with resident morale and calls for police service. Other researchers have examined the distribution of graffiti and abandoned cars or the impact of taverns, schools, and mixed land use on crime. This research is not easy to conduct. There must be acceptable levels of inter-observer agreement on what they observed for us to accept the results of their judgments; also, it is important to ensure the safety of observers. There are limits to what can be observed and what persons living in a neighborhood can be asked about. For example, Richard Taub (Taub et al., 1984) found that his observers could not reliably count junk in front yards and vacant lots that was "smaller than a toaster," so they used that standard. Many of the phenomena we would like to observe can be transitory in character, especially if observers are looking at social behavior rather than physical manifestations of decay. These disorders are events rather than conditions, so brief observations are likely to miss them. They vary enormously by the time of day, the day of the week, and the weather. In one study, during repeated and lengthy observations of specific locations that had been identified as high-disorder hot spots, observers actually saw something disorderly take place very infrequently. A survey example Exhibit 4 reports the results of surveys of five police districts in Chicago, using the "how big a problem" formula described above. It identifies the 4 neighborhood problems that were the most highly ranked in each district from a list of 22 problems that were presented to respondents in 3 different sections of the questionnaire. Several points are illustrated. First, some problems were common across many or most of the districts, including drugs and gang violence. Street drug sales were on the agenda in every community; gang violence, in four of the five. However, the other top problems differed from place to place, and issues that loomed large in some areas were scarcely problems in other districts. In one dense area with little off-street parking, vandalism to automobiles was one of the area's top four problems; only in the wealthiest area was auto theft on the list. Thus, one goal of community policing is to open departments up to local input--so they can effectively discern these variations in local concerns and tailor their operations to respond to them. Second, not all of the problems on people's minds fell in the "conventionally serious crime" category. A wide range of problems were identified as vexing. Car vandalism was near the top of the list in two areas, as was graffiti. Street crime was also highly rated in two areas. Auto theft, burglary, disruptions around schools, abandoned buildings, and "vacant lots filled with trash and junk" each stood near the top of the list in one district. It is interesting to note that only in one district--Morgan Park- -did conventionally serious crimes constitute all four of the area's most highly ranked problems. This was the wealthiest area of the group, one that is the home of many city workers and has strong connections with city hall and municipal service agencies. In the other four districts, two of the top four problems were quality-of-life concerns rather than conventionally serious criminal offenses. Finally, exhibit 4 illustrates that the initial levels of these "biggest problems" varied considerably from district to district. For example, street drug dealing was rated a big problem by more than 60 percent of residents of Englewood, but only by about 13 percent of the residents of Morgan Park, and by 20 percent of those we interviewed in Rogers Park, even though it was among these areas' top-ranked issues. In Morgan Park, burglary was a top-ranked problem, but only 10 percent gave it a high rating. In Morgan Park in particular, there was not much room for improvement on many dimensions, and expectations about the potential impact of community policing on problems had to be tempered by this fact. What was the impact of the program on these problems? Exhibit 5 examines this question. It depicts Wave 1 and Wave 2 survey results for one district and its matched comparison area. The biggest problems in Englewood included drugs, gang violence, abandoned buildings, and trash-strewn lots. The values in parentheses near the bottom of the figure present the statistical significance of Wave 1 to Wave 2 changes within the area. This is the likelihood that the change recorded actually reflects a chance fluctuation in the survey. (We only want to pay attention to changes that were probably not due to chance.) Detailed statistical analyses of the data are not presented here, but they reinforced the patterns that can be observed in exhibit 5. In Englewood, all four of the biggest problems declined, while none went down significantly in its matched comparison area. Street drug sales were ranked a big problem by 62 percent of Englewood residents in 1993, but by only 49 percent in 1994. Abandoned building problems dropped from 43 percent to 27 percent. Gang violence was down only modestly, declining from 41 to 35 percent, but it increased significantly in Englewood's comparison area. Detailed statistical analysis provided additional evidence that these problems all declined significantly after 15 months of community policing. Measuring fear of crime There have been many efforts to clarify the meaning of the concept of "fear of crime" (Ferraro and LaGrange, 1987; Maxfield, 1984). Some are troubled that there is no clear consensus on what the concept means or how it is best measured and that studies that measure the concept in conceptually diverse ways find that different operationalizations of fear are only moderately correlated with one another. However, this heterogeneity of meaning simply reflects the fact that fear of crime is a concept of everyday language, one suited for casual conversation. People commonly talk about fear of crime and its social and political effects; for example, one hears that the elderly are "prisoners of fear," traumatized by the thought of venturing out because of the risks they would face. But the concept needs to be refined for research purposes, and how it is best defined depends upon the purpose of the research. Research on fear of crime conceptualizes it in one of four ways. Three definitions are cognitive in nature, reflecting people's concern about crime, their assessments of personal risk of victimization, and the perceived threat of crime in their environment. The remaining approach to defining fear is behavioral and defines fear by the things people do in response to crime. Dissecting these variations in how fear of crime is defined is important because they make a great deal of difference in what researchers have found. Different definitions of fear can lead to different substantive research conclusions. Concern about crime The "concern" definition of fear focuses on people's assessments of the extent to which crime and disorder are serious problems for their community or society. Concern is a judgment about the frequency or seriousness of events and conditions in one's environment. There are a number of approaches to measuring concern. Opinion surveys ask whether crime or disorder is increasing or decreasing and whether respondents would place them on their list of "most important problems." Most research adopting this definition of fear examines neighborhood conditions. In my research I have asked about "how big a problem" respondents think various conditions are in their immediate area. The 1995 Boston Public Safety Survey asks, "Is crime a problem in Boston?" The British Crime Survey gives respondents a list of crimes and disorders and asks, "how common or uncommon they are in your area?" (Maxfield, 1984). Respondents also are sometimes asked to compare crime in their neighborhood to the city as a whole. Even in the highest crime cities, most report that their own area is "below average." Massive surveys of 13 cities conducted by the U.S. Census Bureau during the 1970s found that only 7 percent thought their neighborhood was more dangerous when compared to others in the metropolitan area (Garofalo, 1977). This is likely to be true because the distribution of crime within cities typically is very skewed, with a few areas driving up the citywide total. Because they ask for a report on neighborhood conditions that is independent of how respondents perceive their own risks, measures in this category are typically unrelated to those that tap the emotive dimensions of fear. Risk of victimization The second common meaning of fear is the perception that one is likely to be victimized. Since the surveys sponsored by the President's Crime Commission in the mid-1960s (Biderman et al., 1967), researchers