Declining Crime and Our National Research Agenda

Jeremy Travis
National Institute of Justice

School of Public Policy and Social Research
University of California, Los Angeles

March 2, 1998
Los Angeles, California

Good morning:

I thank you for the invitation to speak with you this morning and thank Mark Kleiman for arranging this impressive gathering. I realize that my trips to the West Coast are all too infrequent and so took advantage of a trip even further west -- to Australia -- to spend a few hours with you and with Peter Greenwood and his colleagues at RAND.

My topic this morning is "Declining Crime and Our National Research Agenda."

We live in an era of declining crime rates. These declining crime rates are not uniform across the nation, nor across crime types, nor across age groups. Crime rates are declining most steeply in our large cities; yet some medium sized cities are experiencing either stable or rising crime rates. Rates of property crime have been declining fairly steadily for twenty years. Indeed, as Frank Zimring and Gordon Hawkins remind us, the city I was in yesterday, Sydney Australia, has a property crime rate quite similar to Los Angeles. The unique American crime problem is lethal violence, not property crime.1 Over the same twenty year period, rates of violent crime have shown some volatility, with an increase and decrease in the late 70's and early 80's, then another sharp increase beginning in 1985 and sharp decrease beginning in the early 90's. And, according to the Bureau of Justice Statistics, the decrease in serious violent crime we have experienced over the past several years reflects a sharper decrease among adult offenders than among juvenile offenders.

Yet despite these variations by jurisdiction, crime category and offender age, the overall picture is of historic importance. According to the National Crime Victim Survey, our rates of violent crime are the lowest since the early 70's when the victimization survey was first conducted.2 And in some cities, such as my hometown New York City, the numbers are simply staggering. In 1990, when crime seemed out of control and caused a crisis in governance resulting in a decision to increase the police department by 5,000 officers, there were 2,245 homicides in New York City. Last year, there were 767, a drop of 66 percent.3

Yet, despite the best efforts of some effective publicists, we recognize that this decline in crime rates is not a New York story alone. Indeed, the city experiencing the greatest decline in homicide between 1990 and 1995 was not New York City, but Jacksonville, Florida. In fact, two thirds of U.S. cities over 200,000 in population, Los Angeles included, have experienced double digit drops in homicide rates during that five year period.4

What is happening with our crime rates?

We all get asked this question. Some of us may even have our pet theories. Some of the explanations offered in the public debate reflect the perspective of the commentator. Police officials in New York claim that new policing strategies in that City are totally responsible -- but have no insights into declines (or increases) in other cities. Advocates for tougher imprisonment policies point to declining crime rates as vindication for policies that have resulted in a quadrupling of our prison populations over the past twenty years, but can't square that explanation with the results in a city like Boston where deadly juvenile handgun violence has been virtually eliminated without harsher sentencing policies. Some explanations make the crime reduction appear to be the result of laws of mathematics or markets -- that this is merely regression to the mean, or akin to the self-corrections seen in an overheated stock market.

I am particularly struck by the reluctance of the research community to try to understand this phenomenon. At last year's conference of the American Society of Criminology, with one or two exceptions, I could find no plenary session or panel presentation or paper that offered academic perspective on crime's decline. The academic reluctance to engage this important question of crime's decline sometimes approaches disdain for the question itself. When Bill Bratton first asserted that the NYPD was responsible for New York's declining crime rates, prominent researchers scoffed privately that he must be cooking the books, and stated publicly that crime was beyond the influence of the police.

