Adolescent Femicide: A Population-Based Study | Adolescent Medicine | JAMA Pediatrics | JAMA Network
[Skip to Navigation]
Sign In
April 2003

Adolescent Femicide: A Population-Based Study

Author Affiliations

From the Departments of Pediatrics and Internal Medicine, Division of Community Pediatrics, School of Medicine (Dr Coyne-Beasley), the Injury Prevention Research Center (Drs Coyne-Beasley and Moracco), and the Departments of Maternal and Child Health (Dr Moracco), Health Behavior and Health Education (Dr Moracco), and Environmental Sciences and Engineering (Dr Casteel), School of Public Health, University of North Carolina at Chapel Hill. Dr Moracco is now with the Pacific Institute for Research and Evaluation, Chapel Hill.

Arch Pediatr Adolesc Med. 2003;157(4):355-360. doi:10.1001/archpedi.157.4.355

Background  Homicide is the third leading cause of deaths for girls aged 11 to 14 years and the second leading cause of death for girls aged 15 to 18 years. However, few studies examine the contextual issues of adolescent femicide, especially among 11- to 14-year-old victims.

Objectives  To obtain quantitative and contextual information about adolescent femicide, and to compare the context of femicide in younger vs older adolescents.

Methods  Data from the North Carolina medical examiner were analyzed for all 11- to 18-year-old female homicide victims during 1990 to 1995. Police interviews were conducted for 1993 to 1995 cases to determine context, the relationship of victim and perpetrator, and criminal histories.

Results  There were 90 victims; 63 were aged 15 to 18 years, 55 were killed with firearms, and 40 were behind in school. Of 37 femicides for which law enforcement interviews were conducted, the most common contexts were altercation (n = 9), broken or desired relationship (n = 8), reckless behavior with a firearm (n = 6), retaliation (n = 5), and drug related (n = 3). Most perpetrators were men (89%; n = 33), were older than their victims (mean age difference, 8 years), and had criminal records (59%; n = 21). Seventy-eight percent of victims (n = 29) were killed by an acquaintance or intimate partner.

Conclusions  Femicide contexts differed by age. Younger adolescents (aged 11-14 years) were more likely to be killed by a family member in the context of an argument than by an intimate partner or acquaintance in the context of a broken relationship or reckless behavior with a firearm. Many victims were engaged in high-risk behaviors, including dropping out of school, running away from home, using drugs, and dating much older men with criminal records. Intervention specialists targeting high-risk female adolescents should be aware that this population may also be at increased risk of femicide.

ADOLESCENT HOMICIDE rates in the United States increased sharply in the mid-1980s, followed by a steady decline in rates for persons aged 12 to 17 years, with a 59% decrease between 1991 and 2000.1,2 Adolescent boys who trafficked in illegal narcotics in urban environments were identified as being at high risk for homicide, primarily by firearms, and intervention strategies based on this research may have been partly responsible for the decline in homicide rates.3

However, adolescent homicide rates in the United States remain unacceptably high, exceeding those of other industrialized countries.4 In 1999, homicide was the second leading cause of death for African American boys aged 10 to 14 years and the leading cause of death for African American boys aged 15 to 19 years.5 From 1995 to 1998, homicide was the third leading cause of death for girls aged 11 to 14 years and the second leading cause of death for those aged 15 to 18 years.5

Most adolescent victims are boys; therefore, boys' experiences drive observed victimization trends, which obscures the unique characteristics and patterns of femicide, the killing of girls and women. Adult homicides and femicides have been shown to have differing contexts and causes.6-11 For example, women are much more likely than men to be killed by intimate partners, typically preceded by a history of domestic violence and the women's recent separation from their partners.11-19

Femicide studies have generally focused on women 18 years and older or have included a subset of adolescent girls within an age range (eg, 15 to 24 years).20,21 Some studies have demonstrated that most adolescent femicide victims are killed by an intimate partner or someone they know, either an acquaintance or family member.22 However, it is not known whether the circumstances of adolescent femicides are the same as those for adult women, nor whether they differ for younger vs older female adolescents.

Further elucidation of the epidemiological characteristics and contexts of adolescent femicide, including victims aged 11 to 14 years, would improve our understanding of adolescent femicide and potentially guide the design and implementation of prevention strategies specifically tailored to adolescent girls. The present study was undertaken to improve our understanding of adolescent femicide by obtaining quantitative and contextual information about the circumstances surrounding these femicides and by comparing the contexts of femicide in younger vs older adolescents.