have asked people to rate their chances of being victimized. For example, survey respondents may be asked to rate "how likely" they are to be attacked or burglarized, on a scale ranging from "not very likely" to "very likely." Assessments of risk are respondents' perceptions of the likelihood of things happening to them, and these are frequently recommended as measures of "fear." In the 1988 British Crime Survey, respondents were asked to rate their risk of being victimized in the next year using a six-point scale from "certainly not" to "certain to be victimized" (Mayhew et al., 1989). Risk measures appear to factor in what respondents have done to protect themselves from victimization. As a result groups like the elderly--who report high levels of fear on other dimensions--do not perceive of themselves as particularly at risk because they are much less exposed to victimization (Skogan, 1993). Threat of crime Definitions of fear focusing on threat emphasize the potential for harm that people feel crime holds for them. Threat levels are high when people believe that something could happen to them, if they exposed themselves to risk. The concept of threat is distinct from those of risk and concern. People may adopt various tactics to reduce their vulnerability to victimization; as a result, they may not rate their risk as particularly high because they avoid exposure to risk. However, they might rate the threat of crime as high if they were to be exposed to risk. Because many people believe they are capable of dealing with crime, threat is also distinct from concern about the issue. Threat is measured by questions that ask, "How safe would you feel if you were out alone?" or, "How would you feel if you were approached by a stranger on the street or heard footsteps in the night?" (Taub et al., 1984). Numerous surveys have found that the threat of crime is felt most strongly by the elderly, and in comparison to measures of risk or concern, questions measuring threat clearly differentiate senior citizens from the remainder of the adult population. Fear as behavior A final, important conceptualization of fear of crime is what people do. This operational definition of fear focuses on the behavioral, rather than cognitive, aspects of the attitude. From this perspective, fear is best assessed by how it manifests itself in the frequency with which people go out after dark, restrict their shopping to safer commercial areas, fortify their homes against invasion, and avoid contact with strangers. The International Crime Survey, which has been conducted in almost 30 countries, asks if respondents avoid certain areas, go out with an escort, have a burglar alarm, leave their lights on when away from home, and ask neighbors to watch their homes when they are away (Van Dijk and Mayhew, 1993). This research usually examines two general classes of reactions to crime: those that limit risk of personal attack by avoiding potentially threatening situations and those defensive tactics that reduce the vulnerability of households to burglary and home invasion. This distinction was first drawn by Furstenberg (1971), who dubbed them "avoidance" and "mobilization." Avoidance definitions emphasize behaviors aimed at reducing risk of personal crime, such as avoiding dangerous places and people and walking only with an escort (rather than alone) after dark. Mobilization includes the extent to which people fortify their homes against crime by adopting security measures such as special outdoor lights, door locks, window bars, and interior lights and by marking their property with a special identification number. Which measure to use It makes a difference what measure is used. For example, research on the effects of mass media coverage of crime is contingent upon the conceptualization of fear that is used. Tyler and Cook (1984) found that exposure to media stories about crime increased people's concern about crime (as it is defined here, the belief that crime is a growing community problem). However, they found that it did not affect people's perception that their own neighborhood was unsafe or that their personal safety was at risk. Other researchers have found that political attitudes and measures of ideological position are correlated with concern measures, but not with risk or threat measures. Victimization, on the other hand, has clearer effects on both risk and threat measures. Interestingly, the elderly's well-known fear of crime is manifested only on the threat measure; they do not rate their own risk of being victimized as particularly high, they do not perceive their neighborhoods as particularly disorderly, and they are much less likely than others to be concerned about crime (Skogan, 1993). As this summary implies, it is important that evaluators pick and choose fear measures carefully. To evaluate the impact of visible patrol, it would be wise to use threat measures, which assess perceived risk "outside." On these measures, almost no one feels very unsafe during the day, so after-dark fears--and after-dark programs--need to be assessed. Domestic violence programs would call for tailored behavioral measures that would assess, for example, things victims do to distance themselves from abusive partners. The fear-of-crime measure employed by the National Opinion Research Center, the Roper poll, and others ("Is there a place nearby"--that is, within a mile--where you would be afraid to walk alone after dark?") would be a useful hot spot question, especially in conjunction with a followup open-ended question identifying the location. Specific interventions might call for fear measures linked to specific types of crime; for example, house burglary or robbery near automatic teller machines. Offense-specific measures of fear are more strongly linked to one another than are broad or heterogeneous measures (Warr, 1984). An example Can better policing affect fear of crime? This is an area where I think the common research wisdom is wrong. The notion that visible policing does not make a difference in fear and attitudes toward police stems from early experiments conducted in Kansas City. Police there were selectively withdrawn from some experimental precincts and their numbers beefed up in others to gauge the effect of the extent of routine (largely motorized) patrol on crime and fear. Researchers found no differences in the subsequent views or victimization experiences of residents of the experimental and comparison areas. Residents also did not notice that the number of police assigned to their area had changed. There has been research before and since that ran counter to these conclusions, but the Kansas City findings (Kelling et al., 1974) became famous. However, researchers working with survey data on the visibility of policing and contacts between the public and the police quickly note that associations between visibility, contacts, satisfaction, and fear are strong, persisting even when a long list of alternative correlates are controlled for. This can be illustrated by the findings of an ongoing evaluation of community policing in Chicago (see Skogan and Hartnett, 1997). Unlike Kansas City, the evidence in this case is correlational rather than experimental. But it also involves a program that suddenly increased--this time visibly--the level of police activity in selected areas. The apparent consequences of police visibility in Chicago contradict the Kansas City results. In this evaluation, respondents were questioned twice, once before the program began and again after about 15 months. The research examined the impact of experiences the respondents personally had between the two waves of interviews. Fear of crime was measured each time by responses to three questions about localized, outdoor crime threats: o How safe would you feel being alone outside in your neighborhood at night? [four responses, ranging from "very safe" to "very unsafe"] o Is there any particular place in your neighborhood where you would be afraid to go alone, either during the day or after dark? [yes or no] o How often does worry about crime prevention prevent you from doing things you would like to do in your neighborhood? [four responses, ranging from "very often" to "never at all"] The reliability of the composite scale combining these items was 0.66. Before the program began, levels of fear were higher among women, low-income and less educated people, African-Americans, and renters. Statistical analysis found that the impact of visible community-oriented police efforts (walking on foot, talking with residents, patrolling the alleys) on this fear measure was large and highly significant. Controlling for many other factors, residents who subsequently observed the police involved