I should not be too harsh on the research community. Research has made important contributions. Our understanding of the rapid increase in violent crime rates in the mid-80's was significantly advanced, in my view, by the development of the"diffusion hypothesis" by Al Blumstein of Carnegie Mellon. He published an article in 1994 showing that, although robbery and burglary rates for juveniles have remained basically the same for twenty years, rates of juvenile violent crime significantly increased beginning in the mid-80's. Between the years 1985 and 1992, the homicide rate for those 18 and younger doubled;5 during the same period, the number of juvenile homicides with guns also doubled;6 during that period the arrest rate for nonwhite juveniles for drug offenses also doubled.7 The Blumstein "diffusion hypotheses" posits that, as crack cocaine hit urban America, established drug markets were destabilized, young people were drawn into the new crack trade, guns were needed to defend turf and to protect street-level operations, and these guns "diffused" into the adolescent culture until they were used to settle traditional adolescent squabbles over coats and girlfriends.8

Despite the contributions of this hypothesis to understanding the rise in violent crime, it only begs more questions when considering its decline. Have markets stabilized? where have the guns gone? how are those squabbles being settled now? why are adult rates of violent crime falling more sharply than juvenile rates?

Again, we have some insights from research. The work of Golub and Johnson on the cycle of the crack epidemic is very important research. Using survey data and urinalysis results from NIJ's Arrestee Drug Abuse Monitoring (ADAM) program (formerly known as the Drug Use Forecasting program), they have shown that the rates at which adolescent arrestees (those aged 18-20) test positive for cocaine (and self-report for crack use) have been declining significantly in some cities, while use rates for older arrestees have declined only slightly.9 So we have support for the hypothesis that a younger generation of adolescents is resisting the risky behavior of their elder siblings. And I must mention the recent report by my colleagues at the National Institute of Justice -- our study of homicides in eight cities -- which also found a very powerful statistical correlation between the rise and fall of crack cocaine as reflected in positive urine tests of arrestees and the rise and fall of homicide rates.

So, we have developed some important insights on crimes' rise and decline from research. And I am pleased that NIJ and the Journal of Crime and Criminology will be cosponsoring a symposium on crime's decline later this month at the Northwestern University. Yet I would like to argue this morning that we know too little about changing crime rates, and know that too late, and that we should think about designing a much more ambitious research agenda to understand crime in its local context as our challenge for the next century.

I would like to lay the foundation for this argument by talking first about the recent research findings that I think have the greatest significance for shaping the direction of both research and practice. These are drawn from the Project on Human Development in Chicago Neighborhoods which is cofunded by NIJ and the MacArthur Foundation. This study is a very ambitious, long-term inquiry into the relationship between community, crime, delinquency, family and individual development. At the community level, the Project has surveyed more than 8,700 adult residents in 343 neighborhoods throughout Chicago. In addition, researchers have identified 80 neighborhoods as the focus for a longitudinal cohort study to be conducted over the next eight years. As part of the first wave of this longitudinal study, researchers have conducted interviews with 7,000 children and adolescents and their primary care givers.10

This past August, the research team, headed by Dr. Felton Earls of the Harvard School of Public Health, published their first findings in Science magazine.11 They reported that the largest predictor of violent crime rates was "collective efficacy" -- a term they used to mean a sense of trust, common values, and cohesion in a neighborhood. They found that there are lower rates of violence in neighborhoods that have a strong sense of community and values, where adults are likely to intervene when children are missing from school or scrawling graffiti on building walls. According to Prof. Robert Sampson, a co-author of the report, "cohesion, or efficacy, seems to be a shared vision, a fusion of shared willingness of residents to intervene and social trust, a sense of engagement and ownership of public space." According to Earls, the most important characteristic of "collective efficacy" is a "willingness by residents to intervene in the lives of children." He was referring to a willingness to supervise, or step in, and monitor the children in a neighborhood.12

The finding of collective efficacy is important standing alone, but is doubly important when placed in context. Collective efficacy, when assessing predictors of violence, has an impact OVER AND ABOVE traditional predictors such as the racial or ethnic composition of a neighborhood, poverty levels, or residential instability.

These findings have had considerable influence on our thinking within NIJ -- and indeed throughout the Justice Department -- as we are designing an agenda for research and practice.

I will not dwell at length this morning on the implications for practice. In sum, our colleagues within different offices of the Department find substantial support in the Chicago research for their efforts to strengthen communities as part of larger crime prevention efforts. A number of comprehensive community programs sponsored by the Clinton Administration -- at least one of which, the Weed and Seed program, predates the Clinton team -- are explicitly premised on the notion that the appropriate role of the federal government is to work with communities to strengthen their infrastructure -- their "collective efficacy" -- as the first order of business in producing safety.

