Ideas in American POLICE FOUNDATION Policing July 1998

Evidence-Based Policing

By Lawrence W. Sherman

Most of us have thought of the statistician’s work as that of measuring Abstract and predicting . . . but few of us have thought it the statistician’s duty to try to The new paradigm of “evidence-based medicine” holds bring about changes in the things that he important implications for policing. It suggests that just doing [or she] measures. —W. Edwards Deming research is not enough and that proactive efforts are required to push accumulated research evidence into practice through national Ideas in American Policing presents and community guidelines. These guidelines can then focus in- commentary and insight from house evaluations of what works best across agencies, units, leading criminologists on issues of interest to scholars, practitioners, victims, and officers. Statistical adjustments for the risk factors and policymakers. The papers shaping crime can provide fair comparisons across police units, published in this series are from the Police Foundation lecture series of including national rankings of police agencies by their crime the same name. Points of view in prevention effectiveness. The example of domestic violence, for this document are those of the author and do not necessarily which accumulated National Institute of Justice research could represent the official position of the lead to evidence-based guidelines, illustrates the way in which Police Foundation. agency-based outcomes research could further reduce violence ©1998 Police Foundation and Lawrence W. Sherman. All rights reserved. against victims. National pressure to adopt this paradigm could Lawrence W. Sherman is come from agency-ranking studies, but police agency capacity to professor and chair of the adopt it will require new data systems creating “medical charts” Department of Criminology for crime victims, annual audits of crime reporting systems, and and Criminal Justice at the in-house “evidence cops” who document the ongoing patterns University of Maryland at and effects of police practices in light of published and in-house College Park. He was the research. These analyses can then be integrated into the NYPD Police Foundation’s director Compstat feedback model for management accountability and of research from 1979 to continuous quality improvement. 1985. Of all the ideas in policing, million in print (Sackett and Medicine, in fact, seems just one stands out as the most Rosenberg 1995)—as the highly as resistant to the use of evidence powerful force for change: police rigorous scientific evidence used to guide practice as are fields with practices should be based on to guide medical practices. He lower educational requirements, scientific evidence about what has suggested that policing such as policing. The National works best. Early in this century, should therefore be more like Institutes of Health (NIH) Berkeley Police Chief August medicine. Consensus Guidelines are a case Vollmer’s partnership with his Sherman was right about the in point. NIH convenes advisory local university helped generate need for many more randomized boards to issue to physicians this idea (Carte and Carte 1975), experiments in policing, but recommendations that are based which was clearly derived from wrong about how much medicine on intensive reviews of research that era’s expansion of the was really based on scientific evidence on specific medical scientific method into medicine, research. New evidence shows practices. These recommendations management, agriculture, and that doctors resist changing usually receive extensive publicity, many other fields (Cheit 1975). practices based on new research and are reinforced by mailings of While science had greater initial just as much as police do, if not the guideline summaries to some impact in those other professions more so. Closer examination one hundred thousand doctors. during the first half of the reveals medicine to be a But according to a RAND century, policing in recent battleground between research evaluation, doctors rarely change decades has been moving rapidly and practice, with useful lessons their practices in response to to catch up. However, any for policing on new ways to publication of these guidelines assessment of this idea in modern promote research. Those lessons (Kosecoff et al. 1987, as cited in policing must begin with an come from a new strategy called Millenson 1997). Thus three accurate benchmark: catching up “evidence-based medicine,”1 years after research found that to what? More complete evidence “widely hailed as the long-sought heart attack patients treated with on the linkage between research link between research and calcium antagonists were more and practice suggests a new practice” (Zuger 1997) to solve likely to die, doctors still paradigm for police improvement problems like the following prescribed this dangerous drug to and for public safety in general: (Millenson 1997, 4, 122, 131): one-third of heart attack patients. evidence-based crime prevention. Eight years after antibiotics were • An estimated 85 percent of For years, Sherman (1984, shown to cure ulcers, 90 percent medical practices remain 1992) and others have used of ulcer patients remained untested by research evidence. medicine as the exemplar of a untreated by antibiotics profession based upon strong • Most doctors rarely read the (Millenson 1997, 123–25). scientific evidence. Sherman has 2,500 medical journals available, and instead base their praised medicine as a field in Evidence Cops which practitioners have advanced practice on local custom. training in the scientific method • Most studies that do guide The struggle to change and keep up-to-date with the practice use weak, non- medical practice based on most recent research evidence by randomized research designs. research evidence has a long reading medical journals. He has history, with valuable implications cited the large body of for policing. In the 1840s, Ignaz randomized controlled Semmelweiss found evidence that 1 The term “evidence” in this mono- experiments in medicine—now graph refers to scientific, not criminal, maternal death in childbirth estimated to number almost one evidence. could be reduced if doctors