Data sources

Medical Examiner Data

We reviewed all 1990 to 1995 case files in the North Carolina Medical Examiner database with the manner of death listed as "homicide" for female victims aged 11 to 18 years. Cases were designated as femicides based on the medical examiner's records, independent of the legal outcome. Case files contained the medical examiner's investigation report detailing the femicide circumstances, death certificate, autopsy and toxicology reports, and other documents.

Information was abstracted from the medical examiner database by trained research assistants and one of us (T.C.B.) and included the victims' sociodemographic information, the femicide setting, and the femicide method. The victim's blood alcohol level (from the toxicology report) was also recorded.

Law Enforcement Interviews

For the femicides that occurred from 1993 through 1995, additional information about victim and perpetrator characteristics, such as prior criminal records, weapon accessibility, relationship between the victim and perpetrator, and femicide circumstances, was obtained from telephone interviews with the investigating law enforcement officers. Before the interviews, the investigating officer and his or her chief received a letter that explained the study, asked the officer to review a specified case or cases, and told the officer to expect a telephone call for a 15-minute interview within 2 weeks after receipt of our letter. Structured interviews with standardized questions were conducted by 2 trained research assistants, supervised by one of us (T.C.B.). Law enforcement interviews were conducted only for the most recent 3 years to minimize problems with recall.

Operational definitions

Victim and Perpetrator Characteristics

"Younger adolescents" were those aged 11 to 14 years (middle-school age) and "older adolescents" were those aged 15 to 18 years (high-school age).

Adolescents 2 or more years behind their expected completed school grade (based on beginning first grade at age 6 years) were classified as "age-grade discrepant." A 2-year difference was used to allow for individuals with birthdays in the middle of the year.

Femicide Settings

The day and time of the fatal injury rather than the day and time of death were included because time may have passed before a victim died. Counties were designated as urban or rural based on the established metropolitan statistical areas. A metropolitan statistical area is an economically and socially integrated geographic unit centered on an urban area with a population of 50 000 or more residents.23

Femicide Contexts

Each femicide was assigned to a "context" category by the research team after a review of the case information. The context categories were created by one of us (T.C.B.) and are consistent with our previous adolescent homicide research.22 If information regarding the circumstances was missing from the medical examiner database, the context was determined exclusively by police reports. When contexts overlapped, we determined which theme predominated.

Drug-Related. The femicide directly involved drug sales, trafficking, or turf battles. This category also included retaliations and altercations that involved drugs (eg, misallocation of drugs or drug payments).

Altercations. The homicide took place during the heat of an altercation or argument regarding some issue other than drugs.

Retaliations. The homicide was precipitated by a prior real or perceived non–drug-related event or injury against the perpetrator (eg, transmission of a sexually transmitted infection or an altercation). A homicide following an altercation was coded as a retaliation when the victim and perpetrator left each other for a period and the perpetrator later returned and killed the victim.

Reckless Behavior. The homicide occurred as a consequence of behavior reflecting reckless disregard for firearm safety and lethality, such as playing Russian roulette or playing with, handling, or showing a loaded firearm to friends or family.

Unintended Victim. The victim was not involved in a confrontation or dispute but was related to someone involved in the dispute or was in the vicinity of the dispute. The perpetrator was allegedly trying to harm or injure someone else.

Robbery. The homicide occurred while personal property (other than drugs) was being stolen, whether the homicide victim was the robber or the one being robbed.

Desired or Broken Relationship. The victim severed a real or perceived romantic relationship with the perpetrator or would not enter into a romantic relationship.

Data analysis

Data were analyzed using the Stata statistical analysis system, version 5 (Stata Corp, College Station, Tex). Univariate and bivariate proportions were compared using Pearson χ2 or Fisher exact tests, as appropriate. Analyses of femicide contexts were conducted using information from the police interviews only for the 1993 to 1995 cases because the data were more detailed and comprehensive. Except where indicated, denominators used to calculate percentages included cases with missing data. The Institutional Review Board at the University of North Carolina, Chapel Hill, deemed the study exempt.


There were 90 femicide victims aged 11 to 18 years in the medical examiner's database from 1990 through 1995. Interviews with law enforcement officers were completed for 37 (86%) of the 43 femicides committed from 1993 through 1995.

