NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Office of the Surgeon General (US); National Center for Injury Prevention and Control (US); National Institute of Mental Health (US); Center for Mental Health Services (US). Youth Violence: A Report of the Surgeon General. Rockville (MD): Office of the Surgeon General (US); 2001.

Cover of Youth Violence

Youth Violence: A Report of the Surgeon General.

Show details

Chapter 3 -- The Developmental Dynamics of Youth Violence

Most violence begins in the second decade of life. This chapter looks closely at childhood and adolescence as critical periods of development to trace how violence unfolds -- its onset, duration, and continuity into adulthood. It also examines violence in relation to other risky behaviors that emerge in adolescence.

The dynamics of youth violence are best understood from a developmental perspective, which recognizes that patterns of behavior change over the life course. Adolescence is a time of tumultuous change and vulnerability, which can include an increase in the frequency and means of expression of violence and other risky behaviors. Understanding when and under what circumstances violent behavior typically occurs helps researchers craft interventions that target those critical points in development.

Our understanding of developmental patterns depends in large part on longitudinal studies, which track the same group of individuals over long periods of time, sometimes a decade or more. Four major longitudinal studies are described in this chapter. They add new dimensions to the surveillance statistics presented in Chapter 2, and they provide essential background for Chapter 4, which deals with why young people become involved in violence.

Early- and Late-Onset Trajectories

Longitudinal research has detected two prominent developmental trajectories for the emergence of youth violence, one characterized by an early onset of violence and one by a late onset. Children who commit their first serious violent act before puberty are in the early-onset group, whereas youths who do not become violent until adolesence are in the late-onset group. While other developmental trajectories have been identified (D'Unger et al., 1998; Nagin & Tremblay, 1999), this report focuses on the early- and late-onset trajectories because they are recognized by most researchers, they debunk the myth that all serious violent offenders can be identified in early childhood, and they have strikingly different implications for prevention.

In the early-onset trajectory, problem behavior that begins in early childhood gradually escalates to more violent behavior, culminating in serious violence before adolescence. A child's first serious violent act may have been officially recorded, or it may have been reported by the child to researchers in a confidential survey. The early-onset group, in contrast to the late-onset group, is characterized by higher rates of offending and more serious offenses in adolescence, as well as by greater persistence of violence from adolescence into adulthood (reviewed in Stattin & Magnusson, 1996, and Tolan & Gorman-Smith, 1998). The National Youth Survey shows that nearly 13 percent of male adolescents in the early-onset trajectory engaged in violence for two or more years, compared to only 2.5 percent in the late-onset trajectory (Tolan & Gorman-Smith, 1998).

Between 20 and 45 percent of boys who are serious violent offenders by age 16 or 17 initiated their violence in childhood (D'Unger et al., 1998; Elliott et al., 1986; Huizinga et al., 1995; Nagin & Tremblay, 1999; Patterson & Yoerger, 1997; Stattin & Magnusson, 1996). A higher percentage of girls who were serious violent offenders by age 16 or 17 (45 to 69 percent) were violent in childhood (Elliott et al., 1986; Huizinga et al., 1995). This means that most violent youths 1 begin their violent behavior during adolescence. However, the youths who commit most of the violent acts, who commit the most serious violent acts, and who continue their violent behavior beyond adolescence begin during childhood (Loeber et al., 1998; Moffitt, 1993; Tolan, 1987; Tolan & Gorman-Smith, 1998).

The greater prevalence of late-onset youth violence refutes the myth that all serious violent offenders can be identified in early childhood. In fact, the majority of young people who become violent show little or no evidence of childhood behavioral disorders, high levels of aggression, or problem behaviors -- all predictors of later violence.

The implications of these findings for prevention are clear: Programs are needed to address both early- and late-onset violence. Targeting prevention programs solely to younger children with problem behavior misses over half of the children who will eventually become serious violent offenders, although universal prevention programs in childhood may be effective in preventing late-onset violence (see Chapters 4 and 6).

Onset and Prevalence of Serious Violence

Much of what is known about the onset, prevalence, and other characteristics of serious violence during the adolescent years comes from four important longitudinal surveys. The only nationally representative one is the National Youth Survey (NYS), an ongoing study of 1,725 youths age 11 to 17 in 1976, when the survey began (Elliott, 1994). These youths have been tracked by researchers for more than two decades and through nine waves, or points at which they were interviewed and/or their official records were sought to corroborate self-reported violence. 2

The other three longitudinal studies cited here are city surveys sponsored by the U.S. Office of Juvenile Justice and Delinquency Prevention and the National Institutes of Health (Huizinga et al., 1995; Thornberry et al., 1995). 3 Beginning in 1988, three teams of researchers began to interview 4,500 youths age 7 to 15 in three cities -- Denver, Pittsburgh, and Rochester (New York). These youths were monitored at different points from 1988 to 1994. Each sample disproportionately represents youths at high risk of delinquency to ensure that it is large enough to draw valid conclusions about delinquency and violence, but each also uses weighting procedures to yield locally representative estimates. The estimates presented here are based on weighted data.

These four surveys define serious violence as aggravated assault, robbery, gang fights, or rape; an individual is labeled a serious violent offender if he or she reports committing any one or more of these offenses. 4 Gang fights are included because follow-up information on these fights reveals that most of them involve injury serious enough to require medical attention (Elliott, 1994).

Only the NYS reports the hazard rate for serious violence during the first two decades of life. 5 The hazard rate is the proportion of persons who initiate serious violence at a given age. Serious violence begins mostly between the ages of 12 and 20 (Figure 3-1). In fact, 85 percent of people who become involved in serious violence by age 27 report that their first act occurred between age 12 and 20. The onset of serious violence is negligible after age 23 and before age 10 (only 0.2 percent of arrests for serious violent crime in 1997 involved a child under age 10 [Maguire & Pastore, 1999]).