——2—— Increased pressure for “reinventing government” to One way to describe people focus on measurable results is reflected in the 1994 U.S. who try to apply research is the Government Performance Results Act (GPRA), which requires all role of “evidence cop.” federal agencies to file annual reports on quantitative indicators of their achievements. Education is under growing pressure to raise test scores as proof that children washed their hands before Hospital in Los Angeles. As are learning, which has led to delivering babies. He then tried to director of the hospital’s Center increased discussion of research apply this research to medical for Applied Health Services evidence on what works in practice in Vienna, which led to Research, Weingarten is an education (Raspberry 1998). And his being driven out of town by evidence-cop-in-residence. His the U.S. Congress has required his boss, the chief obstetrician. job is to monitor what the 2,250 that the effectiveness of federally Hundreds of thousands of women doctors are doing to patients at funded crime prevention died because the profession the hospital and to detect programs be evaluated using refused to comply with his practices that run counter to “rigorous and scientifically evidence-based guidelines for recommendations based on recognized standards and some forty years. The story shows research evidence. He does this methodologies” (House 1995, the important distinction between through prodding rather than sec. 116). All this sets the stage merely doing research and punishment, convening groups of for a new paradigm for making attempting to apply research to doctors who treat specific research more useful to policing redirect professional practices. maladies to discuss the research than it has ever been before. One way to describe people evidence. These groups then who try to apply research is the produce their own consensus Key Questions role of “evidence cop.” More like guidelines for practices that a traffic cop than Victor Hugo’s become hospital policy. Thirty- In suggesting a new paradigm detective Javert, the evidence five such sets of guidelines were called evidence-based policing, cop’s job is to redirect practice produced in Weingarten’s first there are four key questions to through compliance rather than four years on the job (Millenson answer: What is it? What is new punishment. While this job may 1997, 120). about it? How does it apply to a be as challenging as herding cats, What NIH, Weingarten, and specific example of police it still consists of pointing the 1995 founders of the new practice? How can it be professionals to practice “this way, journal called Evidence-Based institutionalized? not that way.” As in all policing, Medicine are all trying to do is to the success rate for this job varies push research into practice. Just What is it? widely. Fortunately, the initial as policing has become more Evidence-based policing is the failures of people like Semmelweiss proactive at dealing with crime, use of the best available research paved the way for greater success researchers are becoming more on the outcomes of police work in the 1990s. proactive about dealing with to implement guidelines and Consider Scott Wein- practice. This trend has developed evaluate agencies, units, and garten, M.D., of Cedars-Sinai in many fields, not just medicine. officers. Put more simply,

——3—— evidence-based policing uses far inferior to a systematic, public, revealing a strong practice research to guide practice and evidence-based checklist used by effect: the more operations evaluate practitioners. It uses the nurses. The mythic power of doctors and hospitals did each best evidence to shape the best subjective and unstructured year, the lower the risk-adjusted practice. It is a systematic effort wisdom holds back every field death rate. Using this clear to parse out and codify and keeps it from systematically correlation to push low-frequency unsystematic “experience” as the discovering and implementing surgeons and hospitals out of this basis for police work, refining it what works best in repeated tasks. business altogether, hospitals by ongoing systematic testing of A prime example of the were able to lower the death rate hypotheses. power of systematic, ongoing in these operations by 40 percent Evaluation of ongoing evaluations comes again from in just three years (Millenson operations has been the crucial medicine. In 1990, the New York 1997, 195). missing link in many recent State Health Department began Evidence-based policing is attempts to improve policing. If it to publish death rates for about two very different kinds of is true that most police work has coronary bypass surgery grouped research: basic research on what yet to go “beyond 911” by hospital and individual works best when implemented (Sparrow, Moore, and Kennedy surgeon. This action was properly under controlled 1990), the underlying reason may prompted by research showing conditions, and ongoing be a lack of evaluation systems that while the statewide average outcomes research about the that clearly link research-based death rate was 3.7 percent, some results each unit is actually guidelines to outcomes. It is only doctors ran as high as 82 percent. achieving by applying (or with that addition that policing Moreover, after adjusting for the ignoring) basic research in can become a “reflexive” or risk of death by the pre-operation practice. This combination creates “smart” institution, continuously condition of the patient caseload, a feedback loop (fig. 1) that improving with ongoing patients were 4.4 times more begins with either published or feedback. likely to die in surgery at the least in-house studies suggesting how The basic premise of successful hospitals than at the policing might obtain the best evidence-based practice is that we best hospitals. Despite enormous effects. The review of this are all entitled to our own opposition from hospitals and evidence can lead to guidelines opinions, but not to our own surgeons, these data were made taking law, ethics, and community facts. Yet left alone to practice individually, practitioners do come up with their own “facts,” which often turn out to be wrong. A recent survey of 82 Figure 1. Evidence-Based Policing. Washington State doctors found Literature