Victim and perpetrator sociodemographic characteristics

Of the 90 femicide victims, 63 (70%) were 15- to 18-years-old, and 51 (57%) were black; 40 victims (44%) were age-grade discrepant (ie, behind in school) (Table 1). Law enforcement data revealed that 33 alleged perpetrators (89%) were male, and 22 (59%) were black. The age range was 13 through 71 years (mean ± SD age, 25 ± 2.3 years). The mean ± SD age difference between the victim and perpetrator was 8 ± 3.4 years; in all cases except 2, the victim was younger. In 9 cases (24%), individuals other than the identified perpetrator (most commonly 1-3 others) were present at the time of the homicide.

Table 1. 
Victim Sociodemographic Characteristics and Location of Femicides From Medical Examiner Data
Victim Sociodemographic Characteristics and Location of Femicides From Medical Examiner Data

Femicide settings and medical examiner data


Of the 100 counties in North Carolina, 42 were represented among the 90 cases. Sixty-four female adolescent victims (71%) were killed in urban counties (Table 1), which make up only 35% of North Carolina counties overall, and 72 victims (80%) were killed in the county where they resided.


The medical examiner's database indicated the location in 77 (86%) of 90 cases. Fifty adolescent femicides (65%) occurred within a residence or on residential property; in 35 (70%) of these cases, it was the victim's residence (Table 1), and 16 femicides (32%) occurred in a bedroom. Nonresidential locations included a street or highway, a natural body of water, and a forest. Younger adolescent femicide victims were more likely to be killed at home than were older victims (65% vs 50%; P = .01).

Femicide Method

Fifty-five victims (61%) were killed with firearms, of which 34 (62%) were handguns, 13 (23%) were shotguns, and 6 (11%) were rifles. Two (4%) were of unknown type because the firearm or bullet was not recovered. Other means of death included asphyxia, fighting, blunt instrument, drowning, and child abuse or neglect (Table 2).

Table 2. 
Femicide Methods From Medical Examiner Data
Femicide Methods From Medical Examiner Data


Alcohol was found in the blood of 15 victims (17%). Alcohol levels were not determined in 3 victims because of advanced decomposition. Blood alcohol levels ranged from 20 to 300 mg/dL (4-64 mmol/L) (median, 69 mg/dL [15 mmol/L]). Only one of the younger adolescent victims had a positive blood alcohol level.

Law enforcement interview data

Victim-Perpetrator Relationship

Of the 37 femicides for which police interviews were conducted, 12 (32%) victims were younger adolescents, and 25 (68%) were older adolescents. Seventeen (46%) of the alleged perpetrators were friends or acquaintances, 12 (32%) were intimate partners, and 7 (19%) were family members, most commonly a father or stepfather (n = 3). Older adolescents were more often killed by an acquaintance (12 [48%] of 25) or an intimate partner (11 [44%] of 25), whereas younger adolescents were more likely to be killed by family members (6 [50%] of 12) or acquaintances (5 [42%] of 12) (Table 3).

Table 3. 
Femicide Contexts and Victim-Perpetrator Relationships From Law Enforcement Interview Data
Femicide Contexts and Victim-Perpetrator Relationships From Law Enforcement Interview Data

Criminal Record

Twenty-one perpetrators (59%) had known criminal records, compared with 3 victims (8%). Most perpetrators had committed multiple offenses. The most common offenses were assaults (37%), drug-related offenses (32%), breaking and entering (23%), and trespassing (14%). One perpetrator was known to have committed another homicide. The victims' criminal offenses were nonviolent, such as shoplifting, truancy, and running away. All victims with known criminal records were aged 15 to 18 years.

Femicide Contexts

Based on the law enforcement interview data (n = 37), the 3 most common femicide contexts were altercation (9 [24%]), desired or broken relationship (8 [22%]), and reckless behavior (6 [16%]) (Table 3). In addition, sexual assault or rape was a component of 7 homicides (19%), and at least 4 victims were known to be pregnant when they were killed.

For the 12 younger adolescents, the most common femicide contexts were altercations (5 [42%]) and retaliations (3 [25%]). Only 1 young teenager was killed as a result of a desired or broken relationship, and none were killed in drug-related contexts. In contrast, the most common motive for the 25 older adolescents was a desired or broken relationship (7 [28%]), followed by reckless behavior with a firearm (5 [20%]).

Females were the alleged perpetrators in only 6 (16%) of 37 cases. The contexts of female-perpetrated femicides were restricted to altercations and retaliations. Female perpetrators used guns in only 2 cases; in 2 cases, they used knives, and asphyxia and fighting were the methods in 1 case each.