Figure 3-1. Hazard rate for initiating serious violence, by age, National Youth Survey.


Figure 3-1. Hazard rate for initiating serious violence, by age, National Youth Survey. Source: Elliott, 1994. 95% confidence intervals are all less than or equal to ± 1.0%.

The peak age of onset is 16, when about 5 percent of male adolescents report their first act of serious violence. The age of onset peaks somewhat later for white males (age 18) than for African American males (age 15). The hazard rate at the peak age also varies somewhat by race/ethnicity. It is lower for white males (5 percent) than for African American males (8 percent) (Elliott, 2000a). A similar finding is reported in the Pittsburgh Youth Survey (Huizinga et al., 1995). 6 No comparable hazard rates have been published for female youths, but other studies have found that they are generally lower. 7

Age-specific prevalence -- that is, the proportion of youths at any given age who report having committed at least one serious violent act -- is also greatest in the second decade of life. The NYS and the three city surveys find that, broadly speaking, age-specific prevalence among male youths ranges from about 8 to 20 percent between the ages of 12 and 20 (Figure 3-2). Among females, it ranges from 1 to 18 percent (Table 3-1). There is some variability across surveys, however. In general, the NYS has lower rates than the three city surveys, reflecting the difference between a national sample and local samples drawn from urban areas. The differences may also reflect the timing of the studies. NYS estimates cover the years 1976 to 1986, whereas the three city surveys cover the years 1986 or 1988 to 1994, the peak years of the violence epidemic (see Chapter 2). Nevertheless, the estimates for age 17 across these longitudinal surveys are in the same range as those for high school seniors in the Monitoring the Future survey (see Chapter 2).

Figure 3-2. Prevalence of serious violence among male youths, by age: four longitudinal surveys.


Figure 3-2. Prevalence of serious violence among male youths, by age: four longitudinal surveys. Sources: Denver Youth Survey (DYS), Pittsburgh Youth Study (PYS), Rochester Youth Development Study (RYDS): 1986-1999 (Huizinga et al., 1995); National Youth (more...)

Table 3-1. Prevalence of serious violence by age, sex, and race/ethnicity: four longitudinal surveys.


Table 3-1. Prevalence of serious violence by age, sex, and race/ethnicity: four longitudinal surveys.

Another key difference between the national and city surveys is the maturation effect, or the age at which serious violence begins to decline sharply during the transition to adulthood. The NYS shows a decline in age-specific prevalence starting in the late teen years and a steep drop-off by age 20. In contrast, the city surveys, which were begun more recently, do not show a decline in the late teen years (Huizinga et al., 1995), and they have not yet published data on prevalence in early adulthood. Therefore, it is too soon to tell whether or at what age more recent groups of youths will mature out of violence. It is possible that young people are staying violent longer.

Some important differences in age-specific prevalence by sex have emerged from the data (Table 3-1). Female adolescents have lower rates of serious violence throughout the second decade. For the NYS and Denver surveys, rates at age 12 are about twice as high for boys as for girls. Between age 12 and 15 in Rochester, the rates are fairly similar. For all studies, the rates for females at age 17 are about one-quarter the rates for males. In addition, the peak age of serious violent offending occurs a few years earlier among females, and their maturation out of serious violence is both earlier and steeper than males'.

Age-specific prevalence also varies by race/ethnicity (Table 3-1). The NYS finds a significant racial gap between ages 14 and 17, when rates for African American youths are 36 to 50 percent higher than those for white youths. The city surveys show an even wider gap between African American and white youths (Huizinga et al., 1995). Rates among Hispanic youths, reported only for Denver and Rochester, are similar to or lower than those reported by African American youths in these cities. The prevalence reported by Hispanic youths ranges from 6 to 12 percent in Denver and about 10 to 20 percent in Rochester. Possible reasons for developmental differences by sex, race, and ethnicity are discussed in Chapter 4. None of these comparisons takes into account the effects of poverty, education, housing, or other environmental conditions.

Cumulative Prevalence

Cumulative prevalence refers to the proportion of youths at any particular age who have ever committed a serious violent offense. 8 As a measure of violence, it tends to equalize rather than magnify differences across populations because it counts youths only once, regardless of when or how often they engaged in violent acts.

The most striking feature of the cumulative prevalence is its sheer magnitude: About 30 to 40 percent of male and 16 to 32 percent of female youths have committed a serious violent offense by age 17 (Figure 3-3). Although these rates are only slightly higher than those found in international studies, the represent a more serious set of offenses (Junger-Tas et al., 1994).

Figure 3-3. Cumulative prevalence of serious violence, by age, sex, and race: four longitudinal surveys.


Figure 3-3. Cumulative prevalence of serious violence, by age, sex, and race: four longitudinal surveys. Sources: National Youth Survey (NYS), 1976-1993 (Elliott, 1994); Rochester Youth Development Study (RYDS), 1986-1999; Denver Youth Survey (DYS), (more...)

The cumulative prevalence of youth violence is generally consistent across the four surveys. For male youths, the NYS shows it rising to about 40 percent and then leveling off beyond age 22. In the city surveys, it reaches more than 40 percent of male and 32 percent of female youths by age 17. Not only is the rise in cumulative prevalence by age 17 steeper in the city surveys, the magnitude is substantially higher.

These differences between a nationally representative sample and city samples are to be expected. The timing of the surveys may also explain some of the differences. For example, 17-year-olds in the NYS were interviewed at some point between 1976 and 1982, whereas 17-year-olds in the city surveys were interviewed at some point between 1988 and 1994, the era during which the self-reported prevalence of serious youth violence increased somewhat (see Chapter 2).