in a list of community-oriented activities (not just driving by) felt safer. The most important control factors took advantage of the fact that the respondents were interviewed twice: The analysis controlled for a measure of how fearful they were before the program began and what they reported seeing police in their area doing before the program began. Controlling for past experience, residents of the target community policing neighborhoods were less fearful and more satisfied with police responsiveness to community concerns; they also thought police were more effective at dealing with crime. The effect of police visibility on fear was of about the same magnitude as the effects of age and sex, two of the strongest determinates of fear. To illustrate the magnitude and generality of the involved effects, exhibit 6 charts Wave 2 responses to the first fear question listed above, "How safe would you feel being alone outside in your neighborhood at night?" It shows the percentage of respondents who replied "unsafe" or "very unsafe." The visibility of community-oriented policing during the period between the interviews is represented by a count of sightings (ranging from zero to four) of two different kinds of foot patrol--police checking buildings and alleys, and officers having informal conversations with citizens. Whites were less fearful than African-Americans or Hispanics, most notably when police visibility was very low. However, levels of fear were lower for all groups when the police were more visible. Also, the downward slopes of the lines for African-Americans and Hispanics were somewhat steeper than the slope for whites. This suggests the effect of police visibility was greater for minorities than for white respondents. Police-related measurement issues Having developed useful indicators of the extent of crime, disorder, and fear, is the evaluator's task done? What we have reviewed is just the beginning. A thoroughgoing evaluation may have to attend to many more issues that call for systematic measurement. The list is long, and some issues--such as those related to assessments of the quality of police service, the visibility of policing, police-citizen contacts, and satisfaction with encounters with police--are worthy of a conference in their own right. The following section addresses some of the issues that evaluators have found crucial. Visibility of police Since the Kansas City preventive patrol experiment, surveys have routinely included questions about observation of various police activities. No research has addressed the accuracy of these measures, which is probably fairly low. Visibility should be mostly related to how frequently people are positioned to see police, and it is typically much lower among older people, the unemployed, and women. In our Chicago study, we used a checklist of seven common police activities that neighborhood residents might observe, including driving through the area, patrolling a nearby commercial area, pulling over an auto or searching or frisking someone, patrolling an alley or checking garages, and having an apparently friendly chat with people from the neighborhood. All of these were commonly observed in the dense, not-well-off areas that we surveyed. Over time, the activities commonly associated with community-oriented policing (conversations, foot patrols, and alley or garage checks) were observed more frequently in the program areas than in the comparison areas. Those activities were also linked to reduced fear of crime (as illustrated in exhibit 6), while visible motorized patrol seemed to have no consequences at all. Encounters between police and the public The survey approach screens for encounters between police and the public within a specified recall period (e.g., "the last 6 months"), using a list of typical contact situations. The British Crime Survey, which is conducted in person, presents respondents with a checklist of 17 scenarios-- ranging from reporting a crime to asking for directions--and asks if they have been involved in them during the past 12 months. More than 50 percent of Britons recalled such a contact during 1992. Almost 40 percent contacted the police, while an overlapping 33 percent were stopped by police or were contacted in the course of an investigation (Skogan, 1994). There are no comparable national figures for the United States. In our Chicago surveys, we screen respondents for nine types of citizen-initiated contacts, ranging from reporting a crime to contacting the police to ask for information. We also ask about their involvement in motor vehicle stops and being stopped while they are on foot. In April 1993, 61 percent of adult Chicagoans recalled one or more of these direct contacts with police during the past year. In addition, almost 30 percent indicated they had received a parking ticket in the city during the previous year, but we did not include that indirect contact in the 61 percent figure. Assessments of the quality of police service Remarkably little attention has been focused on developing measurements of public assessment of police service. In Chicago, we have asked "how good a job" respondents think the police do at a variety of tasks and under a variety of circumstances, "how satisfied" people are with specific police efforts, and how well the police behave "toward people in this neighborhood." Typically, 15 to 20 percent of respondents insist that they "don't know" about these things; analytically, they turn out to be older, to have had no recent contact with police, to watch little or no television, and to be uninvolved in neighborhood life. Assessments of encounters with police Following a contact screen like that described above, respondents recalling an encounter can be questioned about what transpired. If they have had multiple contacts, they should be asked about the most recent one. These data are particularly useful because they can provide a detailed "consumer report" of recent encounters with police. The British survey asks those who contacted the police about response time, efforts that police made at the scene, the interest the police seemed to show in the case, if the respondent had any followup contacts with police about the matter, and how politely the respondent was treated. People who were stopped by the police are asked if they were given reasons for being stopped; if they were questioned, searched, or breath-tested; and if they were arrested, prosecuted, or otherwise sanctioned. In Britain, all of these factors are closely related to how satisfied people who have had contacts are with the quality of police service (Skogan, 1994). One complication is that many crime victims who contact the police have also been stopped or even arrested by them in the recent past, complicating how they judge the quality of the service they receive (Maxfield, 1988). References Biderman, Albert D., Louise A. Johnson, Jennie McIntyre, and Adrianne W. Weir. Report on a Pilot Study in the District of Columbia on Victimization and Attitudes Toward Law Enforcement. U.S. President's Commission on Law Enforcement and Administration of Justice, Field Survey I. Washington, DC: U.S. Government Printing Office, 1967. Biderman, Albert D., and James P. Lynch. Understanding Crime Incidence Statistics: Why the UCR Diverges From the NCS. New York: Springer-Verlag, 1991. Boston Police Department. The 1995 Public Safety Survey. Boston: Boston Police Department, 1995. Bureau of Justice Statistics. Criminal Victimization in the United States, 1993. Washington, DC: U.S. Department of Justice, Office of Justice Programs, 1996. Farrell, Graham. "Preventing Repeat Victimization." In Building a Safer Society: Strategic Approaches to Crime Prevention, Crime and Justice: A Review of Research, vol. 19, ed. Michael Tonry and David P. Farrington. Chicago: University of Chicago Press, 1995: 469-534. Federal Bureau of Investigation. Crime in the United States 1994. Washington, DC: U.S. Department of Justice, Federal Bureau of Investigation, 1995. Ferraro, Kenneth F. Fear of Crime: Interpreting Victimization Risk. Albany, NY: State University of New York Press, 1995. Ferraro, Kenneth F., and Randy LaGrange. "The Measurement of Fear of Crime." Sociological Inquiry 57 (1987): 70-101. Furstenberg, Frank. F., Jr. "Public Reaction to Crime in the Streets." American Scholar 40 (1971): 601-610. Garofalo, James. Public Opinion About Crime. Washington, DC: U.S. Department of Justice, National Criminal Justice Information and Statistics Service, 1977. Groves, Robert M., and Robert L. Kahn. Surveys by Telephone: A National Comparison with Personal Interviews. New York: Academy Press, 1979. Kelling, George. "Measuring What Matters: A New Way of Thinking About Crime and Public Order." City