137 different strategies for ➤ treating urinary tract infections Best Guidelines (Berg 1991). No doubt the same ➤ Evidence ➤ result could be found for

➤ Outputs

handling domestic disturbances. In-House➤

A study evaluating the accuracy ➤

of strep throat diagnoses based ➤ on unstructured examination by ➤ Outcomes experienced pediatricians found it

——4—— culture into account. These guidelines would specify measurable “outputs,” or Evidence-based policing is practices that police are asked to follow. Their varying degrees of clearly different from, but success at delivering those outputs can then be assessed by very helpful to, all three tracking risk-adjusted “outcomes,” or results over a present paradigms of policing. reasonably long follow-up period. These outcomes may be defined in several different ways: offenses per 1,000 residents, repeat victimizations per 100 victims, outcomes research (like bypass use such outcomes to justify repeat offending per 100 surgery death rates). No other longer time spent on each call on offenders, and so on. The paradigm—not even NYPD’s the basis of an officer’s average observation that some units are Computerized Crime Comparison results, rather than issuing a getting better results than others Statistics (Compstat) strategy crude demand that he or she stay can be used to further identify (Bratton with Knobler 1998)— within an average time limit. It factors associated with success, uses scientific evidence to hold could also place much more which can then be fed back as professionals accountable for emphasis on learning how to deal new in-house research to refine results in peer-reviewed and even with each call most effectively the guidelines and raise the public discussions of outcomes and preventively, a question that overall success level of the agency. evidence. currently gets little attention. Such research could also be Evidence-based policing is Community policing, published in national journals or clearly different from, but very however defined, is not clearly at least kept in an agency helpful to, all three present linked to evidence about database as institutional memory paradigms of policing. Incident- effectiveness in preventing crime. about success and failure rates for specific policing, or 911 It is much more about how to do different methods. responses, currently lack any police work—a set of outputs— outcomes measure except time than it is about desired results, or What is new about it? out of service. Police officers who outcomes. Working with the Skeptics may say that there is take too much time to handle a community and listening to and nothing new in evidence-based call are sometimes accused of respecting community members policing, and that other shirking and are urged by are all important elements of the paradigms already embrace these supervisors to work faster.2 But paradigm. But that paradigm principles. On closer examination, no one tracks the rate of repeat alone has been easy for many however, we will see that no calls by officer or unit to see how officers to ignore. Adding the other paradigm contains the effective the first response was in accountability systems from the principles for its own preventing future problems. paradigm of evidence-based implementation. No other Evidence-based policing could policing could actually make paradigm contains a principle for police far more active in working both changing practices and with the community. 2 This sounds oddly like the pressure measuring the success of those for drive-in, drive-out childbirth health Problem-oriented policing is changes with risk-adjusted insurance now barred by federal law. clearly the major source for

——5—— evidence-based policing. Herman NYPD’s Compstat strategy controlled trial, both its strength Goldstein’s writings (1979, (Bratton with Knobler 1998) has and its limitations. The strength 1990), as well as John Eck and pushed the results accountability of the research design, pioneered William Spelman’s SARA model principle farther than ever before, in policing by the Police (1987), clearly emphasize but it has not used the scientific Foundation, is its ability to assessment of problem-solving method to assess cause and effect. reduce uncertainty about the responses as a key part of the Successful managers are average effects of a policy on vast process. Yet there is no clear rewarded, but successful methods numbers of people. The statement about the use of are not pinpointed and codified. limitation of the research design scientific evidence either in What evidence-based policing is that it cannot escape variability selecting strategies for responding adds to these paradigms is a new in treatments, responses, and to problems or in monitoring the principle for decision making: implementation. implementation and results of scientific evidence. Most police The variability of treatments those strategies (Sherman 1991). practice, like medical practice, is in policing is much like that in Reports on problem-oriented still shaped by local custom, surgery, which stands in sharp policing have so far produced opinions, theories, and subjective contrast to pharmaceuticals. little evidence either from impressions. Evidence-based While the chemical content of controlled tests or outcomes policing challenges those medical drugs is almost always research. Because the paradigm principles of decision making and identical, the procedural content stresses the unique characteristics creates systematic feedback to of surgery varies widely. Similarly, of each crime pattern, problem- provide continuous quality the style and tone each officer oriented policing has not been improvement in the achievement brings to a citizen encounter used to respond to highly of police objectives (see Hoover varies enormously and can make a repetitive situations like domestic 1996). Hence the inspiration for big difference in the outcome of assaults or disputes. Few this paradigm is not only a specific case. Dosage, timing, comparisons of different methods medicine and its randomized and follow-up of both drugs and for attacking the same problem trials, but also the principles of police work can vary widely in have been developed. Few officers quality control in manufacturing practice. are even held accountable for not developed by Walter Shewhart Even holding treatment implementing a problem-solving (1939) and W. Edwards Deming constant, there is evidence that plan they have agreed to (1986). These principles were both patients and offenders undertake. Problem-oriented initially rejected by U.S. business respond to treatments with wide policing has clearly revolutionized leaders, but were finally embraced variations. Some of these the way many police think about in the 1980s after Japanese responses, allergic reactions, can their objectives, moving them industries used them to far kill some people with treatments away from a narrow focus on surpass U.S. manufacturers in the that cure most others. Offenders each incident to a broader focus quality of their products. are known to vary in their on patterns and systems. But in What makes both policing responses to police actions by the absence of pressure from an and medicine different from individual, neighborhood, and evidence-based approach to manufacturing, of course, is the city. And implementation of new evaluating success and far greater variability in the raw practices based on controlled management accountability, material to be processed—human experiments in both medicine and problem-oriented policing has beings. That is what gives the policing varies according to how been kept at the margins of gold standard of evaluation well research is communicated, police work. research, the randomized how much information is created

