You are viewing a Web site, archived on 07:46:23 Oct 16, 2004. It is now a Federal record managed by the National Archives and Records Administration.
External links, forms, and search boxes may not function within this collection.
Public Health Seal report title shim
contents search order press resources links home curve end shim

Chapter 5:
Prevention and Intervention

Promoting Healthy, Nonviolent Children

Methods of Identifying Best Practices

Scientific Standards for Determining Program Effectiveness

Stategies and Programs: Model, Promising, and Does Not Work



Going to Scale


Appendix 5-A: Consistency of Best Practices Evaluations

Appendix 5-B: Descriptions of Specific Programs That Meet Standards for Model and Promising Categories

Model Programs: Level 1 (Violence Prevention)

Model Programs: Level 2 (Risk Prevention)

Promising Programs: Level 1 (Violence Prevention)

Promising Programs: Level 2 (Risk Prevention)

Chapter 5
Prevention and Intervention

Shootings and deaths in schools throughout the United States have left parents believing that their communities are no longer safe from the most extreme examples of youth violence (Gallup, 1999). This perception, combined with the increased lethality of youth violence in the early 1990s, has lent urgency to the search for effective violence prevention efforts. Hundreds of youth violence prevention programs are being used in schools and communities throughout the country, yet little is known about the actual effects of many of them (Gottfredson et al., 2000; Tolan & Guerra, 1994). Few such programs have been rigorously evaluated, including many ongoing efforts (Elliott, 1998). The evaluations that have been done indicate that much of the money America spends on youth violence prevention is spent on ineffective—sometimes even harmful—programs and policies (Mendel, 2000).

At the same time, researchers know much more today about how to prevent youth violence than they did two decades ago, when some declared that "nothing works" to prevent violence (Lipton et al., 1975; Sechrest et al., 1979). This is clearly no longer the case. Over the past few decades, social scientists have made great strides in uncovering the causes and correlates of youth violence.

Unfortunately, the news about effective programs has been slow to bring about change in school, community, and juvenile justice system prevention efforts, where precious resources continue to be spent on ineffective programs. Some experts believe that youth crime and violence rates could be "substantially" reduced simply by reallocating the money now spent on ineffective policies and programs to those that do work (Mendel, 2000, p. 1).

The strategy of using prevention resources to their fullest potential presents many challenges. The first lies in identifying effective prevention approaches and programs. Differentiating between effective and ineffective ones can be a difficult chore for schools, communities, and juvenile justice authorities. Numerous agencies and organizations have published recommendations on "what works" in youth violence prevention, but in many cases there is little consistency regarding the specific programs they recommend. The reason for this inconsistency is a lack of uniformly applied scientific standards for what works.


This chapter identifies a set of standards based on scientific consensus and applies those standards to the literature on youth violence prevention in order to identify with confidence general strategies and programs that work, that are promising, or that do not work to prevent youth violence. This information can be used by schools, communities, juvenile justice agencies, program funders, and others interested in youth violence prevention to aid their programming decisions. With this information in hand, it may be possible to fulfill the prediction that better use of existing prevention resources can substantially reduce the problem of youth violence.

The first section of this chapter describes the methods used in this report to identify best practices in youth violence prevention. The second describes currently accepted scientific standards for determining program effectiveness. The third section applies those standards to the existing youth violence prevention literature and presents findings on best practices—what works, what is promising, and what does not work. The information in that section is based on currently available research and is not intended to be the final word on the subject. As more programs are evaluated, the standards outlined in this report can be used to identify additional programs and strategies that work in preventing youth violence.

The fourth section, on cost-effectiveness, is intended to enhance the information provided in the best practices section by adding another dimension to the determination of what works. The conclusion discusses the need to take the next step in preventing youth violence by learning how to preserve the benefits of successful prevention programs when implementing them on a national scale.


Identifying the best practices for preventing youth violence involves two approaches, each with its own limitations. The first is meta-analysis, a rigorous statistical method of combining the results of several studies to obtain more reliable estimates of the effects of a general type of treatment or intervention. This quantitative approach can be used to summarize program evaluation evidence and draw overall conclusions about the strength and consistency of the influence, or effect size, that particular types of programs have on violent behavior. In the field of youth violence, meta-analysis has been used primarily for evaluations of interventions with violent or delinquent youths.

The second, less empirical approach is to review the evaluation research and identify the general strategies that characterize effective programs. While such reviews are not quantitative, they are more easily conducted than meta-analyses, and they offer useful information for generating hypotheses and drawing general conclusions about the effectiveness of various strategies for preventing youth violence.

In identifying best practices, this report relies heavily on recently published reviews and focuses mainly on strategies and programs with demonstrated effects on youth violence and on the major risk factors for youth violence.

Strategies and programs are first classified as effective or ineffective. Effective strategies and programs are then further broken down into Model programs, which meet very high standards of demonstrated effectiveness, and Promising programs, which meet a minimum standard. Finally, within Model and Promising categories, a distinction is made between strategies and programs that have demonstrated effects on violence and serious delinquency (Level 1) and those that have demonstrated effectiveness on known risk factors (Level 2).

The decision to include serious delinquency along with violence as a criterion for Level 1 programs was based upon the major meta-analysis of program effectiveness (Lipsey & Wilson, 1998), which did not differentiate between those two outcomes. Serious delinquency was a major risk factor for violence in both the early and late onset of risk (see Chapter 4). Level 2 Model programs are those that address any of the other risk factors with large effect sizes (substance use, weak social ties, antisocial or delinquent peers, gang membership).1 For Promising strategies, Level 1 again refers to programs with demonstrated effects on violence and serious delinquency. Level 2 programs are those with demonstrated effects on any risk factor with an effect size of .10 or greater.