Journal 2 (2) (Spring 1992): 21-34. Kelling, George, and Catherine Coles. Fixing Broken Windows. New York: Free Press, 1996. Kelling, George, Tony Pate, Duane Dieckman, and Charles Brown. The Kansas City Preventive Patrol Experiment: A Summary Report. Washington, DC: Police Foundation, 1974. Kraemer, Helena C., and Sue Thiemann. How Many Subjects? Statistical Power Analysis in Research. Newbury Park, CA: Sage Publications, Inc., 1987. Maxfield, Michael G. Fear of Crime in England and Wales. London, England: Home Office Research Study No. 78, 1984. Maxfield, Michael G. "The London Metropolitan Police and Their Clients: Victim and Suspect Attitudes." Journal of Research in Crime and Delinquency 25 (1988): 188-206. Mayhew, Pat, David Elliott, and Lizanne Dowds. The 1989 British Crime Survey. Home Office Research and Planning Unit Report No. 111. London, England: HMSO, 1989. Mirrlees-Black, Catriona, Pat Mayhew, and Andrew Percy. The 1996 British Crime Survey. London: Home Office Research and Statistics Directorate, 1996. National Criminal Justice Information and Statistics Service. Criminal Victimization in the United States 1976. Washington, DC: U.S. Department of Justice, National Criminal Justice Information and Statistics Service, 1976. Schneider, Anne L. "Victimization Surveys and Criminal Justice System Evaluation." In Sample Surveys of the Victims of Crime, ed. Wesley G. Skogan. Cambridge, MA: Ballinger Publishing Co., 1976: 135-150. Shapland, Joanna. "Preventing Retail-Sector Crimes." In Building a Safer Society: Strategic Approaches to Crime Prevention, Crime and Justice: A Review of Research, vol. 19, ed. Michael Tonry and David P. Farrington. Chicago: University of Chicago Press, 1995: 263-342. Skogan, Wesley G. Issues in the Measurement of Victimization. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics, 1981, NCJ 74682. Skogan, Wesley G. Disorder and Decline: Crime and the Spiral of Decay in American Cities. New York: Free Press, 1990a. Skogan, Wesley G. The Police and Public in England and Wales: A British Crime Survey Report. London, England: Home Office Research Study No. 117, 1990b. Skogan, Wesley G. "The Various Meanings of Fear." In Fear of Crime and Criminal Victimization, ed. Wolfgang Bilsky, Christian Pfeiffer, and Peter Wetzels. Stuttgart, Germany: Enke, 1993: 131-140. Skogan, Wesley G. Contacts Between Police and Public: Findings From the 1992 British Crime Survey. London, England: Home Office Research Study No. 134, 1994. Skogan, Wesley G. Resident Survey Methods Report. Program Paper No. 7. Evanston, IL: Institute for Policy Research, 1995. Skogan, Wesley G., and Andrew C. Gordon. "A Review of Detective Division Reporting Practices." In Crime in Illinois 1982. Springfield, IL: Illinois Department of Law Enforcement, 1983. Skogan, Wesley G., and Susan M. Hartnett. Community Policing, Chicago Style. New York: Oxford University Press, 1997. Skogan, Wesley G., et al. Community Policing in Chicago, Year One. Chicago: Illinois Criminal Justice Information Authority, 1994. Skogan, Wesley G., et al. Community Policing in Chicago, Year Two. Chicago: Illinois Criminal Justice Information Authority, 1995. Taub, Richard P., D. Garth Taylor, and Jan Dunham. Paths of Neighborhood Change: Race and Crime in Urban America. Chicago: University of Chicago Press, 1984. Taylor, Ralph B. "Going Over Grime Reduction: Conceptual and Practical Implications of Measuring and Distinguishing Community Disorder." Paper presented at the National Institute of Justice and Office of Community Oriented Policing Services Meeting, Measuring What Matters, November 28, 1995. Taylor, Ralph B. "The Impact of Crime on Communities." Annals of the American Academy of Political and Social Science 539 (May 1995): 28- 45. Trickett, Alan, Denise K. Osborn, Julie Seymour, and Ken Pease. "What is Different About High Crime Areas?" British Journal of Criminology 32 (1992): 81-90. Tyler, Tom R., and Fay L. Cook. "The Mass Media and Judgments of Risk: Distinguishing Impact on Personal and Societal Level Judgments." Journal of Personality and Social Psychology 47 (1984): 693-708. Van Dijk, Jan J.M., and Pat Mayhew. "Criminal Victimisation in the Industrialized World: Key Findings of the 1989 and 1992 International Crime Surveys." In Understanding Crime, Experiences of Crime and Crime Control, ed. Anna del Frate, Ugljesa Zvekic, and Jan J.M. van Dijk. United Nations Publication No. E.93.III.N.2. Rome, Italy: United Nations Interregional Crime and Justice Research Institute, 1993: 1-50. Warr, Mark. "Fear of Victimization: Why Are Women and the Elderly More Afraid?" Social Science Quarterly 65 (1984): 681-702. Wilson, James Q. "The Urban Unease: Community Versus the City." The Public Interest 12 (1968): 25-39. ------------------------------- Measuring What Matters Darrel W. Stephens In recent years, discussions of policing among practitioners and scholars have begun to emphasize the importance of outcome and impact measures. These discussions have pointed out that the police have developed a series of performance measures that, for the most part, have little relationship to results. James Q. Wilson, in "The Problem of Defining Agency Success," says it this way: Most of the efforts to improve performance measures for policing have concentrated on finding either real measures of overall effectiveness or plausible proxy measures. Not much has come of these efforts for reasons that should be obvious. There are no "real" measures of overall success; what is measurable about the level of public order, safety, and amenity in a given large city can only partially, if at all, be affected by police behavior. (For example, if the murder or robbery rates go up, one cannot assume that this is the fault of the police; if they go down, one should not necessarily allow the police to take credit for it.) Proxy measures almost always turn out to be process measures--response time, arrest rates, or clearance rates--that may or may not have any relationship to crime rates or levels of public order. (Wilson, 1993) Many practitioners and scholars would agree with Wilson. Nevertheless, the police continue to face the challenge of dealing with the impact of crime, fear, and disorder in their communities and the public's belief that it is their responsibility. The police are the first, and frequently the only, government agency the public looks to for answers when crime rates change, a heinous crime occurs, or citizens are afraid to go out of their houses after dark. Like many other aspects of their job, even when the police do not have a clear answer, there is an expectation that they say or do something that will provide a sense that things are either under control or will be in short order. A survey of Florida residents by the St. Petersburg Times (November 4, 1995) indicates that 85 percent of the respondents say the problem of greatest concern to them is crime. Over the past 10 to 15 years, national public opinion surveys routinely indicate that crime and drug abuse are among the highest priority concerns. The police are truly on the front line in dealing with the crime, fear, and disorder that have such a great impact on a community's quality of life. Before the police can address these problems, however, they face the significant challenge of measuring them. This challenge, along with the impact of these problems on the quality of community life, is the subject of this paper. The problems associated with measuring the levels of crime, fear, and disorder in the community are discussed in separate sections devoted to each of these areas, followed by an examination of the impact of these problems on the quality of life in the community. The concluding section discusses how these measures can be applied to specific neighborhoods in a way that allows the police to gain a sense of both the overall community problems and the efforts to deal with them. Measuring crime How do the police measure the level of crime in their community? For all intents and purposes the police measure the level of crime, and any change in crime, through the Federal Bureau of Investigation's Uniform Crime Reports (UCR). In many cities, monthly, quarterly, and annual reports are released to the public to show the number of serious crimes (Part 1 or index crimes) that citizens have reported to the police during each timeframe. These reports often provide comparisons to the same period in the previous year so anyone interested can see if reported crime has increased or decreased. News media stories about these crime statistics usually include quotes and sound bites from police representatives who attempt to explain any significant variations from one timeframe to the next. Occasionally, the stories include