Plenary Presentation to the American Society of Criminology

Jeremy Travis
Director, National Institute of Justice

November 12, 1998
Washington, D.C.

Dear Colleagues:

I am pleased to have this opportunity to speak to you about the important issue of the relationship between research and policy and practice. This is a timely topic, particularly since the American Society of Criminology has chosen as its theme this year "Crime, Justice and Public Policy: Examining Our Past and Envisioning Our Future" and since this conference is being held in the Nation's capital, the city where policy-making is the main industry and research is a major activity.

This year we are mindful of a number of historical markers. This year we celebrate the 50th anniversary of the American Society of Criminology's annual meeting; this year also marks the 30th anniversary of the establishment of the National Institute of Justice. So, as this anniversary conference celebrates the growth of the field of criminology, it is appropriate to ask what contributions this field has made to the development of criminal justice policy. And, as the National Institute of Justice enters its fourth decade, after a period of unprecedented growth my colleagues at NIJ and I are once again focused on the value that our research activities provide to the development of sound policy and practice.

I am intrigued by the insistence of the question itself. Why do we expect that there will be any relationship between research and practice? Why do we (or at least some of us) feel that we have failed if the translation between research and practice does not occur? What exactly do we envision when we posit a relationship between these two distinct and different activities?

This afternoon I would like to construct a case for a new relationship between researcher and practitioner — a relationship of constructive engagement, of partnership in the development of useful knowledge, of symbiosis in the testing of ideas. I hasten to add that I do not wish to be understood as not supportive of basic research — on the contrary, I believe that we ought to expand our commitment to basic research. We see this new relationship between research and practice in embryonic form around the country and see its value, both to the research endeavor and to the development of policy. But before setting forth the elements of this new partnership model, I would like to explore with you the landscape of possible relationships between research and practice.

I suppose we could imagine a world in which the academic community and the practice community did not communicate, except to criticize each other. We could imagine a world in which the academic community goes about its research, maintaining appropriate distance from the nitty-gritty world of practice, publishing in academic journals, and criticizing the crime policies of the country. In this same world, we could envision a community of practitioners who go about their business, doing what they thought best, criticizing the aloofness of academe, and ignoring the findings of research. Yet we clearly do not want to live in this world — we want to see some interaction between research and practice. Is that because we think the quality of practice is improved by exposure to research? Or the quality of research is improved by exposure to practice? Or both?

Before addressing those questions, let's quickly ask ourselves whether we want to live in a world where there is total congruence between research and practice. Can we imagine a world in which the criminal justice policies of the Nation are determined by science? How would those policies actually be developed? What would they look like? Could we expect a scientific consensus about how to handle sex offenders? How to deter juvenile offending? How to structure criminal sanctions? How to deploy police resources? How to respond to child abuse?

Certainly we would face, at a minimum, the inevitable reality of a lack of consensus in the research community about some of these policies — it would, of course, be silly to resolve these conflicts by taking a vote of the Board of Directors of the ASC to determine our society's criminal justice policies. But, more fundamentally, we must recognize at the outset that the process of making policies in this and other areas of public life is necessarily messy and far removed from the discipline of scientific inquiry, involving difficult choices between competing claims on public goods and governmental authority, requiring democratic consensus to achieve political legitimacy, and necessitating that action be taken before objective evaluation is possible.

So, if neither world seems possible or desirable — if we prefer a world characterized by some engagement between research and practice — how should that engagement be structured? Are we close to achieving the right relationship? What are the appropriate divides between these two activities? Thus, the question we should address today is how those relationships should be structured, and how they are evolving.

Thirty years ago, when the President's Commission on Law Enforcement and Administration of Justice recommended the creation of the federal agencies that today are the Bureau of Justice Statistics and the National Institute of Justice, the rationale set forth in the Commission's report was quite clear: The "revolution of scientific discovery has largely bypassed the problems of crime and crime control."1

Those words could be spoken today with equal conviction — yet we should acknowledge that the world has changed. Over the intervening thirty years, the capacity of the Nation's researchers to implement this vision of scientific relevancy to crime policy has expanded enormously. Other disciplines — health, economics, political science, education, and geography are playing increasing roles. We have seen quantum increases in the numbers of criminology and criminal justice departments in our universities. Other disciplines — health, economics, political science, education, geography — are playing an increasing role. We have seen operations research units institutionalized within police departments and other criminal justice agencies. We have watched as the systems for measuring crime have become more sophisticated with the maturation of victimization surveys, indicators of fear and disorder, collective efficacy and other measures of community resiliency, hot spots, and geographic analysis.