No attempt was made to systematically search the literature for programs and strategies that affect small risk factors for youth violence, such as academic failure or anxiety and depressive disorders. Therefore, some effective programs or strategies that target small risk factors may not be included. However, such programs and strategies occasionally came to our attention; those that did were included if the risk factors they affected had an effect size of .10 or greater. Thus, risk factors such as child abuse and neglect, which have an effect size of less than .10, are not covered in this report.2 Also excluded are clinical trials of psychotropic medications, which have proved effective in treating some affective disorders that are risk factors for youth violence but have not been shown in rigorous clinical studies to reduce youth violence specifically. A review of these interventions can be found in Mental Health: A Report of the Surgeon General (1999).

Several major reviews of youth violence prevention and intervention programs have been published in the past decade. This chapter draws mainly upon the following:3

  • Preventing Crime: What Works, What Doesn’t, What’s Promising. A Report to the United States Congress (Sherman et al., 1997)
  • The Office of Juvenile Justice and Delinquency Prevention’s A Sourcebook: Serious, Violent, and Chronic Juvenile Offenders (Howell et al., 1995)
  • The Office of Juvenile Justice and Delinquency Prevention’s Guide for Implementing the Comprehensive Strategy for Serious, Violent, and Chronic Juvenile Offenders (Howell, 1995)
  • The Centers for Disease Control and Prevention’s Best Practices of Youth Violence Prevention: A Sourcebook for Community Action (Thornton et al., 2000)
  • The American Youth Policy Forum’s Less Hype, More Help: Reducing Juvenile Crime, What Works—And What Doesn’t (Mendel, 2000)
  • The Center for the Study and Prevention of Violence’s Blueprints for Violence Prevention (Elliott & Tolan, 1999)
  • "School-Based Crime Prevention" (Gottfredson et al., in press)

To improve readability, the reviews listed above are cited here and in the section below on scientific standards for effectiveness, but they are not referred to repeatedly in the section on best practices. Citations not listed above will be included where appropriate.

Many reviews use either no explicit criteria or only minimal criteria for identifying the programs they recommend, making it difficult to draw clear conclusions about the effectiveness of individual violence prevention strategies. This report limits its identification of effective programs to those meeting a minimum standard of study design similar to the maximum standard described in Sherman et al. (1997) and Gottfredson et al. (in press). In those two reports, a scientific methods score of 4 or 5 indicates that an evaluation used an experimental or quasi-experimental design. This report applied that design standard to the extent possible, given the limitations described above.

This report also attempts to distinguish between absolute deterrent effects, in which an intervention is compared to no treatment, and marginal deterrent effects, in which the intervention or strategy is evaluated against another treatment. Marginal deterrent effects are effects of the intervention over and above the effects of another treatment strategy and thus may underestimate the true effects of the intervention, compared to receiving no treatment at all. In such cases, the nature of the comparison group is identified in the text.

Both meta-analyses and reviews can be used to identify successful strategies, approaches, or types of programs used to prevent youth violence. However, this general approach has a critical limitation: Within any given category of programs, there may be specific programs that are effective and others that are not; moreover, programs may be effective for some populations but not others (males but not females, for example). The general effect size is an estimate of the average effect and thus may not characterize any particular program or its use for a particular population. Often the effects of individual programs within each general strategy vary widely. For this reason, it is necessary to take one more step in the identification of best practices—focusing specifically on individual programs that work.

Identifying specific programs that work requires a clear set of standards for judging effectiveness. This chapter describes the scientific community’s consensus regarding standards and how this report applied those standards to the evaluation literature to place programs in one of three categories: Model (demonstrates a high level of effectiveness), Promising (meets minimal standards of effectiveness), or Does Not Work (consistent evidence of no effects or harmful effects).

Few existing violence prevention and intervention programs have met the qualifications of a Model program. Many more have met the standards for a Promising program, and even more would probably meet these standards if evaluated appropriately. The fact that a program is not identified in this report as Promising or Model does not mean it is ineffective; in most cases it means only that it has not been rigorously evaluated. Those evaluated and found to be ineffective are identified as ineffective. While hundreds of programs are employed throughout the United States to prevent youth violence and treat young offenders, only those with a credible scientific evaluation are highlighted in this report. This shortfall underscores the need for a renewed focus on evaluation in the field of youth violence prevention.



As indicated in Chapter 4, drug use has a large effect size as an early risk factor but a small effect size as a late risk factor. It is included here as an outcome criterion for level 2 Model programs because most of the existing drug prevention programs begin in the early period, or before adolescence.


This limitation in the review of programs and strategies should not be interpreted as a judgment that such programs are unimportant. Programs that successfully address multiple risk factors, even those with very small individual effect sizes, may be very useful and should be supported and disseminated. Given limited funding, however, it seems prudent to invest in those programs that have greater potential effects on violence prevention.


Reviews by the Hamilton Fish Institute (2000), Drug Strategies Research Institute (1998), and the Center for Substance Abuse Prevention's (2000) online list of model programs were also considered. For the most part, these sources used criteria for selecting the programs they recommend different from those used in this report, or their recommendations overlap those in the primary sources for violence prevention listed above.

Back to Top

Home | Contents | Previous | Next