observations about the statistics from political figures and academic experts. Political figures are most often available to the media when crime reports are down from the previous reporting period. The UCR data represent the official level of crime in the community. These reports and the news media stories about them can have a significant impact on the community. They often serve as grist for the political mill--local elections have been greatly influenced by crime reports. In some cases, the careers of police chiefs and sheriffs have been affected in either positive or negative ways by these statistics. Because of their potential impact, UCR data have been the subject of considerable debate, discussion, and criticism as a measure of crime in the community. The criticism of the UCR has been focused primarily on a number of well-known limitations of the reporting system (Silberman, 1978; Kelling, 1996). First, the UCR represents only that portion of crime that is reported to the police. Although well known, this fact is not usually noted in either the reports provided by the police or the news media stories about them. In many residents' minds, these statistics represent the actual level of crime in their communities, particularly if there are significant increases from one year to the next. The second criticism is that only eight crimes have been included as Part 1 offenses. Crimes that the public cares a great deal about such as narcotic offenses are not included in the reports. Third, a series of program rules contribute to confusion about what the reports actually mean. For example, a bicycle or lawn mower stolen from an open garage is classified as a burglary. If these same items are stolen from the driveway a few feet from the open garage door, the offense is called a larceny. Some are also critical of the "hierarchy rule," which requires that an incident be classified as the most serious crime if multiple crimes occur at the same time. The fourth and perhaps most significant criticism is that crimes are reported to the police, who classify them, tabulate them, and send them to the State or directly to the FBI. Those suspicious of the police argue that this provides the opportunity for intentional manipulation of the numbers or mistakes in classification. The possibility of crime reports being manipulated by the police is not without some basis in fact. One example is the Kansas City, Missouri, police chief who had served with distinction for a number of years and was indicted by a county grand jury in 1960 for manipulating the UCR. The indictment was eventually dismissed, but he lost his job in the process. The same problem has surfaced in other cities over the years and continues to be one of the most significant concerns about crime reports. After all, there are subtle differences between attempted burglary and vandalism. A window might be broken in both, but there are different motives for each type of crime, and the motive may not be immediately clear. There is also a slim margin of difference between a strong-arm robbery and a purse snatching. It is clear when the victim is knocked to the ground in the process of taking the purse. In many cases, though, the difference is the degree of resistance involved in hanging onto the purse. There are similar distinctions that can be made in shoplifting cases where the suspect is confronted and resists apprehension. These are important issues because the seriousness of the crimes can be influenced by the benefit of the doubt going to the less serious incident. In the case of burglary or vandalism, if the latter classification is used, the incident drops completely out of the Part 1 crime category. All of the other limitations of the UCR are just that--limitations that need to be taken into account when using the data as a measure of crime. At the local level, intentional manipulation of the reports, however, is an entirely different matter. Manipulation of the reports renders them virtually useless as a measure of crime in a city. This, in turn, casts a dark shadow on the only measure of crime that most cities have and raises serious questions about the overall integrity of the police. Although local victimization surveys might be helpful, their cost puts them well beyond the ability of most police departments to conduct them with any regularity. Given the limitations of the UCR, how useful is it to the police and community as a measure of crime? In one sense, the question is academic: Until someone develops a suitable replacement, the UCR is the best available measure of reported crime--even with the flaws. A substitute for the UCR is not likely to be available anytime soon. An alternative system developed in the mid-1980s by the Police Executive Research Forum with the support of the Bureau of Justice Statistics failed to attract sufficient interest to serve as a viable replacement. No other initiatives are under way to develop a crime reporting and measurement system to take the place of the UCR. Therefore, it is important to reach a consensus on how significant the limitations of the UCR are to measuring crime in the community. If police departments pay close attention to proper collection and classification methods, the UCR can be a valuable and useful measure of reported crime. In fact, so much time is spent criticizing the system, little attention is given to the useful aspects of a reporting process that provides a good indication of the matters the public believes is important enough to bring to the attention of the police. Several aspects of the UCR provide helpful information to the police. One useful aspect is that it provides a relatively simple method of classifying criminal incidents that are brought to the attention of the police by the public. Even with the limitations, it provides a common language that most people, police officers and citizens alike, can understand. Using State statutory definitions presents some of the same problems as the UCR, and generally State definitions are more complex. For example, in some States, a burglary is limited to building structures; in others, a theft from a vehicle can be a burglary. State statutes contain many overlapping definitions for similar incidents, which can result in several criminal charges from one incident. A key criticism of the UCR is that it measures only the crime that is reported. That criticism would exist with any system unless it included victimization surveys, which are generally not practical for police departments. Moreover, one might want to explore just how valuable it would be for a police department to invest the resources to know what citizens have failed to report. How helpful would victimization data be for a police department? For the most part, knowing about every fight that takes place between two juveniles on the way home from school that might be classified as an assault is probably not particularly helpful to the police or the community. To be sure, most citizens will report what they believe is important for the police to know. If the police routinely encourage citizens to report incidents, what is reported might be a useful measurement of the level of crime in the community that the public believes is important for the police to know. Given the challenges of measuring crime, the UCR has been and can continue to be a useful way of measuring reported crime in a community. One of the greatest difficulties with the UCR is not the system itself but how the police and politicians use the information that comes from the system. Criticism of the UCR is loudest when reported crime is increasing. In spite of the cautions against comparisons from one city to another, it is done with great regularity, and it is naive to believe that will not continue. In fact, police, academics, and the news media regularly engage in the practice. The limitations of making such comparisons are rarely pointed out, except when reported crime is increasing. During these periods of increasing crime, it is often said that the primary reason the comparisons are not useful is because other cities may not give the same amount of attention to the accuracy of the reports. Although most police executives have learned to be cautious about what they say about UCR crime statistics when reported index offenses are declining, some are quite vocal about police contributions to the decline and look to the most recently implemented program as the source of the change. An important question that begs for some professional resolution in dealing with the issue of measuring what matters is who gets the credit--or the blame--for fluctuations in reported crime. Are police executives entitled to take credit for a decline in reported crime? If