The majority of deaths in most context categories occurred in the home environment. All femicides with the context of desired or broken relationships or reckless use of a firearm occurred in homes, as did all but 1 femicide with the context of robbery or altercation. All drug-related events occurred outside the home. There were no known drug-related femicides in rural areas.

Altercations. There were 9 fatal altercations; in 5 (56%), a firearm was used. Five altercations occurred in the context of intimate partner or family violence. In 2 such cases, girls were killed when trying to protect their mothers who were being physically assaulted by intimate partners. In 2 other cases, victims were killed by their much older boyfriends. One girl was killed by her father for talking back to him.

Causes of the other altercations included 2 girls arguing over a boy, the victim telling on the perpetrator for smoking in school, calling someone a "bitch," and talking negatively about the perpetrator's friend.

Desired or Broken Relationships. A total of 8 adolescent femicides occurred when the victims had separated from the perpetrators (n = 6) or when they would not participate in a relationship that the perpetrator desired (n = 2). Guns were used in 6 (75%) of these cases. Five of the alleged perpetrators were current or former partners, 2 were acquaintances, and 1 was a stepfather. Three perpetrators shot themselves after they killed the victim, 3 cases involved a rape or sexual assault, and all but 1 of the victims (who was killed by her stepfather) were aged 15 to 18 years.

Reckless Behavior. Five of 6 deaths in this category resulted from individuals' playing around with guns or showing guns to others and allegedly not realizing that the gun was loaded. In 1 case, the victim was allegedly playing Russian roulette. Interestingly, in 3 of 6 cases reckless behavior and use of firearms resulted in the death of female adolescents at the hands of an intimate partner. Most of these guns were owned legally by the perpetrator or the perpetrator's parent and were not stored locked up. All these events took place inside a residence.

Retaliations. Five cases occurred in the context of retaliation. Two sisters were sexually assaulted and stabbed when their boyfriends became enraged upon finding pills they believed were for treating chlamydia. In 1 case, a victim had agreed to testify against her ex-boyfriend in court; in another, the victim threatened to tell the police about the perpetrator's (a tenant of her mother's) involvement in a series of armed robberies committed with stolen guns. In the remaining retaliation case, the victim was killed because her niece died of sudden infant death syndrome while she was baby-sitting.

Drug-Related. Three femicides occurred in drug-related contexts. Two friends were abducted and killed for allegedly stealing drugs from the perpetrators. One victim was raped and killed by her drug dealer boyfriend and his 2 friends because they thought she owed the perpetrator money for drugs. Two of 3 drug-related femicides were committed with handguns.

Unintended Victim. There were 3 cases in which the adolescent girl was not the intended victim. Of these, 1 girl died when a fire that was set by man engaged in a domestic dispute with his wife spread to her family's apartment. Another occurred when the victim's brother set their trailer on fire to retaliate against his mother. In the last case, a man shot his cousin because he was angry with her mother for breaking up with him.

Robbery. Two victims died during non–drug-related robberies. One girl who was "known to carry a lot of money on her" was accosted by 4 acquaintances in a park. The other case involved an 11-year-old girl who was raped and killed when a coworker of her father's broke into her house to rob it.


This study is one of the few population-based examinations of adolescent femicide to date. As with studies of adult femicide victims,8,11-13,24 we found that most adolescent victims were killed in their homes by someone they knew, often an intimate partner or family member. Although most were killed with firearms, the proportion is smaller than what has been reported nationally for both adolescent and adult males but is similar to percentages for adult femicide.

We also found that the circumstances surrounding femicides of younger adolescents (age range, 11-14 years) differed from those of older adolescents (age range, 15-18 years), particularly in terms of victim-perpetrator relationship and context. Younger adolescents were much more likely to be killed by family members and in the context of altercations, domestic violence, and child abuse, situations more typical of child homicides than adolescent homicides.25

Despite the complex, overlapping contexts for adolescent femicides, a number of themes emerged. Many victims appeared to live in chaotic home environments, often with the absence of adults or a lack of appropriate adult supervision. For example, all the "reckless behavior" femicides occurred in the victims' or perpetrators' homes, often in the middle of the night with an accessible firearm.

