You are viewing a Web site, archived on 16:44:28 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. Note that this document was downloaded, and not saved because it was a duplicate of a previously captured version (09:19:37 Oct 16, 2004). HTTP headers presented here are from the original capture.
Public Health Seal report title shim
contents search order press resources links home curve end shim
Chapter 1:

Scope, Focus, and Overarching Themes

Secondary Areas of Concern

Youth Violence: The Public Health Approach

Myths About Youth Violence

Sources of Data and Standards of Evidence

Overview of the Report's Chapters

Chapter Conclusions

Preparation of the Report


Chapter 1


In October 1985, Surgeon General C. Everett Koop convened an unprecedented Workshop on Violence and Public Health (U.S. DHHS, 1986). The participants agreed strongly that it was time public health perspectives and expertise were brought to bear on questions of crime and violence. Throughout much of the last century, these questions had been dominated by the social sciences and the criminal justice system. For the most part, health care efforts were restricted to the rehabilitation of convicted offenders (Sechrest et al., 1979; U.S. DHHS, 1986). Dissatisfaction with both the timing and the outcomes of the "rehabilitation ideal" spurred the search for a more effective role for health care in addressing violence.

With its emphasis on prevention of disease or injury, the public health approach to violence offers an appealing alternative to an exclusive focus on rehabilitation. Primary prevention identifies behavioral, environmental, and biological risk factors associated with violence and takes steps to educate individuals and communities and protect them from these risks. Central to education and protection is the principle that health promotion is best learned, performed, and maintained when it is ingrained in individuals' and communities' daily routines and perceptions of what constitutes good health practices.

Public health practitioners and advocates have taken the lead in encouraging alliances and networks among academic disciplines, professions, organizations, and communities to make health concerns permanent public priorities and part of personal practices. In that tradition, participants at the 1985 Surgeon General's conference emphasized the importance of convincing the public that violence should be treated as a public health problem. As Marvin Wolfgang, a distinguished leader in the field of criminology, told conferees, "Our nation must feel as comfortable in controlling its violent behavioral urges and practices as it does in controlling bacterial, viral, and physical manifestations of morbidity and death" (U.S. DHHS, 1986).

Just as the application of public health principles and strategies has reduced the number of traffic fatalities and deaths attributed to tobacco use (CDC, 1999), the public health approach can help reduce the number of injuries and deaths caused by violence. Broader than the medical model, which is concerned with the diagnosis, treatment, and mechanisms of specific illnesses in individual patients, public health offers a practical, goal-oriented, and community-based approach to promoting and maintaining health. To identify problems and develop solutions for entire population groups, the public health approach:

  • Defines the problem, using surveillance processes designed to gather data that establish the nature of the problem and the trends in its incidence and prevalence;
  • Identifies potential causes through epidemiological analyses that identify risk and protective factors associated with the problem;
  • Designs, develops, and evaluates the effectiveness and generalizability of interventions; and
  • Disseminates successful models as part of a coordinated effort to educate and reach out to the public (Hamburg, 1998; Mercy et al., 1993).

The chapters in this report are keyed to each of these components of the public health approach. Chapter 2 presents research describing the magnitude of the problem of violent behavior by young people. Chapter 3 explores how violence develops and emerges over time. Chapter 4 summarizes research on risk and protective factors for youth violence; Appendix 4–B elaborates on the effects of exposure to media violence (including violence in interactive media) as a risk factor for aggressive and violent behavior. Chapter 5 focuses on the design, evaluation, and refinement of numerous programs and strategies that seek to reduce or prevent youth violence; Appendix 5–B provides details on specific programs discussed in the chapter. Chapter 6 suggests future courses of action, including the necessary next steps in research. A glossary of technical and discipline-specific terms follows.


An important reason for making research findings widely available is to challenge false notions and misconceptions about youth violence. Myths such as those listed below are intrinsically dangerous. Assumptions that a problem does not exist or failure to recognize the true nature of a problem can obscure the need for informed policy or for interventions. An example is the conventional wisdom in many circles that the epidemic of youth violence so evident in the early 1990s is over. Alternatively, myths may trigger public fears and lead to inappropriate or misguided policies that result in inefficient use of scarce public resources. An example is the current policy of waiving or transferring young offenders into adult criminal courts and prisons.