so, under what circumstances? While some in policing believe the police are essentially powerless to do much about crime, others argue that the police can make significant contributions to reducing crime in specific neighborhoods and circumstances. Focused, thoughtful responses to specific crime problems at the neighborhood level that involve those affected by the problem can contribute to reductions in reported crime. The police also should be able to accept some of the credit or responsibility for changes in reported crime. At the citywide level, it may be appropriate for the police to share in the credit for a decline in reported crime under at least two circumstances. First, the police should share in the credit if they address a problem in a small geographic area and changes in reported crime in the area affect the citywide totals. A good example of this is what happened with thefts from autos in the downtown area of Newport News, Virginia, in the mid-1980s. As a part of the department's problem-oriented policing effort, officers focused on the issue of thefts from vehicles parked in the area of the shipyard that employed more than 35,000 people. A careful analysis of the problem and the implementation of solutions tailored to the various aspects of the thefts resulted in a 52 percent decline in theft reports over a 12-month period (Eck and Spelman, 1987). That decline corresponded with a significant decline in the total number of thefts from vehicles in the city. While there are other possible explanations for this, it seems it is appropriate for the police to say this initiative is likely to have had some impact on the overall reduction in thefts from vehicles in the city. Moreover, since the larceny category was a major part of overall crime, it could be argued the subsequent decline in property index offenses was due in part to the initiative at the shipyard. It is also important to note in this example that the solutions implemented relied heavily on the contributions of others--the shipyard, the city, owners of the vehicles--to take steps to change the environment; thus, they should share in the credit for reducing the problem. Second, the police should share in the credit for declines in a specific crime on a citywide basis if they have implemented a specific response to the problem and the problem declines. Gasoline driveoffs have been affected by pay-before-you-pump policies advocated by police in many cities. In the mid-1970s, most urban areas enacted exact-change policies for public buses, and the once frequent bus robberies stopped. In neither case can other factors be ruled out because change and displacement influence overall numbers, but it seems appropriate for the police to accept some of the credit for the outcome. The UCR is perhaps the best available tool to address the question of how the police measure crime in a community. Given careful attention to the process and how the information is used by officials, some of the concerns can be addressed. In addition, the UCR can gain greater credibility, which might enhance its value. The UCR, however, has taken on a role as a measure of police impact that is well beyond what it should be--even if it works exactly as it was designed and everyone understands its limitations. Community measurements of crime and fear do not seem to be influenced to a great extent by the fluctuations in Uniform Crime Reports. The community uses other barometers. Measuring disorder How does the public measure crime? How much influence do official police reports have on citizen perceptions of crime? Do police annual rituals of providing UCR statistics to the public create a sense of relief or contribute to concern about crime? Part of the answer to these questions lies in how citizens define crime. Experience in working with citizens in a number of communities suggests that citizens define crime in very different terms than the police, and, by and large, official periodic pronouncements of the level of crime in the community have little influence on citizens' feelings about crime. In fact, these experiences lead one to believe the average citizen's perspective is influenced to a much greater extent by the amount of disorder they encounter, what they hear from friends and family members, their personal victimizations, and news media reports. The combination of these and other factors influence both their sense of the significance of the crime problem and their level of fear. Perceptions of disorder clearly seem to have an effect on citizens' views of crime and its impact on the quality of community life. Therefore, it is important for the police to define disorder, gain a better understanding of its influence on citizens' perceptions, and make stronger efforts at measurement. In "The Impact of Community Policing on Neighborhood Residents," Wesley G. Skogan looked at disorder through the use of survey questions that each of the projects included as a part of their evaluations (1995). The amount of disorder was determined by questions on public drinking, begging, street harassment, truancy, and gang activity. Surveying is one good way to understand citizens' views of disorder and its impact in a neighborhood or community. In fact, surveys of neighborhoods by the police in cooperation with residents are both practical and useful tools that are well within a department's capacity to conduct. There are other ways of measuring disorder as well. One helpful way to measure disorder is through simple observation of neighborhood or area conditions. It would not be difficult for police officers or motivated citizens to conduct a disorder assessment of the neighborhood by systematically recording what they see in a drive or walk through an area of concern. In St. Petersburg, neighborhood groups have volunteered to conduct surveys of residents as well as record the physical aspects of the area. If security is a concern, and it almost always is, they routinely walk the neighborhood at night to do an inventory of street lights, noting those that need repair as well as identifying locations where they believe additional lighting is needed. To measure disorder in a neighborhood, consideration might be given to the presence of graffiti, groups of people loitering on the street, the level of noise (from boom boxes or loud car stereo systems, for example), boarded and vacant structures, abandoned vehicles, homeless or street people, and litter. The presence of these elements in a neighborhood tends to contribute to a sense that the situation is out of control and to heighten the level of fear. The police also have an abundant source of information about disorder that would provide a sense of both its extent and location. Police call records, arrests, and reports are all good sources of information on public concerns about disorder (Skogan, 1990). Police call data is little used but is one of the best sources of information that police have about citizen concerns and their views of what police work should be. Calls about noise disturbances, street corner drug dealing, drinking on the street, graffiti, and gunfire are all good indications of pubic concern about disorder. Regular analysis of call information--frequency, type, location, and time--can give police a strong indication of the nature of the problems and, in some cases, insight into what might be done to improve the situation. Perhaps the greatest challenges for police in measuring disorder are to make it a priority and do what they can to change conditions. Wilson and Kelling's theory of "broken windows" is well accepted, and there is evidence that efforts to control disorder have some influence on the level of citizen fear, satisfaction, and reported crime (Houston, Newark, New York City, and St. Petersburg). However, it is often difficult for a street police officer to make the same connection. It is not because they do not have the intellectual capacity--they do. Police officers simply get caught up in the urgency of dealing with robberies, burglaries, auto thefts, and blatant street-level drug dealing. It is not easy for them to step back from the fray far enough to see the relationship between rowdy youths on the street corner, noise calls, and how those activities might contribute to the environment that produces the "real crime" they are most concerned about and believe is of greatest concern to the public. Although a challenge, disorder management is becoming a higher priority in many cities as the police make greater efforts to develop partnerships with the community to solve problems. Interaction with residents about neighborhood problems helps officers understand the importance of disorder to citizens' sense of safety. As police officers explore