The findings of this study must be considered within the context of its methodological limitations. We included only adolescent femicides committed in North Carolina, and the findings may not be generalizable. The small number of cases, particularly for younger adolescents, prohibited multivariate analyses. We were also limited to information recorded in the medical examiner database and files or obtained through interviews with law enforcement officers; both sources varied in quantity and quality and may be subject to recall bias. Some deaths may have been misclassified. It is possible that some alleged perpetrators classified as acquaintances could have been current or former partners, thus underestimating intimate partner femicides among adolescents.

We often found it difficult to restrict femicides to a single context because of the complex circumstances surrounding these cases. For example, family and/or intimate partner violence was a recurrent theme that crossed multiple contexts. Current and former intimate partners (only one of whom was a husband) made up the single largest category of perpetrators, yet only 1 of the victims who was killed by an intimate partner was 18 years old and could have sought a domestic violence protective order under current North Carolina domestic violence legislation.26 Given the high proportion of adolescent femicides committed by nonmarital intimate partners, legal protection for teenagers involved in dating or domestic violence situations is necessary.

Finally, the contexts in which many of these young women died are not the same situations that are targeted by youth violence prevention programs, which tend to address situations more appropriate for adolescent boys (eg, conflict resolution or employment opportunities).27-31 However, it did appear that many victims were engaging in high-risk behaviors, such as dropping out of school, running away from home, reportedly using drugs, dating much older men who had criminal records, and engaging in high-risk sexual behaviors. At least 4 adolescent girls were pregnant and/or had young children, 15 victims aged 15 to 18 years (17%) had positive blood alcohol levels at the time of their deaths, and 40 victims (44%) were behind in school. Intervention specialists targeting high-risk female adolescents should be aware that these young women may also be at increased risk of femicide.


Corresponding author and reprints: Tamera Coyne-Beasley, MD, MPH, Department of Pediatrics, Division of Community Pediatrics, 130 Mason Farm Rd, 5th floor, Campus Box 7220, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7220 (e-mail:

Accepted for publication October 17, 2002.

This study was supported in part by the Minority Medical Faculty Development Program of the Robert Wood Johnson Foundation, Princeton, NJ (Dr Coyne-Beasley), and the Faculty Scholars Program of the William T. Grant Foundation, New York, NY (Dr Coyne-Beasley).

We thank the North Carolina Office of the Chief Medical Examiner, Chapel Hill, the law enforcement agencies whose participation made this project possible, Carol W. Runyan, PhD, Luenda Charles, PhD, and Mikesia Brown, MPH, for their assistance.

What This Study Adds

Much of the research on homicide has been focused on men's experiences as victims and perpetrators. The limited amount of research on femicide has focused on adult women and has indicated that the contexts of femicide differ markedly from those of homicide. Because preventive interventions are often based on epidemiological findings from population-based studies, it is important to understand the context of adolescent femicide in order to guide the development of appropriate prevention strategies.

Our findings indicate that adolescent femicide contexts are indeed different from those of adolescent homicide, and we identify several common contextual themes, including lack of appropriate adult supervision, accessibility of firearms in the home, and intimate partner violence. Programs aimed at adolescent girls should take into account that these contexts may put adolescents at increased risk for femicide.