Most importantly, we are experiencing an increased demand to know what works. There is a constant call from practitioners to know what works to reduce crime. They have a great faith in the research community to tell them whether a program or policy is effective. At NIJ, we have been very gratified to see the response to the report the Institute commissioned by the University of Maryland to determine what works, what doesn't and what's promising in the area of crime prevention. This report has had continuing reverberations in the Congress, throughout the states, in the professional associations, and with the public at large — even internationally. Although the Maryland report cannot claim credit for all of these developments, we should note the following. Congress has required that the DARE program react to research findings that its curriculum does not reduce drug abuse. The clamor for boot camps seems to have subsided. There is strong interest around the country in early childhood programs, particularly home nurse visitation programs. The Department of Education is limiting its funding under the Safe and Drug-Free Schools program to programs of proven effectiveness. Drug treatment has experienced increased support due to research. Congressional funding for evaluations of crime programs has increased significantly and, as a result, the research portfolio of the National Institute of Justice has increased four-fold in four years. In sum, although we have a long way to go, this is commendable progress over a short period of time.

Although this is all good news, I think we need to be careful that we do not promise more than we can deliver. Ironically, I have come to believe that our faith in science and our reliance on the scientific method sometimes limits our ability to influence practice. Sometimes we lapse into the erroneous belief that crime policy development should resemble the relationship between the Food and Drug Administration and the pharmaceutical industry — that researchers, using the most rigorous scientific models available, should pronounce a particular policy or practice to be effective and then it should be adopted throughout the practitioner community. This model, in my view, has some — but limited — applicability to the world of crime and justice.

At a conference on policing research held here in Washington earlier this week, Wes Skogan spoke about the ideal of the white coat laboratory scientist — an ideal borrowed from the medical field that formed the foundation for the training of most criminologists — the ideal that researchers could retreat to the figurative laboratory, conduct experiments testing the effectiveness of criminal justice interventions using the most rigorous methods possible, preferably randomized controlled experiments, and then pronounce, a few years later, whether an intervention worked or not.

This model suffers from a number of weaknesses. First, as Lawrence Sherman pointed out in a recent lecture at the Police Foundation, even the medical community that we have looked to for inspiration does not live up to the ideal of medical research. He reported that an estimated 85 percent of medical practices remain untested by research evidence, most doctors rarely read the medical journals and instead base their practices on local custom and personal experience, and most studies that influence practice in fact use weak, non-randomized research designs.2 So, before we berate ourselves for not living up to the medical ideal, we should recognize that even in that profession — which has a tradition of espousing a strong relationship between science and practice — is apparently having great difficulty living up to that ideal.

Second, even where we approximate the ideal of the medical model, we should recognize that the business of crime prevention often implicates values that cannot be subjected to empirical assessment. In our business, the policy-making process responds, appropriately in my view, to competing values that tend to limit the impact and reach of research findings.

To illustrate this point, I would like to trace briefly the recent history of policy development in the area of police response to domestic violence. In the early 1980's, the Police Foundation published a landmark research report showing that arrests for misdemeanor spousal assault in Minneapolis significantly reduced subsequent violence.3 Within a short period of time, a large number of police departments shifted their domestic violence policies to ones that favored or mandated arrests. Although there were a number of other factors at work at the same time pushing police policies in the same direction — principally, the fear of litigation and pressure from feminist critics — this rapid shift in police practices was celebrated around the country as an exemplary translation of research into policy.

The replication studies of the original Minneapolis research painted a much more complex picture, however — for certain categories of offenders (specifically, for unemployed offenders), arrests for misdemeanor domestic assault actually resulted in higher levels of violence.4 When police policy did not follow these latest research findings, some researchers saw the failure of police agencies to curtail their mandatory arrest policies as an inexcusable dereliction of duty, bordering on malfeasance, because those policies contributed to domestic violence.