problems--and think about prevention and noncriminal justice responses--they begin to see the links between neighborhood conditions, fear, and crime. The development of a police department environment where officers have not only the expectation but also the opportunity to focus on problems in their areas of responsibility is critical. Police executives, managers, and supervisors have the obligation and responsibility to create this environment. With this environment comes the knowledge and understanding of the importance of measuring and responding to disorder problems. Fear Many would argue that the local government is as obligated to deal with the fear of crime as it is to deal with the actual incidence; that it is important, whatever the basis for existing fears, that citizens feel secure in their home and on their streets. (Goldstein, 1977) Over the past 20 years or so, it has become increasingly clear that the true mission of the police ought not to be "to protect and serve" but to help create a sense of safety in the community. To contribute to the production of safe communities, the police must both acknowledge and take steps to address citizen fear. This is a complicated task indeed, particularly because Skogan showed that the level of fear is not directly related to the risk of victimization (1986). Obviously, citizen surveys are the most helpful tool in measuring citizen fear and, like disorder, are within the capacity of the police to conduct on a neighborhood level. In fact, neighborhood surveys can be designed and conducted in a way that provides information on a variety of issues. The questions in exhibit 1 were included in surveys conducted in St. Petersburg that provided information on fear. While the information is not sufficient to understand the reason for the change in fear, it does give the police and citizens a sense of the level of fear and how it has changed over time. Although measuring fear is a bit more complicated for the police than measuring crime and disorder, data are available that would be helpful if viewed in the context of this problem. Once again, police calls can be a useful source of information about the level of fear in the community. Of particular importance are calls related to suspicious people and vehicles. Alarm calls might also serve as a crude measure of the level of fear in some areas. Alarm calls, particularly false alarm calls, have increased in most cities. While part of that increase is due to faulty systems, the rise in the use of both building and vehicle alarms has contributed to the increase as well. In some communities, ordinances have been enacted that require alarms for structures to be registered with the police. New alarm permits provide an indication of the level of fear in the community. In St. Petersburg, alarm permits increased almost 25 percent in the second year following the enactment of an ordinance requiring alarm systems to be registered. Looking at these data in concert with neighborhood survey data might identify areas where police can engage in specific activities to address citizen fear. While it may be difficult to capture, the investment in or presence of other security measures might be an indication of the level of fear in the community or neighborhood. The use of window bars, dead bolt locks, and demands for increased lighting provide some indication of the level of fear in a neighborhood. The police or other governmental agencies also have information on gun permits, security guard services, and off-duty police employment. All of these areas can provide some indication of the level of fear in the community and offer the potential for identifying specific areas where fear levels seem to be increasing. Although it is very difficult to measure, the impact of the news media, the entertainment industry, and police educational programs on citizen fear must be considered. The media obviously has some influence on how citizens feel about crime and violence and is, at least partially, responsible for contributing to citizen fear. When one considers the attention given to crime in both the print and electronic media, it is reasonable to conclude it affects the fear level in the community. In many metropolitan areas, local television news consumes from 4 to 6 hours of programming time. When combined with national news coverage, as much as a third of programming time is devoted to news. If the lead story is not devoted to crime, at least one of the top two or three stories is likely to deal with crime--generally the most violent or vicious of the day. In addition, a considerable portion of tabloid television shows are devoted to crime and violence. The steady diet of crime, murder, and mayhem reinforces daily the notion that there is good reason to be afraid. A significant portion of the television and movie entertainment industry is focused on crime and violence as well. The police shows like "COPS," "Stories of the Highway Patrol," and "America's Most Wanted" enjoy high ratings and add to the sense that crime and violence are completely out of control. This, of course, is an additional contribution to fear in the community. The police contribute to fear as well. With the best of intentions, the police have made the challenge of dealing with fear even more difficult. Police efforts to convince citizens of the importance of taking precautions to minimize their potential for victimization almost always begin with statistics or anecdotes about crime. The idea is to motivate citizens enough to take reasonable steps to protect themselves or their property. Unfortunately, these efforts have also caused additional fear; a police officer telling a citizen about the risks of crime has an extra amount of credibility. The clear challenge for the police is to educate citizens about their risk of criminal victimization in a way that motivates action--but does not unnecessarily increase their fear. The police must become more thoughtful and aggressive in providing information to the public to mitigate the effects of all the messages that promote fear. One tool that can be helpful is public cable television. Many cities have developed special programming designed to inform citizens about steps that can be taken to reduce the potential for victimization without living in fear. Police departments have also developed a range of methods to provide accurate information to citizens about crime in their neighborhoods. Some use telephone call-in systems allowing residents to access data 24 hours a day by entering the appropriate codes for their neighborhoods. Others provide periodic reports that are included in neighborhood newsletters. Still other departments have made crime and workload data available over the Internet. Many public newspapers in urban areas have returned to the practice of printing a police log that lists calls and crime reports by neighborhoods. The St. Petersburg Times lists crime reports and calls by community policing area in a biweekly neighborhood section. All of these tools are important to help members of the community be mindful of their potential for victimization but not so fearful that they become prisoners in their own homes. The effects of crime, disorder, and fear on the quality of community life What are the effects of crime, disorder, and fear on the quality of community life? Are the choices that people make on where to live, work, shop, or recreate influenced by their assessment of the risk of being a victim of crime? Fear is one effect of crime and disorder that clearly has an influence on how people live their lives. A USA Today poll indicated that 43 percent of Americans no longer shop at night because of the fear of crime. In a recent meeting, St. Petersburg car dealers concerned about crime indicated that citizen fear about the location of their businesses made it more difficult to attract both employees and customers. Concerns about safety in public schools have also had as much or more to do with parents placing their children in private schools than the quality of education. The fear of crime and disorder contributes to neighborhoods declining and dying because people are afraid to invest in them. Those who can afford it escape to the suburbs. Those who are not able to escape watch single-family houses turn into multiple-family dwellings that eventually get boarded up and demolished after absentee landlords reach the point where even minimal investments in meeting codes do not result in profits. Local governments wrestle with the dual problem of meeting increased service demands in these neighborhoods--fire protection, police service, code enforcement, environmental cleanups--while the revenue to support the services decreases. Measuring the