Fox  JAZawitz  MW Homicide Trends in the United States.  Bureau of Justice Statistics, Department of JusticeAvailable at: December 2001
Klaus  PRennison  CM Age Patterns in Violent Victimization, 1976-2000.  Washington, DC Bureau of Justice Statistics, Dept of Justice2002;Available at: 190104Accessed February 2002
Blumstein  ARivara  FPRosenfield  R The rise and decline of homicide—and why.  Annu Rev Public Health. 2000;21505- 541Google ScholarCrossref
Centers for Disease Control and Prevention, Rates of homicide, suicide, and firearm-related death among children—26 industrialized countries.  MMWR Morb Mortal Wkly Rep. 1997;46 ((5)) 101- 105Google Scholar
Anderson  RN Deaths: Leading Causes for 2000.  Hyattsville, Md National Center for Health Statistics2002;9National Vital Statistics Report 50 (16)
Goetting  A Female victims of homicide: a portrait of their killers and the circumstances of their deaths.  Violence Vict. 1991;6159- 168Google Scholar
Barnard  GWVera  HVera  MINewman  G Till death do us part: a study of spouse murder.  Bull Am Acad Psychiatry Law. 1982;10271- 280Google Scholar
Puzone  CASaltzman  LEKresnow  MJThompson  MPMercy  JA National trends in intimate partner homicide: United States, 1976-1995.  Violence Against Women. 2000;6409- 426Google ScholarCrossref
Kellermann  ALMercy  JC Men, women, and murder: gender-specific difference in rates of fatal violence and victimization.  J Trauma. 1992;331- 5Google ScholarCrossref
Block  CRChistakos  A Intimate partner homicide in Chicago, over 29 years.  Crime Delinq. 1995;41496- 526Google ScholarCrossref
Smith  PHMoracco  KEButts  JD Partner homicide in context: a population-based perspective.  Homicide Stud. 1998;2400- 421Google ScholarCrossref
Moracco  KERunyan  CWButts  JD Femicide in North Carolina, 1991-1993: a statewide study of patterns and precursors.  Homicide Stud. 1998;2422- 446Google ScholarCrossref
Arbuckle  JOlson  LHoward  MBrillman  JAnctil  CSklar  D Safe at home: domestic violence and other homicides among women in New Mexico.  Ann Emerg Med. 1996;27210- 215Google ScholarCrossref
Stark  EFlitcraft  A Preventing gendered homicide.  Women at Risk: Domestic Violence and Women's Health Thousand Oaks, Calif Sage Publications1996;121- 153Google Scholar
Browne  AWilliams  K Gender, intimacy, and lethal violence: trends from 1967 through 1987.  Gend Soc. 1993;778- 98Google ScholarCrossref
Wilson  MDaly  M Spousal homicide risk and estrangement.  Violence Vict. 1993;83- 16Google Scholar
Campbell  JCRadford  JedRussell  Ded "If I can't have you, no one can": power and control in homicide of female partners.  Femicide: the Politics of Woman Killing New York Twayne Publishers1992;99- 113Google Scholar
Sev'er  A Recent or imminent separation and intimate violence against women.  Violence Against Women. 1997;3566- 589Google ScholarCrossref
Ellis  DDeKeseredy  WS Rethinking estrangement, interventions, and intimate femicide.  Violence Against Women. 1997;3590- 609Google ScholarCrossref
Fingerhut  LAIngram  DDFeldman  JJ Homicide rates among US teenagers and young adults: differences by mechanism, level of urbanization, race, and sex, 1987 through 1995.  JAMA. 1998;280423- 427Google ScholarCrossref
Dannenberg  ALBaker  SPLi  G Intentional and unintentional injuries in women: an overview.  Ann Epidemiol. 1994;4133- 139Google ScholarCrossref
Coyne-Beasley  TSchoenbach  VJHerman-Giddens  ME The epidemiology of adolescent homicide in North Carolina from 1990 to 1995.  Arch Pediatr Adolesc Med. 1999;153349- 356Google ScholarCrossref
Hewitt  MGesler  WedRicketts  Ted Defining rural areas: impact on health care policy and research.  Health in Rural North America: the Geography of Health Care Services and Delivery. New Brunswick, NJ Rutgers University Press1992;25- 54Google Scholar
Wilt  SAIllman  SMBrodyfield  M Female Homicide Victims in New York City, 1990-1994.  New York New York City Dept of Health, Injury Prevention Program1997;
Herman-Giddens  MEBrown  GVerbiest  S  et al.  Underascertainment of child abuse mortality in the United States.  JAMA. 1999;282463- 467Google ScholarCrossref
Not Available, NC Gen Stat ch 50B, §15A-401.
Aber  JLBrown  JLChaudry  NJones  SMSamples  F The Evaluation of the Resolving Conflict Creatively Program: an overview.  Am J Prev Med. 1996;12(5 Suppl)82- 90Google Scholar
Dahlberg  LL Youth violence in the United States: major trends, risk factors, and prevention approaches.  Am J Prev Med. 1998;14259- 272Google ScholarCrossref
Gabriel  RMHopson  THaskins  MPowell  KE Building relationships and resilience in the prevention of youth violence.  Am J Prev Med. 1996;12(5 Suppl)48- 55Google Scholar
Henggeler  SCunningham  PBPickrel  SGSchoenwald  SKBrondino  JM Multisystemic therapy: an effective violence prevention approach for serious juvenile offenders.  J Adolesc. 1996;1947- 61Google ScholarCrossref
Ringwalt  CLGraham  LAPaschall  MJFlewelling  RLBrowne  DC Supporting Adolescents with Guidance and Employment (SAGE).  Am J Prev Med. 1996;12(5 Suppl)31- 38Google Scholar