This story illustrates a number of points about our expectations regarding the process of translating research into practice. First, it underscores the fact that other values come into play in the policy making process. For example, if the police had promulgated a policy stating that the decision to enforce the laws against spousal assault depended upon the socioeconomic status of the offender, as the research would suggest, that policy, in my view, would have violated an important, overriding principle of our democracy — that the laws should be equally enforced against rich and poor alike. In this instance, I think that the police appropriately demonstrated concern about the principles of equal protection of the laws in deciding to institute pro-arrest policies and to sustain them even in the face of mixed findings regarding the impact of arrest upon subsequent violence.

But I do not want to be understood to dismiss lightly the findings that arrests of unemployed men for misdemeanor domestic assault tended to increase the likelihood of violence. One could argue that continuing a pro-arrest policy would certainly result in some new violence, a high price to pay for observing a legal principle. To me, this dilemma — of choosing between policies based on research and policies based on other sound principles of governance — points to the need for a different relationship between researcher and practitioner.

Imagine with me that the spousal assault replication research had been conducted as part of an ongoing relationship between researchers and practitioners concerned about developing the most effective response to domestic violence. Imagine that the researchers and practitioners were committed to a common goal — developing policies that resulted in reductions in domestic violence and enhancements in the safety of women. And imagine that the researcher recognized that, for a variety of reasons, the police practitioners could not adopt differential arrest policies for different victims or different offenders. In this context, the spousal assault replication findings would trigger a new inquiry — how to fashion and test policies that might assure the safety of the women. A variety of interventions might be tried — different bail policies, orders of protection, relocation of victims, electronic monitoring, substance abuse treatment, job placement. Through a combination of offender and victim focused activities, the agencies of the criminal justice system would be challenged to reduce the levels of violence — and those efforts would be subject to continuous monitoring and evaluation by the independent research team.

This relationship between researcher and practitioner represents the new partnership model that I believe holds such promise. Wes Skogan calls this the strategic feedback model of research. Allow me to tell another story that illustrates this model. Four years ago, the National Institute of Justice funded Prof. David Kennedy of the Kennedy School of Government to work with Boston Police Department to reduce juvenile violence. This was not the typical application for NIJ's research program. Truth be known, there was not much of a research design to sink your teeth into. Yet we took a chance with this proposal, in large part because the Kennedy School proposed to enter into a unique relationship with the Boston Police Department and other criminal justice agencies in that city. They proposed what Skogan now calls "strategic feedback" — working with those agencies to develop an empirical understanding of the juvenile homicide problem, developing testable hypotheses about possible interventions, collecting data while those interventions were implemented, and providing ongoing feedback to the strategic team about the results seen in the streets of Boston. Over the course of the next few years, the research project underwent several transformations. First, it resembled a study of gun markets. Then it morphed into a study of youth gangs. Then into a study of targeted deterrence strategies. These transformations were not dictated by the whims of the research team, rather they responded to evolving insights about the world of juvenile violence in Boston and what might be done to reduce the violence.5

Although David Kennedy and his colleagues would be the first to disclaim credit for these results, the changes in the level of violence in Boston have been stunning. After experiencing record levels of youth handgun homicides, Boston can now can boast that it has cut youthful gun homicides to the point where this kind of homicide is a rare event.

We have been watching Boston with great interest because of its implications for a new relationship between researchers and communities developing anti-crime strategies. With the strong support of Attorney General Reno, we are now engaged in a very ambitious five city experiment called the Strategic Approaches to Community Safety Initiative, or, since everything must have an acronym, SACSI. In these five cities, we have asked the United States Attorney to convene a strategic planning team consisting of federal and local law enforcement officials, criminal justice agencies, and other community stakeholders, to identify the most pressing crime problem facing that community and develop a strategy to bring that crime down. In each city, we have also required that there be a research partner at the table — one who will perform a function very similar to that played by David Kennedy and his Harvard colleagues. These research partners reflect a remarkable range of disciplines — criminology, public health, education, political science and others.