There is a pronounced difference in cumulative prevalence by sex. Among 17-year-olds, the Denver survey found a cumulative prevalence of 16 percent for female youths and 43 percent for male youths, whereas the Rochester survey showed 32 percent and 40 percent, respectively. Data on cumulative prevalence for females are not available in the Pittsburgh survey or the NYS. 9

On the other hand, there are few differences in cumulative prevalence by race over the teen years, according to the NYS. By age 23, white males had a cumulative prevalence of 38 percent and African American males had a cumulative prevalence of 45 percent, a difference that is not statistically significant (Figure 3-3).

Rates of Offending and Violent Careers

Violent youths commit a remarkably high number of crimes (Tolan & Gorman-Smith, 1998). An analysis of NYS data shows that these young people (both male and female) averaged 15.6 rapes, robberies, aggravated assaults, or some combination of these crimes over a 16-year period (1976 to 1992) (Elliott, 2000b). What's more, they averaged just over six serious violent offenses each during the years in which they were active (Elliott, 2000b). 10 This mean annual rate of offending is similar to rates reported in the three city surveys for males (about 5 to 9 serious violent offenses per year) (Huizinga et al., 1995) but much higher than the rates for females (2 to 4.5 per year).

It is noteworthy that the mean annual rate of individual offending appears to be essentially unchanged over the past two decades. This finding is corroborated by a study of trends among juvenile offenders processed by a county court system in Arizona (Snyder, 1998) and by an analysis of both National Crime Victimization Survey data and arrest records (Snyder & Sickmund, 2000). Finally, the Monitoring the Future survey (see Chapter 2) found no significant changes in individual offending rates for robbery or assault with injury between 1983 and 1993. 11

Career length has been variously defined as the number of years of active offending, the maximum number of consecutive years, or the span between the first and last year during which a young person meets the criteria for a serious violent offender (Blumstein et al., 1986). There are relatively few estimates of violent career lengths. In the NYS, the mean career length (number of years of active violent offending) was 2.6 years. The most frequent career length was 1 year (36.8 percent of serious violent offenders). Three-quarters of these serious violent youths had careers lasting 3 years or less, and 15 percent had careers of 5 years or more (Elliott, 2000b). Based on 5 years of data, the Denver survey reports that 42 percent of serious violent youths were active for only 1 year, 22 percent for 2 years, and 31 percent for 3 or more years (Huizinga et al., 1995).

The typical violent career comprises either a single year of continuous offending or a longer period of intermittent offending. Relatively few violent careers are characterized by years of uninterrupted violence. In Denver, well over half of the careers that lasted three or more years had at least one year with no violent offending; three-quarters of those that spanned a 5-year period had an intermittent pattern of offending (Thornberry et al., 1995). Evidence that most careers lasting more than 1 year were characterized by intermittent offending also surfaced in the NYS (Elliott et al., 1986). This intermittent pattern makes it difficult to identify serious violent offenders with cross-sectional studies or with longitudinal studies that have long periods between data collection (Huizinga et al., 1995).

In sum, these studies suggest that in most cases violent careers are relatively short and are characterized by intermittent offending. During active periods, however, most careers are marked by a high rate of violent offending -- up to 10 offenses per year (Elliott et al., 1986; Thornberry et al., 1995).

Developmental Pathway to Violence

Violent youths do not usually begin their careers with a serious violent offense. While the developmental pathway varies, depending on what types of behavior are monitored, studies generally agree that a violent career begins with relatively minor forms of antisocial or delinquent behavior. These acts later increase in frequency, seriousness, and variety, often progressing to serious violent behavior (Elliott, 1994; Loeber et al., 1998; Moffitt, 1993; Tolan & Gorman-Smith, 1998). Several complex pathways to serious violence have been proposed (Loeber, 1996; Elliott, 1994).

The NYS suggests that violence escalates over time. Most serious violent youths who engage in multiple types of violent behavior begin with aggravated assault, then add a robbery, and finally a rape. (Rape appears to be the endpoint of the progression, although there were not enough homicides in the NYS sample to include homicide in the analysis.) Robbery precedes rape in over 70 percent of cases in which both acts have been reported, and about 15 percent of serious violent offenders in the NYS reported having committed a rape (Elliott, 1994, 2000a). This sequence must be considered tentative because it is based on a single study.

When serious violence becomes part of a youth's repertoire of antisocial behavior, it does not substitute for less serious forms of violence; rather, it adds to them and escalates the overall frequency of violent acts. Thus, serious violent youths are high-frequency offenders who are involved in many less serious as well as serious offenses. These youths account for a major share of all criminal behavior, a pattern that is explored more fully in the next section.

Chronic Violent Offenders

A minority of serious violent youths are responsible for the overwhelming majority of serious violent crime, a finding supported by numerous self-report and arrest studies (Tolan & Gorman-Smith, 1998; Tracy & Kempf-Leonard, 1996). In the city surveys, chronic offenders, though representing less than 20 percent of all serious violent offenders, accounted for 75 to 80 percent of self-reported violent crimes (Huizinga et al., 1995). NYS data yield similar findings: Chronic offenders (youths with three or more violent offenses) accounted for 76 percent of all felony assaults and 89 percent of all robberies reported by offenders in 1980 (Elliott et al., 1989).

Chronic violent youths may also account for a disproportionate share of all youth crime. The NYS reveals that in 1980 these serious violent offenders accounted for 79 percent of all felony theft, 66 percent of all illegal services (primarily drug selling), and 50 percent of all self-reported crime (Elliott et al., 1989). In the Philadelphia Birth Cohort Study, 15 percent of youthful offenders accounted for 74 percent of all official crime (Tracy & Kempf-Leonard, 1996).