——6—— about whether practices actually research has also shown that, like treated with courtesy and change, and how much surgery, police practices vary politeness. Do these variations on reinforcement there is for the greatly in their implementation. the theme of arrest make a change, both positive and These variations in practice cause difference? They should, negative. varying results for repeat according to the “defiance” Evidence-based policing offending against victims. Even theory of criminal sanction effects assumes that experiments alone holding practice constant, (Sherman 1993). And they did in are not enough. Putting research responses to arrest vary by Milwaukee, according to into practice requires just as offender, neighborhood, and city. Raymond Paternoster and his much attention to Finally, research shows very poor colleagues (1997). The implementation as it does to compliance with mandatory arrest Milwaukee evidence reveals that controlled evaluations. Ongoing guidelines after they are adopted controlling for other risk factors systems for researching (Ferraro 1989). among some 800 arrested implementation can close the There are many varieties of offenders, those who felt they feedback loop to create the arrest for misdemeanor domestic were not treated in a procedurally principle of industrial quality violence. The offender may or fair and polite manner were improvement. may not be handcuffed, arrested 60 percent more likely to commit in front of family and neighbors, a reported act of domestic How does it apply to a specific given a chance to explain his violence in the future (fig. 2). example of police practice? version of events to the police, or This finding suggests three ways The policing of domestic violence offers a clear illustration of what is new about the evidence-based paradigm. Figure 2. Repeat Domestic Violence and Police Fairness. Domestic violence has been the 50% subject of more police practices research than any other crime problem. The research has 40% 40% arguably had little effect on police practice, at least by the new standards of evidence-based medicine. Yet the available 30% evidence offers a fair and 25% scientifically valid approach for holding police agencies, units, 20% and officers accountable for the results of police work, as measured by repeated domestic violence against the same victims. 10% The National Institute of Justice (NIJ) and the Police Foundation have provided 0% policing with extensive Fair Unfair information on what works to Source: Paternoster, et al. prevent repeated violence. The

——7—— to push research into practice: findings could be changed by 1996), could be used to fashion 1) change the guidelines for further research, but for the new problem-oriented strategies. making domestic violence arrests moment they are the best Most important, the existing to include those elements that evidence available. research can be used to create a would enable offenders to This research evidence could fair system for evaluating police perceive more “procedural support guidelines for policing performance on the basis of risk- justice”; 2) hold police domestic violence that differed by adjusted outcomes. That evidence accountable for using these neighborhood and absence or (fig. 3) shows that the likelihood guidelines by comparing rates of presence of the offender. It could of a repeat offense is strongly repeat victimization associated also support guidelines about linked to the number of previous with different police units; and listening to suspects’ side of the offenses each offender has. 3) compute these rates using story before making arrest Once the risk of repeat statistical adjustments for the pre- decisions and generally treating offending can be predicted with existing level of recidivism risks. suspects with courtesy. Other reasonable accuracy, it becomes The NIJ research provides evidence, such as the extremely possible to use those predictions other evidence for ways that high-risk period for repeat as a benchmark for police police can reduce repeat victimization in the first days and performance. Just as in the bypass offending in misdemeanor weeks after the last police surgery death rates in New York, domestic violence. Rather than a encounter (Strang and Sherman the outcomes of policing can be one-size-fits-all policy, the evidence suggests specific guide- lines to be used under different conditions. Offenders who are Figure 3. Risk of Repeat Domestic Assault by Priors. absent when police arrive—as 80 they are in some 40 percent of 75% cases—respond more effectively to arrest warrants than offenders who are arrested on the scene 60% 60 (Dunford 1990). Offenders who are employed are deterred by 48% arrest, while offenders who are unemployed generally increase 42% 40 their offending more if they are arrested than if they are handled in some other fashion (Pate and Hamilton 1992; Berk et al. 1992; 20 Sherman and Smith 1992). Offenders who live in urban areas of concentrated poverty commit more repeat offenses if they are 0 arrested than if not, while 01 237 offenders who live in more affluent areas commit fewer Percent Repeats repeat offenses if they are arrested Milwaukee Domestic Violence Experiment (Marciniak 1994). All of these