effects of crime, disorder, and fear on the quality of life requires more than just measuring the levels of each of these variables. Once again, surveys can provide an indication of how crime, fear, and disorder affect individuals in the community. In many respects, "quality of life" is a difficult concept to understand. While there will be agreement on many aspects of what a good quality of life might include, individual perspectives will differ considerably. The fear a young man has about crime and disorder is likely to be very different from the fear of an elderly man. A person who is financially well off will not feel the same effects of crime and violence that a poor person will. The wealthy can simply move away from the problem or invest a small portion of income in creating a greater sense of security. Surveys can help sort out these various effects of crime and disorder on the quality of life. One can also monitor population shifts, property value changes, boarded and vacant properties, loss of public revenue, and similar variables that might provide some indication of the effects of fear, crime, and disorder. Another indication might be the willingness of the public to invest resources in public safety. The will to support get-tough policies continues to increase as more of the public treasury is devoted to the prison industry. A focus on neighborhoods When one thinks about crime, violence, drug abuse, fear, and all of the factors associated with them the problems seem overwhelming. The endless debate about what to do about these problems and who is responsible--individuals or society--takes place for the most part at the State or Federal level of government where the primary responsibility for many of the programs to address crime actually lies. And both of these levels of government are, for all intents and purposes, inaccessible to the general public. To effectively deal with crime and disorder and the fear they generate, it seems that a focus on neighborhoods or small geographic areas of the larger community offers the greatest promise of both understanding what is happening and doing something meaningful about these problems The police have been more willing in recent years to acknowledge their limitations in dealing with crime. They have begun to talk about crime and violence in the context of neighborhood conditions, education, the economy, and other demographic factors in areas with the greatest problems. Yet most police departments have not considered changes in these conditions as possible measures of their contributions. Fortunately, some police departments are beginning to look at these factors to determine the effect of initiatives aimed at neighborhood problems. One example is the appearance of the neighborhood. Building on the theory of "broken windows," police departments working with neighborhood associations, other arms of government, and the private sector have begun to consider change in the way a neighborhood looks as a positive impact of their collective efforts. An improvement in the way a neighborhood appears could translate into less fear or higher property values. Both of these variables can be measured at the neighborhood level as can the level of reported crime and amount of disorder. The efforts in St. Petersburg since 1992 have made an important contribution to property values in targeted neighborhoods. Exhibit 2 provides an indication of the change in property values from 1994 to 1996 in eight neighborhoods where citizens and local government developed and implemented specific plans to address problems of concern. The police played a key role in each of these neighborhoods because of the priority that citizens placed on security issues. What is the value of a new or expanded business in a neighborhood from the perspective of crime and disorder? Could new job opportunities help transform some individuals from criminal activities to legitimate forms of work? Can the foot and vehicular traffic associated with new business contribute to safer streets? Can police engage in programs or adopt policies that will enhance neighborhood improvement and investment? Is the police contribution to reducing truancy a valid measure of police performance, and how does that translate into reduced crime and disorder? Does an increase in occupancy of an apartment complex where police have worked on problems reflect a positive contribution? Obviously, the answers to these questions depend in part on the interventions police have initiated in cooperation with the community--but they also might provide greater insight into the ability of the police to affect crime and disorder and the fear they cause. Conclusion Measuring crime, disorder, fear, and their effects on the quality of life in the community is important to the police. It seems, nevertheless, more important to consider a wider range of issues to gain a true sense of the potential impact of the police on contributing to the creation of safe communities. It also appears that the police have the best chance of understanding these issues and making a meaningful contribution to dealing with them if the focus is on neighborhoods. At that level, even difficult, persistent problems do not appear to be quite so overwhelming. At that level, both the public and government can see visible signs of progress or the lack of it. Many baby boomers remember a time when their neighborhoods offered a sense of safety and security and neighbors rallied to provide support to each other in times of need. Many can recall a story of their youth where someone in the neighborhood intervened in a way that enforced standards of acceptable behavior--and then made sure that parents were aware of the incident. These baby boomers also point out that neighborhoods are not what they used to be. In spite of the changes in society, progress is being made in rebuilding neighborhoods and the sense of identity associated with them in cities throughout the United States. That experience suggests that crime, disorder, and fear can be influenced in a positive direction at the neighborhood level. We should be building on that experience. We should measure crime, disorder, and fear at the neighborhood level and develop tailored responses to deal with these problems. In that way, the police can make a substantial and meaningful contribution to the creation of safe communities. References Anderson, D.C. Crimes of Justice. New York: Timesbooks, 1988. Bennett, T. "Community Policing on the Ground: Developments in Britain." In The Challenge of Community Policing: Testing the Promises, ed. D.P. Rosenbaum. Thousand Oaks, CA: Sage Publications, Inc., 1994. Daly, N.C. "Community Survey Comparative Results." Unpublished Report. St. Petersburg, FL: St. Petersburg Police Department, 1994. Eck, J.E., and W. Spelman, with D. Hill, D.W. Stephens, J. Stedman, and J. Murphy. Problem Solving: Problem Oriented Policing in Newport News. Washington, DC: Police Executive Research Forum, 1987. Goldstein, H. Policing a Free Society. Cambridge, MA: Ballinger Publishing Company, 1977. Kelling, G. "Defining the Bottom Line in Policing: Organizational Philosophy and Accountability." In Quantifying Quality in Policing, ed. L.T. Hoover. Washington, DC: Police Executive Research Forum, 1996. Moore, M., and D.W. Stephens. Beyond Command and Control: The Strategic Management of Police Departments. Washington, DC: Police Executive Research Forum, 1991. Silberman, C.E. Criminal Justice, Criminal Violence. New York: Random House, 1978. Skogan, W.G. Disorder and Decline. New York: Free Press, 1990. Skogan, W.G. "The Impact of Community Policing on Neighborhood Residents: A Cross Site Analysis." In The Challenge of Community Policing: Testing the Promises, ed. D.P. Rosenbaum. Thousand Oaks, CA: Sage Publications, Inc., 1994. Skogan, W.G., and M.G. Maxfield. "Fear of Crime and Neighborhood Change." In Crime and Justice: A Review of Research, Communities and Crime, vol. 8, ed. A.J. Reiss and M. Tonry. Chicago: University of Chicago Press, 1986. Stephens, D.W. "Community Problem-Oriented Policing: Measuring Impacts." In Quantifying Quality in Policing, ed. L.T. Hoover. Washington, DC: Police Executive Research Forum, 1996. Taft, P.B., Jr. Fighting Fear: The Baltimore County C.O.P.E. Program. Washington, DC: Police Executive Research Forum, 1986. Wilson, J.Q. "The Problem of Defining Agency Success." In Performance Measures for the Criminal Justice System. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics, 1993, NCJ 143505. Wilson, J.Q., and G.L. Kelling. "Broken Windows: The Police and Neighborhood Safety." The Atlantic Monthly (March 1982). Wilson, J.Q., and G.L. Kelling. "Making Neighborhoods Safe." The Atlantic Monthly (February 1989).