Myth: The epidemic of violent behavior that marked the early 1990s is over, and young people—as well as the rest of U.S. society—are much safer today.

Fact: Although such key indicators of violence as arrest and victimization data clearly show significant reductions in violence since the peak of the epidemic in 1993, an equally important indicator warns against concluding that the problem is solved. Self-reports by youths reveal that involvement in some violent behaviors remains at 1993 levels (see Chapter 2).

Myth: Most future offenders can be identified in early childhood.

Fact: Exhibiting uncontrolled behavior or being diagnosed with conduct disorder as a young child does not predetermine violence in adolescence. A majority of young people who become violent during their adolescent years were not highly aggressive or "out of control" in early childhood, and the majority of children with mental and behavioral disorders do not become violent in adolescence (see Chapter 3).

Myth: Child abuse and neglect inevitably lead to violent behavior later in life.

Fact: Physical abuse and neglect are relatively weak predictors of violence, and sexual abuse does not predict violence. Most children who are abused or neglected will not become violent offenders during adolescence (see Chapter 4).

Myth: African American and Hispanic youths are more likely to become involved in violence than other racial or ethnic groups.

Fact: Data from confidential interviews with youths indicate that race and ethnicity have little bearing on the overall proportion of racial and ethnic groups that engage in nonfatal violent behavior. However, there are racial and ethnic differences in homicide rates. There are also differences in the timing and continuity of violence over the life course, which account in part for the overrepresentation of these groups in U.S. jails and prisons (see Chapter 2).

Myth: A new violent breed of young superpredators threatens the United States.

Fact: There is no evidence that young people involved in violence during the peak years of the early 1990s were more frequent or more vicious offenders than youths in earlier years. The increased lethality resulted from gun use, which has since decreased dramatically. There is no scientific evidence to document the claim of increased seriousness or callousness (see Chapter 3).

Myth: Getting tough with juvenile offenders by trying them in adult criminal courts reduces the likelihood that they will commit more crimes.

Fact: Youths transferred to adult criminal court have significantly higher rates of reoffending and a greater likelihood of committing subsequent felonies than youths who remain in the juvenile justice system. They are also more likely to be victimized, physically and sexually (see Chapter 5).

Myth: Nothing works with respect to treating or preventing violent behavior.

Fact: A number of prevention and intervention programs that meet very high scientific standards of effectiveness have been identified (see Chapter 5).

Myth: In the 1990s, school violence affected mostly white students or students who attended suburban or rural schools.

Fact: African American and Hispanic males attending large inner-city schools that serve very poor neighborhoods faced—and still face—the greatest risk of becoming victims or perpetrators of a violent act at school. This is true despite recent shootings in suburban, middle-class, predominantly white schools (see Chapter 2).

Myth: Weapons-related injuries in schools have increased dramatically in the last 5 years.

Fact: Weapons-related injuries have not changed significantly in the past 20 years. Compared to neighborhoods and homes, schools are relatively safe places for young people (see Chapter 2).

Myth: Most violent youths will end up being arrested for a violent crime.

Fact: Most youths involved in violent behavior will never be arrested for a violent crime (see Chapter 2).


Data Sources
Several comprehensive scholarly reviews of various facets of youth violence were published in the 1990s. Professional organizations, Federal agencies, the National Academy of Sciences, and university-based researchers have invested immense energy in reviewing research on the occurrence and patterns of youth violence, its causes and consequences, intervention strategies, and implications for society.