The first finding from this five city project is that this way of thinking is totally new — new to the prosecutors, new to the police, new to the other city agencies — and, in interesting ways, quite new to the researchers. Some researchers have embraced this challenge — one said this was the opportunity he had been waiting for seventeen years. Others find the challenge quite foreign and intimidating. The second finding from the project — and from Boston — is that this process takes a long time to get underway. In Boston, collaborative work didn't get started until more than nine months after the first meeting between researchers and practitioners. The third important finding is that the crime problems are not what they first seemed. The greatest contribution that the research partners make to this strategic approaches initiative is to help the practitioners understand their crime problems. Who is involved? Where is it happening? What are the different types of criminal behavior involved? Who is killing whom? Where are rapes being committed? Which kids are involved with which gangs? I do not mean to imply that none of these questions could be answered by the police and prosecutors — on the contrary, the front line workers often have the best insights into these problems. Rather, I have observed that the responses to the crime problems that are developed at the agency level are rarely based on the type of analysis — or front line insight — that reflects a true understanding of the phenomenon.

Our hope is that the SACSI project will yield important advances in our understanding of the effective response to crime. This has already been the case with the Boston Gun Project. David Kennedy's "pulling levers" hypothesis has made enormous theoretical and practical contributions to our thinking about deterrence and the role of the criminal justice system in producing safety. More fundamentally, it is our hope that the SACSI project and similar initiatives will add another tool to the scientific toolbox at our disposal. Clearly, this type of research involvement with practitioners is far removed from the medical model of the white lab coat technician. Nor is it the same as a management consultant who does not have the independence and methodological rigor of a university based academic.

Larry Sherman recently pointed me to another useful analogy, first developed by Marcus Felson.6 Maybe we should develop a cadre of criminal justice researchers who perform functions like those performed by the agriculture extension service — research-trained experts based in universities or other research institutions who are intimately involved with the welfare of the practitioners in their community, who keep abreast of the latest scientific findings and convey those findings to practitioners who are interested in improving their yields. Imagine that we could fund the universities around the country to provide a crime prevention extension service that would be engaged in constructive partnerships with the police, prosecutors, courts, corrections, and communities to help them devise and implement strategies to produce safety.

To help this new partnership be effective, we need to develop new ways for them to measure the crime problems facing their community. As we watch the five city initiative unfold, we have come to realize that not all communities are equipped with the same sources of data. Not all of them have incident based crime reports. Not all of them are participating in the Arrestee Drug Abuse Monitoring (ADAM) program that would help them track drug markets and changes in offender behavior. Not all of them are participating in the ATF gun tracing program that would help them track illegal gun markets. They do not routinely conduct fear and victimization surveys, nor do they geocode their data. So, NIJ's next initiative in this area, working with colleagues in the Department of Justice, will be to develop and implement a model set of crime data systems in a small number of jurisdictions to help the strategic planning process move forward. This program, which we call COMPASS — sorry for the acronym, it stands for Comprehensive Planning and Analysis for Safety Strategies — will be developed this year. We will soon be selecting our first COMPASS site — and are looking for a jurisdiction that can quickly implement these various systems of measurement at the jurisdictional level and can implement even more comprehensive measures of collective efficacy at the neighborhood level. In the five cities of SACSI, we have worked with those jurisdictions to catalogue existing data how crime, health, community well-being, schools — anything — and have developed software programs to access all of these different data to support strategic planning. We are looking for a jurisdiction where there is a strong tradition of collaboration and experimentation, both at the jurisdiction and the neighborhood level and is willing to test research based interventions.

As we look forward to the next decade of innovation in the relationship between research and practice, we must recognize the need to construct a national research infrastructure that will support a much more complex set of approaches to the challenges of crime. We believe that the COMPASS initiative will support these new approaches. The police profession has undergone remarkable transformations over the past twenty five years. The police have moved toward a more sophisticated, flexible, problem solving model of policing that involves a wide array of non police entities in both strategic and tactical decision making. At its core, this problem solving technology builds upon a strong research foundation — at its best, the problem solving process begins with an empirical understanding of a crime problem, moves toward articulation of the logic — or the hypothesis — of an intervention, and incorporates an ongoing assessment of the effectiveness of that intervention.