Yet I think the Chicago study's implication for research -- both basic research and applied research -- are quite profound. Let's return to the issue of our crime rates -- first the rise, then the decline. We know very little about how communities were affected by crack's explosive arrival in the mid-80's. Journalistic evidence, anecdotal evidence, and some ethnographic evidence, paint a picture of communities under siege. Having lived in a city besieged by crack -- and working for the Police Commissioner and Mayor of New York City during crack's onslaught -- I recall vividly the sense we could not hold the line against some awful evil force.

If the "collective efficacy" thesis has validity, we should ask whether those informal social controls weakened during that period, and whether they had a role in the re-emergence of safer communities.

There are those who argue that an understanding of the role of community in crime prevention is the missing piece in our research knowledge base. And they argue that the last decade, because of its extreme rise and fall of violent crime, provides the natural setting for understanding those dynamics. Roger Conner, who is a community activist who fought to rescue his own neighborhood in Washington, DC from crack dealers, and is now the Executive Director of the Center for the Community Interest, has coined a wonderful phrase for this phenomenon. He says we witnessed the "release of the antibodies" -- that at some point in time, the healthy forces in these devastated communities said they could take it no more, and mustered their collective strength to fight back. George Kelling recently offered a similar explanation for crime's decline in New York City. At his lecture in December, as part of NIJ's series called Perspectives on Crime and Justice, he said that the first healthy sign in New York's return to safety was the organization of community groups -- tenant organizations, business improvement districts, block watchers, safety coalitions -- that fought to restore order, and their efforts were then met by a police department that had been transformed to be more receptive, flexible and effective in responding to crime conditions.13

We find ourselves looking to other disciplines to import theories and models that might shed light on these community dynamics. In the public health area, epidemiologists construct models of epidemics and can identify "tipping points"14 -- points where the body politic begins to reverse a trend toward unhealthy behavior and reasserts healthy behaviors. Scholars such as Thomas Schelling who studied residential segregation patterns over the past generation identified points at which whole neighborhoods suddenly shifted from one racial group to another. George Galster of the Urban Institute has studied urban redevelopment efforts, both governmental and private, and has shown that those efforts often succeed only after a bad neighborhood as hit bottom, then it can rapidly be reclaimed. Jonathan Crane, a sociologist at the University of Illinois, found that when the number of role models in a community -- people with professional jobs-fell below five percent, certain indicators that had remained stable while the professional presence declined steadily from forty percent to five percent, suddenly exploded. Drop out rates doubled, the rate of child-bearing for teen age girls nearly doubled. The community infrastructure can no longer exert the same power over behavior. Two economists at the University of California at Berkley, George Kerlof and Jane Yellen have developed a complex economic model showing that community resiliency is a key ingredient in crime fighting -- that at a certain point, a community decides that crime has gotten too bad and that it should join forces with the police.15 The question then becomes do we have models of community resiliency as it pertains to crime -- and data to test those models? No.

At NIJ, we are thinking about the implications of the Chicago study, and this notion of community dynamics, for our own research strategy. We are moving in two directions. We are funding different methods of research -- a look at our portfolio will show a lot more ethnographic work, such as that being conducted by Mercer Sullivan and Jeffrey Fagan. We want to bring the research process much closer to the individuals who commit crimes, and the victims, families and communities affected by crimes.16

More important, long term, we are undertaking new measurement and evaluation efforts. Two of these have been publicly announced. I referred earlier to the ADAM program. If funded by Congress, we plan to have a capacity, by the year 2001, to conduct confidential interviews on a quarterly basis with randomly selected arrestees in all 75 cities over 200,000 in population (with annual surveys in rural and suburban jurisdictions), and take urine samples, to measure changes in drug use, gun use, and other criminal behaviors. This year, we expanded the ADAM research program from 23 to 35 cities. Over twenty years ago, our country developed the victimization survey as a standard measure of crime -- it has been invaluable as a research and policy tool because it opened a window into the world of victimization without relying on official records. The ADAM program, although quite different in method, opens a window into the world of offending and will provide invaluable, timely data on the changing picture of crime, particularly changing drug markets.