——8—— By constructing information Figure 4. Observed vs. Expected Risk of Repeat systems for this kind of outcome Domestic Violence. research, police departments can 60 focus on an objective that has only previously been measured in 50% major experiments. Making the 50 goal of policing each domestic assault the outcome of a reduced 40 repeat offending rate rather than the output of whether an arrest is 30 made would have several effects. 25% One is that crime prevention would get greater attention than 20 retribution for its own sake. While not everyone would 10 welcome that, it is consistent with at least some police leaders’ view 0 of the purpose of the police as a Observed Expected crime prevention agency (Bratton with Knobler 1998). Another Percent Repeat effect would be to seek out and

controlled for the risk level inherent in the caseload they face. Figure 5. Observed vs. Expected Ranking by Precinct. Using a citywide database of all 200 domestic assaults, now running over ten thousand cases per year 150% in cities like Milwaukee, a model 150 can be constructed to assess the risk of repeat offending in each 100 case. The overall mix of cases in each police precinct or for each 50% officer can generate an average 50 risk level for that caseload. Each police patrol district can then be 0 evaluated according to the actual versus predicted rate of repeat –25% -50 offending each year (fig. 4). All –50% patrol districts in the city can then be compared on the basis of -100 their relative percentage PCT 1 PCT 2 PCT 3 PCT 4 PCT 5 difference between expected and actual rates of repeat domestic Percent Repeat assault (fig. 5).

——9—— allow stockbrokers to rank publicly-held corporations and The strongest claim about provide those companies with strong incentives for better evidence-based policing is that results, public information about police performance would create it contains the principles of the strongest pressure for improvement.3 its own implementation. One example of how the major city police departments could be ranked on performance can be found in their homicide rates, which already receive even initiate more research on above would have to occur one extensive publicity. What these what works best to prevent police agency at a time, there are statistics lack, however, is any domestic violence. In the world certain national forces that can scientific analysis of expected risk. as we now know it, no one in help start the ball rolling. This Police performance has nothing policing—from the police chief to can be seen, for example, in to do, at least in the short run, the rookie officer—has any direct national rankings of big-city with the social, economic, incentive to reduce repeat police agencies, as well as national demographic, and drug market offending against known victims. mandates for improving police forces that help shape a city’s No one in policing is held data systems to provide better homicide rate. While police accountable for accomplishing, or evidence. Yet even such external performance may also affect those even measuring, that objective. pressures will not succeed homicide rates, the other factors As a result, no one knows without internal evidence cops to must be taken into account. whether repeat victimization rates import, apply, and create research Using risk-adjusted homicide get better or worse from year to evidence. rates provides one indication of year. Using outcomes evidence to No institution is likely to how well a police department evaluate performance would make increase voluntarily its may be doing things like police practices far more victim- accountability except under confiscating illegal weapons, centered, the top priority being strong external pressure. It is patrolling hot spots, regulating that of preventing any further unlikely that evidence-based violent taverns and drug markets, assaults. policing could be adopted by a and monitoring youth gangs. police executive simply because it While the basic research literature How can it be appears to be a good idea. The would increasingly provide a institutionalized? history of evidence-based source of guidance for taking The strongest claim about medicine and education strongly initiatives against homicide, a evidence-based policing is that it suggests that professionals will contains the principles of its own only make such changes under implementation. The principles of external coercion. Nothing seems 3 The 1919 results of the first using evidence both to change to foster such pressure as much as national rankings of hospitals were and evaluate practice can be performance rankings across deemed so threatening that the American applied to a broad institutional agencies (Millenson 1997; College of Surgeons decided to burn the report immediately in the furnace of analysis of implementation. Thus Steinberg 1998). Just as various New York’s Waldorf-Astoria Hotel while the changes described public performance measures (Millenson 1997, 146).