As noted earlier, youths whose violence begins before puberty are more likely to become chronic violent offenders (Loeber et al., 1998). In the Rochester survey, 39 percent of children who initiated violent behavior by age 9 eventually became chronic offenders, 30 percent of those who initiated violence between the ages of 10 and 12 became chronic offenders, and 23 percent of those who initiated violence after age 13 became chronic violent offenders. In Denver, 62 percent of those initiating violence by age 9 and 48 percent of those initiating violence between 10 and 12 became chronic violent offenders. Looking at this another way, 55 percent of all chronic violent offenders in Denver came from the early-onset trajectory (Thornberry et al., 1995). While the late-onset trajectory involves a substantially larger group of youths, fewer than half of all chronic offenders come from this group.

Although most chronic violent offenders in the three city surveys (62 to 77 percent) eventually had contact with the police for some offense (though not necessarily a violent offense), one-quarter to one-third were never arrested (Huizinga et al., 1995). Among those who were arrested for some offense, the first contact came well after they had begun their violent careers. Interventions by the justice system occur too late to prevent such youths from escalating from less serious offenses to serious violence. Fortunately, it appears that at least half of chronic violent offenders can be identified as being at risk in childhood.

Research has found a powerful relationship between membership in a gang and chronic involvement in serious violence (see review in Thornberry, 1998). As noted earlier (see Chapter 2), gang members, a relatively small proportion of the adolescent population, commit the majority of serious youth violence (see Spergel, 1990, and Thornberry, 1998, for reviews).


Between 1983 and 1993, adolescents were committing homicide at dramatically higher rates than in previous years (see Chapter 2). Did those youths represent a new breed of frequent, vicious, remorseless killers? Did the character of violent youths change during that time -- and is it still different today (Bennett et al., 1996)? The answer seems to be no, for several reasons.

First, the increase in homicides was similar across all age groups (see Chapter 2). This suggests that it resulted from a relatively sudden change in the environment that affected all youths rather than from a gradual change in the socialization process, which would have led to progressively more vicious youths with each succeeding age group. Second, the increase in homicides was highly specific to certain youths -- namely, African American males (Zimring, 1998); moreover, it did not take place among females (see Chapter 2). Third, during the violence epidemic, there was a decline in family members killed by youths (Cook & Laub, 1998).

Fourth, a new breed of superpredators should have resulted in more burglaries, auto thefts, and larcenies, but no such increases occurred (Cook & Laub, 1998). It should also have resulted in more homicides involving knives and other weapons, but this did not occur (Zimring, 1998). Fifth, there was no evidence that individual rates of serious violent crime changed during the epidemic. More youths were involved, but the average number of offenses committed by each did not change. Finally, there may be anecdotal evidence that today's youths show less remorse for their violence, but this has not been substantiated by research.

In sum, the epidemic of violence from 1983 to 1993 does not seem to have resulted from a basic change in the offending rates and viciousness of young offenders. Rather, it resulted primarily from a relatively sudden change in the social environment -- the introduction of guns into violent exchanges among youths. The violence epidemic was, in essence, the result of a change in the presence and type of weapon used, which increased the lethality of violent incidents (Wintemute, 2000).

Co-occurring Problem Behaviors

Serious violence is accompanied by a wide range of other problem behaviors, including property crimes, substance use, gun ownership, dropping out of school, early sexual activity, and reckless driving. The co-occurrence of these problem behaviors has been borne out by numerous national and local studies (see reviews in Elliott, 1993; Huizinga & Jakob-Chen, 1998; Tolan & Gorman-Smith, 1998).

The overlap is greatest between serious violence and other forms of crime. In the three city surveys, 82 to 92 percent of chronic violent youths were involved in property crimes, 71 to 82 percent in public disorder crimes, and 26 to 45 percent in selling drugs (Huizinga et al., 1995). Very similar rates were found in the NYS (Elliott et al., 1989). Rates of co-occurrence were much higher among serious violent youths than among less violent youths.

Substance use and abuse are a central feature of a violent lifestyle (Dembo et al., 1991; Elliott, 1994; Elliott et al., 1989; Esbensen & Huizinga, 1991; Fagan, 1993; Johnson et al., 1991). In the Denver survey, for example, about 58 percent of serious violent offenders were alcohol users and 34 percent were marijuana users. The prevalence and frequency of use were much lower in youths who were not seriously violent (Huizinga & Jakob-Chen, 1998). The NYS indicates that 94 percent of serious violent youths in 1980 were using alcohol, 85 percent were using marijuana, and 55 percent were using several illicit drugs. Over half (55 percent) were abusing drugs -- that is, they reported health or relationship problems, or both, associated with their drug use (Elliott et al., 1989).

Similar findings regarding the overlap of substance use and serious violence hold for the Rochester study (Thornberry et al., 1995). Moreover, chronic violent youths in Rochester and violent youths in the NYS had higher rates of dropping out of school, gun ownership and use, teenage sexual activity and parenthood, tobacco use, driving under the influence of alcohol or drugs, and gang membership than nonserious offenders or nonoffenders (Elliott, 1993; Thornberry et al., 1995).

In sum, these studies show that a sizable proportion of serious violent youths have co-occurring problem behaviors -- and at rates significantly higher than those of their less violent counterparts. However, by no means all serious violent youths or even all chronic violent youths have co-occurring problems. Moreover, not all youths with problem behaviors are seriously violent. The fact that serious violence and problem behaviors tend to occur together does not necessarily mean that one causes the other (see Chapter 4) (Elliott, 1993; Reiss & Roth, 1993).

Violence and Mental Health

The relationship between violence and mental health has been studied more intensively in adults than in young people. An earlier U.S. Surgeon General's report on mental health, after weighing the evidence, emphasized that the contribution of mental disorders to overall violence in the United States is very small. In fact, public fear is out of proportion to the actual risk of violence, which contributes to the stigmatizing of people with severe mental disorders (Link et al., 1999). Even though the risk of violence is low overall, it is greatest for adults with serious mental disorders who also abuse substances (Steadman et al., 1998; Swanson, 1994).