Key contributions to this rich information base include:

  • NIMH Taking Stock of Risk Factors for Child/Youth Externalizing Behavior Problems (Hann & Borek, in press)
  • Serious and Violent Juvenile Offenders (Loeber & Farrington, 1998). A report of the Office of Juvenile Justice and Delinquency Prevention (OJJDP) Workgroup on Violence and Serious Offending
  • The National Academy of Sciences' four-volume report Understanding and Preventing Violence (Reiss & Roth, 1993)
  • The American Psychological Association's report Violence and Youth (APA, 1993) and Reason to Hope (Eron et al., 1994)
  • Preventing Crime: What Works, What Doesn't, What's Promising. A Report to the United States Congress (Sherman et al., 1997)
  • The OJJDP national report Juvenile Offenders and Victims (Snyder & Sickmund, 1999)
  • The American Sociological Association's Social Causes of Violence: Crafting a Science Agenda (Levine & Rosich, 1996)

This report draws extensively—but not exclusively—on concepts, general information, and data contained in these documents. The authors gratefully acknowledge the contributors to and publishers of these earlier studies. Whenever the report draws heavily on one of these master sources, that fact is noted. Specific references to these documents are provided where appropriate.

Contributors to and editors of this report have also consulted peer-reviewed journals, books, and government reports and statistical compilations. Some information not considered in prior reviews is contained in this report. When appropriate, the editors have drawn on dissertations and forthcoming work that they judged to be of high quality.

During the development of this report, special data analyses were obtained from established surveys of U.S. adolescents. The key data sources for these analyses are the following:

  • Monitoring the Future survey conducted annually by the University of Michigan's Institute for Social Research (Johnston et al., 1995)
  • Youth Risk Behavior Surveillance Study sponsored by the Centers for Disease Control and Prevention in collaboration with Federal, state, and local partners (Brener et al., 1999)
  • The National Center for Injury Prevention and Control's Firearm Injury Surveillance Study (CDC, NCIPC, 2000)
  • Several longitudinal databases generated by the Program of Research on the Causes and Correlates of Delinquency, Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice (Huizinga et al., 1995)
  • The National Center for Juvenile Justice's up-to-date information on juvenile arrests for violent crimes (Snyder, 2000)
  • The National Crime Victimization Survey (Rand et al., 1998)

Standards of Scientific Evidence for Multidisciplinary Research
The public health approach relies on a multidisciplinary, multijurisdictional knowledge base. Thus, in preparing this report, it was necessary to draw conclusions from research in psychology (social, developmental, clinical, and experimental), sociology, criminology, neuroscience, public health, epidemiology, communications, and education. Integrating findings and conclusions across disciplinary lines is never easy. The questions under study generally determine what approach scientists will take to designing and conducting research, and the approach often determines how investigators report their findings and conclusions. Even when scientific approaches are similar, investigators in different disciplines frequently employ different terminology to describe similar concepts.

In striving to apply scientific standards consistently across the many fields of research reviewed, this report has emphasized two criteria: appropriately rigorous methods of inquiry and sufficient data to support major conclusions. The need for rigor is obvious: The tools or strategies employed in research—like the conclusions reached—are only as good as the precision with which research questions are framed. But the quality of a given study depends on other factors as well, including:

  • General data collection design. Data may be obtained through four major types of study design: experimental, longitudinal, cross-sectional, and case study. This report relies primarily on experimental and longitudinal designs, with some use of cross-sectional studies. (These three methods are described below.)
  • Sampling, or the selection of persons to be studied. Individuals in a study may be recruited or identified through probability or nonprobability sampling, or they may be assigned to experimental or control groups by a random process, a precision or group-matching process, or some other means. This report refers to probability samples as representative samples.
  • Validity and reliability of measures or instruments used in the research.
  • Appropriateness and level of control incorporated into the analysis of findings. Level of control refers to efforts to take into account other factors that might be influencing data or responses from subjects.
  • Appropriateness and significance of generalizations.

As noted earlier, four of the chapters in this report—those concerned with magnitude, demographics, risk and protective factors, and intervention research and evaluation—mirror components of the public health approach to youth violence. Each of these areas involves research from different disciplines and scientific approaches; therefore, the types of research designs and forms of analysis presented differ somewhat from chapter to chapter.