As you may know, we have also established a Crime Mapping Research Center within NIJ. Although we are very interested in the implications of mapping for practice, we are principally interested in the research yield that comes from understanding crime in its spatial and temporal dimensions, and placing it in the context of other social developments. The Mapping Center staff are planning to inaugurate a Neighborhood Safety Sentinel Program which will model changes in those variables to enhance our ability to predict change in crime patterns at the neighborhood level.

The most ambitious new effort now under development at NIJ is called "Building Safety, Building Knowledge: Block by Block", or more affectionately by its nickname, "Block by Block." As currently envisioned, the "Block by Block" initiative would look like this. Over a five year period, NIJ would establish, at the neighborhood level in a small number of cities, a comprehensive set of measures of community well-being -- very similar to those currently being developed by the Project on Human Development in Chicago Neighborhoods. We would include traditional measures such as crime, fear of crime, and community disorder; we would also measure public attitudes, social cohesion, offender behavior, and victim responses to crime. In essence, we are adopting the "community" as the unit of analysis for a larger research effort.

Yet, Block by Block is more than a basic research program. The working title, you will recall, is "Building Safety, Building Knowledge: Block by Block". We want also to learn about what works to prevent crime -- how to "build safety" -- and to ground our research knowledge about the impact of interventions in the community context. So, we are also planning to work closely with the leadership of the neighborhood and the larger community to test, simultaneously, a limited number of interventions that have strong foundations in the research literature and have demonstrated effectiveness. The key here is to examine the impact of these interventions by using a very diverse set of measures of impact and, as important, to examine the interactive effects of more than one intervention.

Let's be a bit more concrete. We have strong reasons to believe that mentoring programs may be effective, that police and other criminal justice interventions targeted against young people at high risk of involvement in gangs may be effective, and that conflict resolution programs may be effective. All of these programs, in different ways, are efforts to change the dynamics of the youth culture in a community and change the life course of individual young people. Is it possible that the power of each intervention will be greater if they are undertaken simultaneously? Take a second example. We have strong reasons to believe that there have been significant changes in the relationship between levels of crack cocaine and levels of violence. Yet, beyond understanding the statistical correlations between the prevalence of crack cocaine and the levels of homicide, we have little understanding of how deliberate policy interventions might influence that relationship. Can a combination of police intervention, youth workers, and drug treatment change the attitudes and behaviors of drug users and people at risk of drug use?

In its current conception, therefore, Block by Block is both a research program and a demonstration program. We want both to measure and to test research hypotheses. Let me hasten to acknowledge the risks inherent in this venture. We are acutely aware that any effort to test more than one intervention at the same time poses risks of contamination of effects. We know that we will have difficulty -- perhaps approaching impossibility -- establishing a control or comparison community.

Yet we think these challenges are worth embracing. The poverty we face here is not a poverty of policy options -- there are abundant good ideas for what should be done to reduce crime. Rather we suffer from a poverty of research design. We have insufficient measures, over an insufficient length of time, with insufficient willingness to try more than one intervention at one time. In short, we are not presently able to conduct research in a way that reflects the complexity -- and the potential power -- of real life and real communities. This is not an argument to abandon the medical model of evaluation research. On the contrary, I believe strongly we need to geometrically increase our investment in testing research hypotheses using the most rigorous research designs that can be implemented in real world settings. Nor is this the occasion for a full critique of the medical model. Yet, I note with great interest that one of our staunchest advocates for the model of medical trials in crime prevention research -- Prof. Lawrence Sherman -- publicly announced a shift in his position last month in a talk at the Police Foundation:

"For years, Sherman (1984, 1992) and others have used medicine as the exemplar of a profession based upon strong scientific evidence. Sherman has praised medicine as a field in which practitioners have advanced training in the scientific method, which they use to keep up to date with the most recent research evidence by reading medical journals. He has cited the large body of randomized controlled experiments in medicine -- now estimated to number almost one million in print (Sackett and Rosenberg, 1995) -- as the highly rigorous scientific evidence used to guide medical practices. He has suggested that policing should therefore be more like medicine.