Although violence is relatively widespread among adolescents, few studies have been undertaken on the co-occurrence of violence and mental health problems or disorders among U.S. adolescents. Such population-based studies are important because they avoid the bias inherent in surveying hospitalized patients or convicted offenders.

Both the NYS and the Denver survey examine the co-occurrence of serious violence and mental health problems. In the NYS, 28 percent of serious offenders age 11 to 17 were classified as having mental health problems, compared to 13 to 14 percent of nonserious delinquent youths and 9 percent of nonoffenders. Youths were classified as having mental health problems on the basis of their responses to questions about emotional problems, social isolation, and feelings of loneliness (Elliott et al., 1989). (The questions were not designed to arrive at a diagnosis of a mental disorder.) Serious violent offenders were more likely than either nonserious offenders or nonoffenders to report having these types of mental health problems.

In the Denver study, serious violent youths were found to have higher rates of psychological problems, based on parents' responses to the Child Behavior Checklist (Achenbach & Edelbrock, 1983). These problems included externalizing and internalizing behavior, depression, uncommunicativeness, obsessive-compulsive behavior, hyperactivity, social withdrawal, and aggressiveness. The rates at which most of these problems occurred in serious offenders were no different from the rates at which they occurred in nonviolent delinquent youths; however, rates in nondelinquent youths were lower. Thus, delinquent youths in general were more likely to have psychological problems than nondelinquent youths (Huizinga & Jakob-Chen, 1998).

Two problems were linked directly to violent behavior -- externalizing symptoms and aggressive behavior. Approximately half of all serious violent offenders display these problems, although the link with externalizing behaviors is statistically significant only for boys. In addition, parents of violent offenders report seeking help for mental health problems more often than parents of nondelinquent or nonviolent delinquent youths. These parents did not go to mental health professionals or school counselors; rather, they sought the advice of friends, relatives, and spiritual leaders (ministers, rabbis, or priests). A similar finding is reported in the Pittsburgh study (Stouthamer-Loeber & Thomas, 1992).

The Denver study found no differences between the self-esteem of serious violent offenders and nonviolent offenders or nonoffenders (Huizinga & Jakob-Chen, 1998). In general, there is little evidence that low self-esteem causes violence or that violent offenders have low self-esteem. On the contrary, the evidence is more consistent with the position that high self-esteem and threats to high esteem lead to violence (Baumeister et al., 1996). This has important implications for treatment and intervention programs and the use of esteem-building activities in these programs.

A population-based study in New Zealand found that in young adulthood (age 21), serious violent offenders are more likely than nonoffenders to exhibit substance dependence disorders, schizophrenia-spectrum disorders, 12 or both (Arseneault et al., 2000). These New Zealand findings are consistent with the studies of U.S. adults showing that the greatest risk of violence stems from the combination of serious mental disorder and substance dependence. However, about 10 percent of serious violent offenders 13 in the New Zealand study exhibited schizophrenia-spectrum disorders without substance dependence or other psychiatric conditions. The researchers concluded that while the contribution of serious mental illness to violence in young adults remains small, it may be slightly higher than it is in adults. One possible reason for the difference is that the overwhelming majority of young adults with mental disorders in the New Zealand study had not been treated or hospitalized within the previous year.

Another recent community-based study found a link between personality disorders (a group of severe mental disorders) and violence. Adolescents with personality disorders, 14 as determined by diagnostic interviews, were more likely than other adolescents to commit violent acts such as assault with injury and robbery (Johnson et al., 2000). For example, about 36 percent of adolescents with personality disorders versus 16 percent without the disorders committed a violent act against others 15 during adolescence. The relationship between personality disorders and violence remained after taking many factors into account, including co-occurring depression, anxiety, and substance disorders. Only a few adolescents (13 percent) with personality disorders had received mental health services during the previous year (Johnston, personal communication, 2000).

Thus, there is some evidence of a relationship between serious mental disorders and violence in adolescents or young adults in the general population. Young people with serious mental disorders may be at risk of becoming violent if they also abuse substances or if they have not received treatment for their mental disorder. More research is needed to understand the relationship between serious youth violence and mental illness.

Offending and Victimization

Violent offenders are frequently victims of violence (Esbensen & Huizinga, 1991; Lauritsen et al., 1991; Sampson & Lauritsen, 1990, 1994). Data from the NYS reveal that victimization is highest among African Americans, males, and frequent offenders (Lauritsen et al., 1991). In addition, youths who report abusing drugs and alcohol, hanging out with delinquent peers, and participating in social activities with little adult supervision are at greater risk of being victims of violence (Gottfredson, 1984; Lauritsen et al., 1991; Sampson & Lauritsen, 1990). A delinquent lifestyle greatly increases the likelihood of being a victim and appears to account for some of the disparities observed in offending and victimization by race/ethnicity and sex. The Denver survey shows that 42 percent of serious violent offenders are also victims of violence (Huizinga & Jakob-Chen, 1998), with higher rates among male offenders than female offenders.

There are many reasons for the overlap between offending and victimization. Perhaps the most common is that the offender is injured by the intended target -- either during the offense or later, in retaliation. Another reason is that offenders tend to live in more violent environments or their lifestyles take them into high-risk environments. The predictive relationship between victimization and offending, as well as the relationship with early child abuse, is discussed in Chapter 4.

Transition to Adulthood

The transition from adolescence to adulthood features a fairly abrupt discontinuation of serious violence, at least according to the NYS. Rates of onset and age-specific prevalence show dramatic declines, and the cumulative prevalence levels off, as discussed above. Only about 20 percent of serious violent offenders continue their violent careers into their twenties (Elliott, 1994).

While there are no differences by sex in the apparent termination of violent offending, there are significant differences by race. Twice as many African American as white youths continue their violent behavior into the adult years (Elliott, 1994). Preliminary analyses suggest that cessation of offending is related to having a stable job and a stable intimate relationship.