Experimental research is the preferred method for assessing cause and effect as well as for determining how effectively an intervention works. Many of the violence prevention programs reviewed in Chapter 5 meet the standard of rigorous experimental (or well-executed quasi-experimental) designs. In an experimental study, researchers randomly assign an intervention to one group of study participants, the experimental group, and provide standard care or no intervention to another group, the control group. A study with a randomly assigned control group enables researchers to conclude that observed changes in the experimental group would not have happened without the intervention and did not occur by chance. The difference in outcome between the experimental and control groups, which in this case may be the reduction or elimination of violent behaviors, can then be attributed to the intervention.

Ideally, researchers assign study participants to the experimental intervention or the control group at random. Randomization eliminates bias in the assignment process and provides a way of determining the likelihood that the effects observed occurred by chance. In this report, most weight is given to true experimental studies. In some cases, true experiments may be too difficult or expensive to conduct, or they may pose unacceptable ethical problems. In such cases, carefully designed and executed quasi-experimental studies are accepted as meeting the standard.

Evidence from an experimental study is considered stronger when, in addition to analyzing the main effects of an intervention, researchers analyze the mediating effects. This analysis permits researchers to determine whether a change in the targeted risk or protective factor accounts for the observed change in violence—that is, did the intervention work because it changed the degree of risk? Without this information, researchers cannot explain the success of a program.

Chapters 4 and 5 make use of meta-analyses. Meta-analysis describes a statistical method for evaluating the conclusions of numerous studies to determine the average size and consistency of the effect of a particular treatment or intervention strategy common to all of the studies. The technique makes the results of different studies comparable so that an overall effect can be identified. A meta-analysis determines whether there is consistent evidence that a treatment has a statistically significant effect, and it estimates the average size of that effect.

Epidemiological research, reviewed in Chapters 2 and 3, focuses primarily on general population studies that use probability samples and cross-sectional or longitudinal designs (Kleinbaum et al., 1982; Lilienfeld & Lilienfeld, 1980; Rothman & Greenland, 1998). Probability samples let researchers generalize from their study to the entire population sampled. Cross-sectional studies involve a single contact with participants for data collection at a given point in time. Multiple cross-sectional studies involve several waves of data collection over time (annually, for example) but typically with different participants at each contact and therefore with no way to link a given person's responses at one time with those at a later time. Prospective longitudinal and panel designs involve multiple contacts with the same study participants over time. Responses at one data collection point can be linked to responses at a later point. Longitudinal studies are used for research on individual development or growth.

Longitudinal designs are necessary to estimate the predictive effect of a given risk or protective factor on later violent behavior. Although cross-sectional designs are sometimes used, they cannot provide estimates of individual-level predictive effects. They can establish simultaneous relationships between risk factors and violence, but conclusions drawn from cross-sectional studies are not as strong as those drawn from longitudinal studies. In cross-sectional studies, cause and effect are unclear and reciprocal effects may inflate the estimates.

Experimental studies are sometimes used to estimate the effects of risk and protective factors, but this practice is rare because of ethical and cost considerations. For example, it would be unthinkable to introduce drug use to a group of adolescents to see whether drugs are a risk factor for violence. However, it would be ethical to conduct a predictive study that selects persons who are not violent and follows them over time. Those who began to use illicit substances would be compared with those who did not, to determine whether drug users are more likely to become involved in violent behaviors at some later date. If they were, then the results would indicate that drug use predicts violence or that drug use increases the probability of future violence.

Level of Evidence
No single study, however well designed, is sufficient to establish causation or, in intervention research, efficacy or effectiveness. Findings must be replicated before gaining widespread acceptance by the scientific community. The strength of the evidence amassed for any scientific fact or conclusion is referred to as the level of evidence.

This report does not rely on any single study for conclusions. Only findings that have been replicated in several studies, consistently and with no contrary results, are reported as part of the contemporary knowledge base. When the report cites unreplicated studies that are of high quality, that have not been refuted by other evidence, and that point in a clear direction, the findings are described as tentative or suggestive. These findings may point to future research needs and directions, but the report takes a conservative approach to drawing conclusions from them.

Back to Top

Home | Contents | Previous | Next