The research profession needs to catch up with policing and to define a role for itself in the problem solving process — the research community needs to find ways to bring its analytical skills, its objectivity, its rigor, its independence, its ability to link theory and practice, into the messy arena of contemporary practice. Larry Sherman call this approach "evidence-based policing," a phrase he borrows from the medical field; in the Department of Justice these days, we are talking about the need for data-driven crime prevention policies. Both of these phrases capture a new relationship between research and practice — an interactive, symbiotic relationship that requires new skills on both side of the partnership.

To succeed, this new partnership also requires support from both professions — those practitioners in the crime control and prevention business need to support the involvement of researchers in the development of their strategies and tactics. Those in academia need to support their colleagues who seek to undertake this form of research, and not relegate them to second class citizenship because they are not engaged in the more traditional research functions. I don't know which is more difficult — changing police culture or changing academic culture — but both will have to find ways to value the role of the other.

A final thought. I believe this transformation of the partnership between research and practice is important, not merely because of the implications for the research and practice professions, but because of the promise this partnership holds for the future of the communities in which we live and work. We have witnessed some remarkable declines in crime over the past few years to record lows, but we are a long way from achieving the levels of safety that befit this country and our grand experiment in democracy. Although we can take some solace in the fact that our property crime rates are lower than those in Europe, Australia and New Zealand, we should be appalled at the levels of lethal violence in our society. Although we can take some solace in the successes of community policing, we should be profoundly troubled by our country's fourfold increase in the rate of imprisonment over the past twenty-five years. Our levels of violence and levels of imprisonment undermine our ability to build a safe, healthy, tolerant and respectful national community.

We need to recognize that a national effort to produce a safer and more just society will not be successful if that effort is limited to a compilation of effective crime prevention programs — the types of programs that can be evaluated in our laboratories of social science. We will only produce the safe and just society we envision if we work at the community level, on complex problems, testing multiple interventions simultaneously. This is real life — where people's behavior is not the result of a program, but the result of a complex interaction of a number of influences — where levels of crime, fear and safety are not determined merely by the actions of the police and the criminal justice system, but by a complex interaction of forces of formal and informal social control, of culture and history. In the messy laboratory of real life in real communities, white coat style research is unlikely to provide satisfactory answers — we will have great difficulty determining causation, measuring effect, or distinguishing the contributions of competing interventions. Yet in those messy laboratories, there is great need for the skills and values that the research community can provide. People are trying new crime strategies every day, without government funding, without programs that have fancy acronyms, without the benefit of the latest NIJ publication. They may not even think they are implementing a crime prevention program, they may simply think they are trying to get their kids through the difficult years of adolescence alive. They may not think they are experimenting with new theories of deterrence, they may simply think they have found a way to link the police and probation to supervise kids at risk of getting killed.

These messy laboratories provide our modern-day strategic research opportunities. I urge you and your colleagues to accept the challenges presented by these research opportunities, and, if you choose another path, at least to value the work of your colleagues who do. The risks are great, but the potential benefits are even greater.

Thank you.

  1. The Challenge of Crime In a Free Society. A Report By The President's Commission On Law Enforcement and Administration of Justice. Washington, DC. February 1967 Pg. 273.

  2. "Ideas In American Policing." Police Foundation, Washington, DC. July 1998. Pg. 2.

  3. Sherman, Lawrence W., and Richard A. Berk. (1984b). The Minneapolis Domestic Violence Experiment. Washington, DC.: Police Foundation Reports 1.

  4. Sherman, Lawrence W., with Janell D. Schmidt and Dennis P. Rogan. Policing Domestic Violence: Experiments and Dilemmas. 1992. Pg. 155.

  5. Kennedy, David. "Pulling Levers: Getting Deterrence Right." NIJ Journal Issue No. 236, June 1998.

  6. Felson, Marcus. "A Crime Prevention Extension Service." In Crime Prevention Studies. Ronald V. Clarke, Editor. Criminal Justice Press. Monsey, New York. 1994.
Date Created: November 28, 2007