By 1992, the most recent year for which data are available, many people monitored by the NYS had reached their late twenties and early thirties. There is virtually no published information about what patterns of violence may have continued into their adult years. The more recent city surveys have published age-specific and cumulative prevalence findings only up to age 19, but these studies are still being conducted. Some evidence from these surveys (Figure 3-2) suggests that violent careers are lasting longer, but additional waves of data are needed to verify this trend.

Understanding the demographics and dynamics of how patterns of serious violence change with the transition into adulthood is critical to designing programs that enhance the termination of violence.


The prevalence of serious violence by age 17 is startling. About 30 to 40 percent of male and 15 to 30 percent of female youths report having committed a serious violent offense at some point in their lives. This cumulative prevalence is similar among African American and white males, in contrast to other measures of violence, which show racial disparities (see Chapter 2).

Two general onset trajectories emerge from longitudinal studies of youth violence -- an early-onset trajectory that begins before puberty and a late-onset one that begins in adolescence. Youths in the early-onset trajectory generally commit more crimes, and more serious crimes, for a longer time. These young people exhibit a pattern of escalating violence through childhood and adolescence, and frequently into adulthood.

Most youths who become violent, however, begin in adolescence. Their late-onset offending is usually limited to a short period, peaking at about age 16 and dropping off dramatically by age 20. They typically show few signs in childhood that they will become violent later on, laying to rest the myth that all violent adolescents can be identified in childhood.

The rate of individual offending appears to have remained virtually unchanged, both during and since the years of the violence epidemic, which began in 1983 and peaked in 1993. This finding, together with evidence that the epidemic was specific to gun-related violence, challenges the myth that the early 1990s produced a generation of superpredators who were more vicious and who committed dramatically more crimes than earlier generations of young people. At the same time, the finding of a stable individual offending rate indicates that the violence epidemic has not altogether subsided.

Serious violence is frequently part of a lifestyle that includes drugs, guns, precocious sex, and other risky behaviors. Youths involved in serious violence typically commit many other types of crimes and exhibit other problem behaviors, presenting a serious challenge to intervention efforts. Successful interventions must confront not only the violent behavior of these young people, but also their lifestyles, which are teeming with risk.

Prevention and intervention programs must also take into account the different patterns of violence typical of the early- and late-onset trajectories, as well as the relatively constant rates of individual offending. Early childhood programs that target at-risk children and families are critical for preventing the onset of a chronic violent career, but programs must also be developed to combat late-onset violence. The importance of late-onset violence prevention is neither widely recognized nor well understood. Substantial numbers of serious violent offenders emerge seemingly without warning. A comprehensive community prevention strategy must address both onset patterns and ferret out their respective causes and risk factors.