—— 10 —— risk-adjusted outcomes analysis alleged (McAllister 1998). Other possible exception of homicide. (fig. 6) would indicate how well examples include teachers helping No two agencies classify crime that research had been put into students to cheat on their answers the same way. The same event practice.4 to national achievement tests and, may be called an aggravated If a credible national research of course, police departments assault in one agency and a organization would produce such under-reporting crime. The New “miscellaneous incident” in “league rankings” among big-city York City police have removed another. The recent FBI decision police departments each year (like three commanders in the past five to drop Philadelphia from the the U.S. News & World Report years for improperly counting national crime reporting program rankings of colleges and crime to make their performance was not an isolated action. In universities), the predictable look better (Kocieniewski 1998), 1988, the FBI quietly dropped result in the short term would be and several chiefs of police the entire states of Florida and attacks on the methodology used. elsewhere have been convicted on Kentucky. Since the FBI lacks That is, in fact, what continues to criminal charges for similar resources to do on-site audits in go on in New York with the conduct. each police agency every year, death rates in surgery. But the Quite apart from pressures to these examples are just the tip of New York rankings have spread corrupt data, criminologists have a very big iceberg. There are to other states, and consumers long known that police crime already rising suspicions of police have found them quite valuable. reporting is not reliable, with the manipulation of crime data as Doctors—and police—may also find rankings very valuable in the long run. Both professions should enjoy greater public respect as they get better at producing the Figure 6. Homicide by City, Actual vs. Predicted. results their consumers want. 80 The more seriously 60% performance indicators influence 60 the fate of organizations, the more likely they are to be 40 subverted. Recent examples include the U.S. Postal Service in 25% West Virginia, where an elaborate 20 scheme to defeat the on-time mail delivery audit was recently 0

-20 –25% 4 While many of the basic risk factors would be computed from Census data -40 that could be out of date by the middle of each decade, other risk data can be –50% derived from annually updated sources, -60 such as the NIJ ADAM data on drug NYC Baltimore Balt. Chicago Chi LA Dallas abuse among arrestees. Unemployment, school dropout, teen childbirth, and infant mortality data are also available Percent Difference annually for each city and could help Hypothetical Data predict the expected rate of homicide.

—— 11 —— crime rates fall in many cities. More serious pressure from national rankings would threaten In the process of revitalizing data integrity even more. One viable solution to this crime data integrity, there problem is a federal requirement for police departments to retain would be great value in CPA firms to produce annual audits of their reported crime reorganizing police data data. This requirement could be imposed as a condition for systems. receiving federal funds, just as many other federal mandates have already done. Anticipating court challenges about unfunded mandates (such as the Brady Bill), description) for each incident a attempts to reduce repeat Congress could also provide victim presents to a police agency, offending at those places. Similar funds to pay for the audits. perhaps anywhere in the state. records could be kept about a Crime counting standards could The chart would also show what pattern of crimes spread out be set nationally by the police did in response, everything across a wider area, such as accounting profession in from taking an offense report to automatic teller machine collaboration with the FBI. arresting an offender whose robberies. If officer teams or units Alternatively, each state legislature release date from prison is also identify these places or patterns as could require (or even fund) kept, updated, in the crime targets and designate a these audits as a means of computerized victim chart. This control group, these medical assuring fairness in performance information tool could help charts can become the basis for rankings of police departments develop many proactive police estimating how much crime each within the state. State agencies methods for preventing repeat police unit has prevented. such as the criminal justice victimization. Allowing officers to Computers can also help statistical centers could also use these data to keep their own police officers to implement produce such rankings as a private “batting averages” for practice guidelines. Medical service to taxpayers. States already repeat victimization (even computer systems now offer have the option of spending without adjusting for risk) may recommended practice guidelines federal funds on such a purpose encourage them to become in response to a checklist of data, under the broad category of involved and committed to doing as well as warning when drug evaluation funds. a better job at preventing crime. prescriptions fall outside In the process of revitalizing Better records are also needed programmed parameters of crime data integrity, there would about what police do about crime disease type and dosage. The use be great value in reorganizing according to certain patterns of of hand-held computers to advise police data systems. Most offenses. “Medical charts” for officers in the field and to provide important would be the creation violent taverns, frequently robbed instant quality control checks may of a “medical chart” for each convenience stores, and other hot not happen soon, but the growth crime victim. Like computerized spots where most crime occurs of police research may make it patient records, this chart would would be very useful for ongoing inevitable in the long run. show the diagnosis (offense problem-oriented policing Doctors are not expected to keep