Sherman was wrong. Like so many conclusions, this one was reached on the basis of inadequate evidence. Better data on how doctors work reveals medicine to be a battleground between research and practice. The idea that most practice should be based on research has spawned a sweeping new strategy called "evidence-based medicine," "widely hailed as the long-sought link between research and practice" (Zuger, 1997)." 17

I agree with him -- we should encourage methodological pluralism. We should recognize that the work of crime control and prevention does not occur in the closed systems of medical trials, but in the open systems of communities.

At NIJ -- the ADAM research program, Crime Mapping Research Center and Block by Block Initiative -- reflect shifts in our thinking about the research strategy necessary to capture community dynamics. Ironically, our recent experience with the rise and fall of violent crime may have provided the catalyst to our new approaches to NIJ's fundamental mission. Yet this is certainly not our mission alone. The development of useful knowledge about crime and justice is equally the task of researchers, police officials, policy makers, community leaders -- and I welcome your partnership in this important effort.


  1. Zimring, Franklin & Gordon Hawkins. 1997. "Lethal Violence and the Overreach of American Imprisonment." NIJ Research Report. Washington DC: US Department of Justice.

  2. Bureau of Justice Statistics. National Crime Victimization Survey. Washington, D.C.: U.S. Department of Justice.

  3. New York Police Department. 1997. Preliminary Homicide Report. New York.

  4. Lattimore, Pamela K., James Trudeau, K. Jack Riley, Jordan Leiter, and Steve Edwards. 1997. "Homicide in Eight U.S. Cities: Trends, Context, and Policy Implications." NIJ Research Report. Washington DC: US Department of Justice.

  5. Age-Specific Arrest Rates and Race-Specific Arrest Rates for Selected Offenses.1965-1992. Uniform Crime Reporting Program, Federal Bureau of Investigation, Washington, DC: U.S. Department of Justice.

  6. Supplementary Homicide Report. 1965-1992. Uniform Crime Reporting Program, Federal Bureau of Investigation, Washington, DC: U.S. Department of Justice.

  7. Blumstein, Alfred. 1995. "Violence by Young People: Why the Deadly Nexus?" National Institute of Justice Journal. Washington DC: U.S. Department of Justice.

  8. Ibid.

  9. Golub, Andrew & Bruce Johnson. 1997. "Crack's Decline: Some Surprises Across U.S. Cities." National Institute of Justice Research in Brief. Washington DC: US Department of Justice.

  10. Earls, Felton J. & Christy A. Visher. 1997. "Project on Human Development in Chicago Neighborhoods: A Research Update." National Institute Research In Brief. Washington, D.C.: U.S. Department of Justice.

  11. Sampson, Robert & Stephen Raudenbush & Felton Earls. 1997. "Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy." Science, Vol.277. American Association for the Advancement of Science.

  12. Butterfield, Fox. "Study Links Rate of Violence to Cohesion in Community." New York Times. August 17, 1997.

  13. Kelling, George. "Crime Control, the Police, and the Cultural Wars: Broken Windows and Cultural Pluralism." December 2, 1997. National Institute of Justice Perspectives on Crime and Justice Lecture Series. Washington, DC.

  14. Gladwell, Malcolm. "The Tipping Point." June 3, 1996. The New Yorker.

  15. Bennett, Amanda. "Economists Demonstrate That Neighbors, Not Wardens, Hold Keys to Cutting Crime." December 7, 1994. Wall Street Journal.

  16. Fagan, Jeffrey. 1998. "Adolescent Violence: A View From the Street." National Institute of Justice Research Preview. Washington, DC: U.S. Department of Justice.

  17. Sherman, Lawrence W. "Evidence-Based Policing." February 18, 1998. Police Foundation Second Invitational Lecture on Ideas in Policing. Washington, DC.
Date Created: December 3, 2007