  1. Achenbach, T.M & Edelbrock, C. S Manual for the child behavior checklist and revised child behavior profile. Burlington, VT: University of Vermont, Department of Psychiatry. (1983)
  2. Arseneault, L., Moffitt, T.E., Caspi, A., Taylor, P. J., & Silva, P. A Mental disorders and violence in a total birth cohort: Results from the Dunedin Study. Archives of General Psychiatry, . (2000);57:979–986. [PubMed: 11015816]
  3. Battin, S., Hill, K. G., Hawkins, J. D., Catalano, R. F., & Abbott, R Testing gang membership and association with antisocial peers as independent predictors of antisocial behavior: Gang members compared to non-gang members of law-violating youth groups. Paper presented at the annual meeting of the American Society of Criminology, Chicago. (1996)
  4. Baumeister, R.F., Smart, L., & Boden, J. M Relation of threatened egotism to violence and aggression: The dark side of high self-esteem. Psychological Review, . (1996);103:5–33. [PubMed: 8650299]
  5. Bennett, W.J., DiIulio, J. J., Jr., & Walters, J. P Body count. New York: Simon and Schuster. (1996)
  6. Blumstein, A., Cohen, J., Roth, J., & Visher, C (1986) Criminal careers and "career criminals." Washington, DC: National Academy Press.
  7. Cook, P.J & Laub, J. H (1998) The unprecedented epidemic in youth violence. In M. Tonry & M. H. Moore (Eds.), Youth violence. Crime and justice: A review of research (Vol. 24, pp. 27-64). Chicago: University of Chicago Press.
  8. Dembo, R., Williams, L., Getreu, A., Genung, L., Schmeidler, J., Berry, E., Wish, E., & LaVoie, L A longitudinal study of the relationships among marijuana/hashish use, cocaine use, and delinquency in a cohort of high-risk youths. Journal of Drug Issues, . (1991);21:271–312. [PubMed: 2090630]
  9. D'Unger, A.V., Land, K. C., McCall, P. L., & Nagan, D. S How many latent classes of delinquent/criminal careers? Results from mixed poisson regression analysis. American Journal of Sociology, . (1998);103:1593–1620.
  10. Elliott, D. S Violent offending over the life course. A sociological perspective. In N. A. Krasnegor, N. B. Anderson, & D. R. Bynum (Eds.), Health and behavior (Vol. 1, pp. 189-204). Rockville, MD: National Institutes of Health, Office of Behavioral and Social Sciences. (2000a)
  11. Elliott, D. S Special analysis prepared for this report by Delbert S. Elliott, principal investigator, National Youth Survey. (2000b)
  12. Elliott, D. S Serious violent offenders: Onset, developmental course, and termination. The American Society of Criminology 1993 presidential address. Criminology, . (1994);32:1–21.
  13. Elliott, D. S (1993) Health-enhancing and health-compromising lifestyles. In S. G. Millstein, A. C. Petersen, & E. O. Nightingale (Eds.), Promoting the health of adolescents: New directions for the twenty-first century (pp. 119-145). New York: Oxford University Press.
  14. Elliott, D.S., Hagan, J., & McCord, J Youth violence: Children at risk. Washington, DC: American Sociological Association. (1998)
  15. Elliott, D.S., Huizinga, D., & Menard, S (1989) Multiple problem youth: Delinquency, substance use, and mental health problems. New York: Springer-Verlag.
  16. Elliott, D.S., Huizinga, D., & Morse, B. J Self-reported violent offending: A descriptive analysis of juvenile violent offenders and their offending careers. Journal of Interpersonal Violence, . (1986);1:472–514.
  17. Esbensen, F.A & Huizinga, D Gangs, drugs, and delinquency in a survey of urban youth. Criminology, . (1993);31:565–589.
  18. Esbensen, F.A & Huizinga, D Juvenile victimization and delinquency. Youth and Society, . (1991);23:202–228.
  19. Fagan, J Interactions among drugs, alcohol, and violence. Health Affairs, . (1993);12:65–79. [PubMed: 8125449]
  20. Fagan, J Social processes of delinquency and drug use among urban gangs. In C. R. Huff (Ed.), Gangs in America (pp. 183-219). Thousand Oaks, CA: Sage Publications. (1990)
  21. Flanagan, T.J & Maguire, K Sourcebook of criminal justice statistics, 1991 (U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics). Washington, DC: U.S. Government Printing Office. [Also available on the World Wide Web: http://www​] (1992)
  22. Gottfredson, M. R Victims of crime: The dimension of risk (Home Office Research Study No. 81). London, United Kingdom: Her Majesty's Stationery Officer. (1984)
  23. Huizinga, D., & Jakob-Chen, C Contemporaneous co-occurrence of serious and violent juvenile offending and other problem behaviors. In R. Loeber & D. P. Farrington (Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 47-67). Thousand Oaks, CA: Sage Publications. (1998)
  24. Huizinga, D., Loeber, R., & Thornberry, T. P Recent findings from the program of research on the causes and correlates of delinquency (U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention, NCJ 159042). Washington, DC: U.S. Government Printing Office. (1995)
  25. Johnson, B., Wish, E., Schneider, J., & Huizinga, D Concentration of delinquent offending: Serious drug involvement and high delinquency rates. Journal of Drug Issues, . (1991);21:205–291.
  26. Johnson, J.G., Cohen, P., Smailes, E., Kasen, S., Oldham, J. M., Skodol, A. E., & Brook, J. S Adolescent personality disorders associated with violence and criminal behavior during adolescence and early adulthood. American Journal of Psychiatry, . (2000);157:1406–1412. [PubMed: 10964855]
  27. Johnston, L Personal communication. (2000)
  28. Junger-Tas, J., Terlouw, G. J., & Klein, M. W Delinquent behavior among young people in the western world: First results of the international self-report delinquency study. New York: Kugler Publications. (1994)
  29. Kaufman, P., Chen, X., Choy, S. P., Ruddy, S. A., Miller, A. K., Fleury, J. K., Chandler, K. A., Rand, M. R., Klaus, P., & Planty, M. G Indicators of school crime and safety, 2000 (NCES 2001-017/NCJ-184176). Washington, DC: U.S. Department of Education and U.S. Department of Justice. (2000)
  30. Klein, M. W (1995) The American street gang: Its nature, prevalence and control. New York: Oxford University Press.
  31. Lauritsen, J.L., Sampson, R. J., & Laub, J. H The link between offending and victimization among adolescents. Criminology, . (1991);29:265–292.
  32. Link, B.G., Phelan, J. C., Bresnahan, M., Stueve, A., & Pescosolido, B. A Public conceptions of mental illness: Labels, causes, dangerousness, and social distance. American Journal of Public Health, . (1999);89:1328–1333. [PMC free article: PMC1508784] [PubMed: 10474548]
  33. Loeber, R (1996) Developmental continuity, change, and pathways in male juvenile problem behaviors and delinquency. In J. D. Hawkins (Ed.), Delinquency and crime: Current theories (pp. 1-27). New York: Cambridge University Press.
  34. Loeber, R., Farrington, D. P., & Waschbusch, D. A Serious and violent juvenile offenders. In R. Loeber & D. P. Farrington (Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 13-29). Thousand Oaks, CA: Sage Publications. (1998)
  35. Maguire, K., & Pastore, A. L Sourcebook of criminal justice statistics, 1998 (U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics, NCJ 176356). Washington, DC: U.S. Government Printing Office. [Also available on the World Wide Web: http://www​] (1999)
  36. Moffitt, T. E Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, . (1993);100:674–701. [PubMed: 8255953]
  37. Nagin, D., & Tremblay, R. E Trajectories of boys' physical agression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development, . (1999);70:1181–1196. [PubMed: 10546339]
  38. 1997 National youth gang survey: Summary (NCJ 178891). Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. [Also available on the World Wide Web: http://www​​/pdffiles1/ojjdp/178891.pdf] (1999)
  39. Patterson, G.R & Yoerger, K A developmental model for late-onset delinquency. In D. W. Osgood (Ed.), Motivation and delinquency (Vol. 44, pp. 121-177). Lincoln, NE: Nebraska Symposium on Motivation. (1997)
  40. Reiss, A.J. Jr., & Roth, J. A (1993) Understanding and preventing violence. Washington, DC: National Academy Press.
  41. Sampson, R.J & Lauritsen, J. L (1994) Violent victimization and offending: Individual-, situational- and community-level risk factors. In A. J. Reiss, Jr. & J. A. Roth (Eds.), Understanding and preventing violence. Social influences (Vol. 3, pp. 1-114). Washington, DC: National Academy Press.
  42. Sampson, R.J & Lauritsen, J. L Deviant lifestyles, proximity to crime and the offender-victim link in personal violence. Journal of Research in Crime and Delinquency, . (1990);27:110–139.
  43. Snyder, H. N Serious, violent, and chronic juvenile offenders -- An assessment of the extent of and trends in officially recognized serious criminal behavior in a delinquent population. In R. Loeber & D. P. Farrington (Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 428-444). Thousand Oaks, CA: Sage Publications. (1998)
  44. Snyder, H.N & Sickmund, M Challenging the myths (1999 National Report Series). Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. (2000)
  45. Spergel, I. A (1990) Youth gangs: Continuity and change. In M. Tonry & N. Morris (Eds.), Youth violence. Crime and justice: A review of research (Vol. 12, pp. 171-275). Chicago: University of Chicago Press.
  46. Stattin, H., & Magnusson, D Antisocial development: A holistic approach. Development and Psychopathology, . (1996);8:617–645.
  47. Steadman, H.J., Mulvey, E. P., Monahan, J., Robbins, P. C., Appelbaum, P. S., Grisso, T., Roth, L. H., & Silver, E Violence by people discharged from acute psychiatric inpatient facilities and by others in the same neighborhoods. Archives of General Psychiatry, . (1998);55:393–401. [PubMed: 9596041]
  48. Stouthamer-Loeber, M., & Thomas, C Caretakers seeking help for boys with disruptive and delinquent behavior. Comprehensive Mental Health Care, . (1992);2:159–178.
  49. Swanson, J. W (1994) Mental disorder, substance abuse, and community violence: An epidemiological approach. In J. Monahan & J. W. Steadman (Eds.), Violence and mental disorders: Developments in risk assessment (pp. 101-136). Chicago: University of Chicago Press.
  50. Thornberry, T. P Membership in youth gangs and involvement in serious violent offending. In R. Loeber & D. P. Farrington (Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 147-166). Thousand Oaks, CA: Sage Publications. (1998)
  51. Thornberry, T.P., Huizinga, D., & Loeber, R The prevention of serious delinquency and violence: Implications from the program of research on the causes and correlates of delinquency. In J. C. Howell, B. Krisberg, J. D. Hawkins, & J. Wilson (Eds.), Sourcebook on serious, violent and chronic juvenile offenders (pp. 213-237). Thousand Oaks, CA: Sage Publications. (1995)
  52. Tolan, P. H Implications of onset for delinquency risk identification. Journal of Abnormal Child Psychology, . (1987);15:47–65. [PubMed: 3571739]
  53. Tolan, P.H & Gorman-Smith, D Development of serious and violent offending careers. In R. Loeber & D. P. Farrington (Eds.) Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 68-85). Thousand Oaks, CA: Sage Publications. (1998)
  54. Tracy, P.E & Kempf-Leonard, K (1996) Continuity and discontinuity in criminal careers. New York: Plenum Press.
  55. Wintemute, G (2000) Guns and gun violence. In A. Blumstein & J. Wallman (Eds.), The crime drop in America (pp. 45-96). Cambridge, United Kingdom: Cambridge University Press.
  56. Zimring, F. E (1998) American youth violence. Oxford, United Kingdom: Oxford University Press.