—— 12 —— large amounts of research data in We may also find that the in which university scientists visit their heads, nor even medical traditional distance between farms and show farmers new guidelines for each diagnosis. researchers and police officials techniques for improving their Computers will not replace good shrinks when researchers provide crop yields. They echo a Chinese judgment, but they can clearly more immediate managerial proverb: Tell me and I will enhance it. information. Criminologists have forget; show me and I will Federal rules could also long refused to provide police remember; involve me and I will require police departments to managers with data on particular understand. appoint a certified police officers, deeming it contrary to The one test of this principle criminologist (either internally or the ethics of basic research in policing to date is Alex Weiss’s in partnership with a university or (Hartnett 1998). By finally (1997) research on how police research organization), who providing the data in a departments adopt innovations. would become the agency’s scientifically reasonable format, Based on a national survey of evidence cop. Like Scott criminologists may become far police chiefs and their top aides, Weingarten of Cedars-Sinai, the more effective at pushing research Weiss discovered that telephone departmental criminologist would into practice. calls from agency to agency be responsible for putting Criminologists can also act on played a vital role in spreading research into practice, then the finding that doctors tend to new ideas. While written reports evaluating the results. Whether change practices based on may have supplemented the the criminologist is actually an personal interaction and repeated phone calls, word-of-mouth employee or a university professor computerized feedback, and not seems to be the major way in working in partnership with the from conferences, classes, or which police innovations are police may not matter as much as written research reports communicated and adopted. the role itself. The criminologist (Millenson 1997, 127–30). Weiss’s study suggests the could help develop more effective Similar findings have been great importance of gathering guidelines for preventing repeat published about the effectiveness more evidence on evidence. The offending, and could develop of agricultural extension services, empirical question for research is, expected versus actual repeat what practices work best to offending data by offense type for change practices? This inherently each police district or detective reflexive posture may lead us to unit. A criminologist could add empirical comparisons of the the scientific method to the The empirical effectiveness of, for example, NIJ NYPD Compstat process conferences, mass mailings of (Bratton with Knobler 1998), question for research-in-brief reports, or new providing statistics at each one-on-one approaches. One meeting on each patrol district’s research is, example of the latter would be crime trends and patterns (or proactive telephone calls to police even its complaints against police what practices agencies around the U.S. made officers) in relation to the by present or former police district’s risk level. Building the work best to officers; callers could be trained capacity to import, apply, and by research organizations to create evidence within each police change describe new research findings. If agency may be an essential national consensus guidelines for ingredient in the success of this practices? practice were developed by panels paradigm. of police executives and

—— 13 —— researchers, the callers could American Board of Family Goldstein, Herman. 1979. communicate those as well. Other Practice (September– Improving policing: A approaches worth testing might October): 327–30. problem-oriented approach. include field demonstrations in Berk, Richard A.; Alec Campbell; Crime and Delinquency 25: police technique. This training Ruth Klap; and Bruce 236–58. would not be based on Western. 1992. The deterrent ———. 1990. Problem-oriented experience, as is the current Field effect of arrest in incidents of policing. New York: McGraw- Training Officer system, but domestic violence: A Bayesian Hill. rather it would be based on analysis of four field Hartnett, Susan. 1998. Address evidence that the method being experiments. American to the Third National demonstrated has been proven Sociological Review 57: 698– Institute of Justice effective in reducing repeat 708. Conference on Police- offending. Bratton, William, with Peter Research Partnerships, Knobler. 1998. Turnaround: February. Conclusion How America’s top cop Hodge, Melville H. 1990. Direct reversed the crime epidemic. use by physicians of the TDS The test of this paradigm’s New York: Random House. Medical Information System. results is not whether it is Carte, Gene, and Elaine Carte. In A History of Medical adopted this year or in twenty 1975. Police reform in the Informatics, edited by years. As Lord Keynes has United States: The era of Bruce I. Blum and Karen suggested, the influence of ideas August Vollmer. Berkeley: Duncan. New York: ACM. may be far more glacial than University of California Press. Kosecoff, Jacqueline, et al. 1987. volcanic. The pressure for better Cheit, Earl. 1975. The useful arts Effect of the National measures of results is in the spirit and the liberal tradition. New Institutes of Health of the age, and police cannot York: McGraw-Hill. Consensus Development long escape it. All this paper does Deming, W. Edwards. 1986. Out Program on physician is add one inch to the glacier, so of the crisis. Cambridge: practice. Journal of the that we can say of policing what Massachusetts Institute of American Medical Association Dr. William Mayo of the Mayo Technology, Center for 258 (November 20): 2708– Clinic said of his profession Advanced Engineering Study. 13. almost a century ago: “The glory Dunford, Franklyn. 1990. McAllister, Bill. 1998. A “special” of medicine is that it is constantly System-initiated warrants for delivery in West Virginia: moving forward, that there is suspects of misdemeanor Postal employees cheat to always something more to learn.” domestic assault: A pilot beat rating system. study. Justice Quarterly 7: Washington Post, 10 January, 631–53. A1. References Eck, John, and William Spelman. Marciniak, Elizabeth. 1994. Berg, A.O. 1991. Variations 1987. Problem-solving: Community policing of among family physicians’ Problem-oriented policing in domestic violence: management strategies for Newport News. Washington, Neighborhood differences in lower urinary tract infections D.C.: Police Executive the effect of arrest. Ph.D. in women: A report from the Research Forum. diss., University of Maryland. Washington Physicians’ Ferraro, Kathleen J. 1989. Millenson, Michael L. 1997. Collaborative Research Policing woman battering. Demanding medical excellence: Network. Journal of the Social Problems 36: 61–74. Doctors and accountability in

