Stopping Violence Before It Starts

Identifying Early Predictors of Adolescent Violence

The nation's young people are increasingly affected by violence, both as its perpetrators and as its victims. Many violence prevention programs aim to reverse this trend, but few of these programs have been rigorously evaluated, and even fewer have been shown to work. To devise better programs, researchers need more information about what causes young people to become violent--specifically, they need to learn whether there are things about young adolescents' personalities and the environments in which they live that promote or inhibit violent behavior later on. Such information could provide the tools for parents, teachers, youth workers, and others to recognize the young adolescents who are most likely to become violent in the future and select these individuals for improved intervention programs.

To that end, a RAND research team led by Phyllis Ellickson identified "early predictors" of violence--personality traits, demographic characteristics, social influences, and behaviors in early adolescence that increase the likelihood of engaging in violent behavior a few years later. The researchers followed nearly 4,500 students from a wide range of communities in California and Oregon. As seventh-graders the students filled out a questionnaire that asked them to describe their lives and environments; five years later the same young people (now high school seniors or dropouts) completed a second questionnaire, this time detailing their involvement with violence. From the two sets of responses, the researchers distilled characteristics of seventh-graders and their environments that consistently led to violence a few years later. Statistical analysis revealed that certain characteristics and environments of seventh-graders predict

whether youths are likely to become violent
whether their victims are more likely to be family members and acquaintances or strangers
how frequently the youths will engage in violence
different types and amounts of violence for boys and girls.

Some seventh-grade traits can predict subsequent violence

Several characteristics of seventh-graders were consistently associated with an increased risk of violence during high school. Students were more likely to become perpetrators of violence if they did poorly in school, engaged in early deviant behavior such as stealing or getting in trouble in school, attended a middle school or junior high in which drug use was common, or transferred frequently from one elementary school to another (see "Overall Violence" column of Table 1). In addition, just being male increased the likelihood of a violent future.

Different traits predict different types of violence

RB4536. tab1
The study distinguished between predictors of two types of violent behavior--relational violence (violence that arises from disputes between family members, friends, or acquaintances) and predatory violence (violent behavior toward strangers, such as mugging, robbing, and gang assaults). Overall violence combines the two types. Table 1 shows which student and school-level characteristics predict which type of violence five years later.

At the student level, for example, seventh-graders who were slightly older than the majority of their classmates were less likely to commit relational violence later on. Whites were less likely than members of other ethnic groups to engage in predatory violence. At the school level, students who were shifted frequently among elementary schools or attended a middle school where drug use among students was relatively common were more likely to engage in relational violence in high school.

Different early traits predict the amount of subsequent violence

The study also offered clues about characteristics that affect how often adolescents will engage in overall violence, relational violence, and predatory violence.

Three characteristics--poor grades, high elementary school mobility, and early deviance--increased the likelihood that adolescents would become frequent perpetrators of overall violence five years later. Conversely, being female decreased the likelihood of frequent violent behavior. Only one characteristic--the actual prevalence of drug use in the adolescent's middle school--predicted frequent relational violence five years later.

For predatory violence, six characteristics affected the amount of violence adolescents committed: frequency of an adolescent's alcohol, cigarette, and marijuana use during seventh grade; higher levels of perceived drug use by one's middle school peers; being male; being multiracial; coming from a household in which both natural parents were present; and rebelliousness. However, the effects of the last two characteristics were contrary to the researchers' predictions. Adolescents from two-parent households were more likely and rebellious youth were less likely to be frequent perpetrators of predatory violence.

Predictors are different for boys and girls

Several characteristics--engaging in deviant behavior as a young adolescent and attending a middle school with comparatively high levels of drug use--increased the likelihood of future overall violence for both boys and girls. However, other predictors were different for boys and girls (see Table 2). For instance, getting poor grades in middle school increased the odds for different types of violence: relational violence for girls and predatory violence for boys.

Other characteristics acted as risk factors for girls but not for boys, and vice versa. For instance, seventh-grade girls with low self-esteem were more likely to engage in relational violence, but low self-esteem did not predict any type of violence in boys. Similarly, living in a poor neighborhood predicted relational and predatory violence only for girls. Adolescent boys, on the other hand, were more affected by high mobility during elementary school and by offers of drugs during the seventh grade, both of which predicted future relational violence for male students but not for female students. Being white lowered the odds of predatory violence for both sexes but diminished the likelihood of relational violence only for girls.

Conclusions

A deeper understanding of the roots of adolescent violence will allow researchers to target interventions to the adolescents who are most at risk, thereby creating more effective violence prevention programs. For instance, because early deviance and poor grades provide useful warning signals of later violence, prevention programs aimed at younger adolescents should include efforts to prevent or reduce troublesome behavior in school and poor academic performance. Such efforts should begin in elementary school. Similarly, the different effects of predictors on boys and girls suggest that violence prevention efforts should be sensitive to the special needs of both genders. For instance, programs should take into account the special vulnerability of girls with low self-esteem and of boys who frequently moved between elementary schools.

The results of this study also raise the possibility that programs aimed at preventing drug use may yield an added violence-reduction bonus. Further research is needed to determine whether this is the case.

RB-4536 (2001)

This research highlight summarizes work more fully documented in Phyllis L. Ellickson and Kimberly A. McGuigan, "Early Predictors of Adolescent Violence," American Journal of Public Health, Vol. 90, No. 4, April 2000, pp. 566-572.

This highlight is cosponsored by RAND Health and the RAND Child Policy Project. RAND Health furthers this mission by working to improve health and health care systems and advance understanding of how the organization and financing of care affect costs, quality, and access. The Child Policy Project provides easier access to objective information that will help improve policy and decisionmaking on children's issues.

All RAND publications are available from RAND Distribution Services, P.O. Box 2138, Santa Monica, CA 90407-2138 (Telephone: 310-451-7002; FAX: 310-451-6915; or Internet: order@rand.org).

Abstracts of all RAND documents are available for review on the web. RAND publications are distributed to the trade by National Book Network. RAND® is a registered trademark. RAND is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions or policies of its research sponsors.

Published 2001 by RAND

RAND's Home Page