A higher proportion of serious violent offenders are male (Chapter 2).


Several other national longitudinal youth surveys are in progress, but none of them has tracked a sample long enough to provide descriptions of developmental trajectories in adolescence.


Confidence intervals for all four surveys are based on simple binomial distributions and do not reflect the full sampling designs of these studies.


In each survey, follow-up questions were asked to determine the seriousness and appropriateness of the reported event. Reports of nonserious events were not included. This adjustment could not be made for the first 3 years of the NYS, so for those years at least two serious violent offenses were required for a youth to be classified as a serious violent offender.


The rates for age 10 and under are based entirely upon retrospective reports and may not be as reliable as those for age 11 and older.


The actual hazard rates are not presented, but the investigators note that the rates were higher for African Americans between the ages of 12 and 16 (Huizinga et al., 1995).


Using a slightly different definition of serious violent offender (three or more serious violent offenses), the hazard rate for females peaked at age 14 (1.5 percent ± 1.0 percent) and dropped by age 17 (0.6 percent ± 0.5 percent) (Elliott et al., 1986).


Cumulative prevalence is also known as lifetime prevalence or ever-prevalence to a given age.


The NYS reports an 11 percent cumulative prevalence at age 21 among females who are chronic (committing three or more offenses per year) serious violent offenders.


Active involvement is defined as any year during which offenders committed one or more serious violent offenses.


Calculations were done by the senior scientific editor Elliott on the basis of Monitoring the Future prevalence and frequency data on aggravated assault and robbery contained in the 1991 and 1998 Sourcebooks of Criminal Justice Statistics (Flanagan & Maguire, 1992; Maguire & Pastore, 1999). Individual offending rates are based on estimated incident rates. For this calculation, frequencies associated with categorical scores were as follows: not at all = 0; once = 1; twice = 2; 3 or 4 times = 3.5; and 5 or more times = 5. Individual offending rates for robbery in both 1983 and 1993 were 1.8. Rates for assault with injury were 2.3 and 2.6 (not significant). The mean individual offending rate can remain relatively constant despite increases in prevalence and incident rates noted in Chapter 2.


This broad category includes individuals who responded "yes, definitely" when asked if they had positive symptoms of schizophrenia (hallucinations and delusions) and for whom other plausible explanations (such as major depressive episodes or the influence of alcohol or illicit drugs) could be ruled out. The classification of diagnoses was made with the Diagnostic Interview Schedule.


Defined by two or more types of violent offenses (simple assault, aggravated assault, robbery, rape, and gang fighting) or a conviction.


The personality disorders included in this study were the same disorders (e.g., schizotypal disorder) studied by Arseneault et al. (2000).


Included threats to injure others, initiation of physical fights, mugging, robberies, or assaults resulting in injury during the past 1 to 4 years or during the individual's lifetime. The acts were reported in 1985-1986 or 1991-1993.


  • PubReader
  • Print View
  • Cite this Page

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...