—— 14 —— the information age. Chicago: the need for evidence-based Sparrow, Malcolm; Mark Moore; University of Chicago Press. medicine. Health Economics 4: and David Kennedy. 1990. Office of Technology Assessment 249–54. Beyond 911: A new era for of the Congress of the United Sherman, Lawrence W. 1984. policing. New York: Basic States. 1983. The impact of Experiments in police Books. randomized clinical trials on discretion: Scientific boon or Steinberg, Jacques. 1998. Public health policy and medical dangerous knowledge? Law shaming: Rating system for practice. Background paper and Contemporary Problems schools: Some states are OTA-BP-H-22. Washington, 47, no. 4: 61–81. finding that humiliation leads D.C.: Government Printing ———. 1992. Policing domestic to improvement. New York Office. violence: Experiments and Times, 7 January, A19. Pate, Antony M., and Edwin E. dilemmas. New York: Free Strang, Heather, and Lawrence Hamilton. 1992. Formal and Press. W. Sherman. 1996. Predicting informal deterrents to ———. 1993. Defiance, domestic homicide. Paper domestic violence: The Dade deterrence and irrelevance: A presented to the American County Spouse Assault theory of the criminal Association for the Experiment. American sanction. Journal of Research Advancement of Science, Sociological Review 57: 691– in Crime and Delinquency 30: January, in Baltimore, 98. 445–73. Maryland. Paternoster, Ray; Bobby Brame; ——— and Douglas A. Smith. U.S. Congress. House. 104th Ronet Bachman; and 1992. Crime, punishment and Congress, 1st sess., H. Rept. Lawrence W. Sherman. 1997. stake in conformity: Legal 104-387, sec. 116. Do fair procedures matter? and informal control of Weiss, Alexander. 1992. Diffusion Procedural justice in the domestic violence. American of innovations in police Milwaukee Domestic Violence Sociological Review 57. departments. Ph.D. diss., Experiment. Law and Society Shewhart, Walter A. 1939. Northwestern University. Review. Statistical methods from the Zuger, Abigail. 1997. New way Raspberry, William. 1998. Tried, viewpoint of quality control. of doctoring: By the book. true and ignored. Washington Edited by W.E. Deming. New York Times, 16 Post, 2 February, A19. Lancaster, Pennsylvania: December, C1. Sackett, David L., and William Graduate School of the U.S. M.C. Rosenberg. 1995. On Department of Agriculture.

—— 15 —— ABOUT THE POLICE FOUNDATION The Police Foundation is a private, independent, not-for-profit organization dedicated to supporting innovation and improvement in policing through its research, technical assistance, and communications programs. Established in 1970, the foundation has conducted seminal research in police behavior, policy, and procedure, and works to transfer to local agencies the best new information about practices for dealing effectively with a range of important police operational and administrative concerns. Motivating all of the foundation’s efforts is the goal of efficient, humane policing that operates within the framework of democratic principles and the highest ideals of the nation.

OFFICE OF RESEARCH BOARD OF DIRECTORS

David Weisburd, PhD Chairman Senior Research Scientist William G. Milliken Rosann Greenspan, PhD Research Director President Patrick R. Gartin, PhD Hubert Williams Senior Research Associate David G. Olson, PhD Freda Adler, PhD Senior Research Associate Lee P. Brown, PhD Edwin E. Hamilton, MA Senior Research Analyst William H. Hudnut III Michael Clifton, MA Research Associate W. Walter Menninger, MD Jennifer C. Nickisch, MA Research Associate Victor H. Palmieri Justin Ready, MA Research Associate Henry Ruth Annette C. Miller, MA Stanley K. Sheinbaum Research Assistant Rachel Dadusc, BA Alfred A. Slocum Administrative Assistant Sally Suchil

Kathryn J. Whitmire

POLICE FOUNDATION 1201 Connecticut Avenue, NW, Washington, DC 20036 (202) 833-1460 • Fax: (202) 659-9149 • e-mail: pfi[email protected]

—— 16 ——