Abstract
Objectives: Given the high rates at which adolescents engage in violence, the strong link between adolescent and adult violence, and the financial and social costs of violence, the prevention of violent behavior is a national priority. Methods: The authors conducted a comprehensive review of evaluations utilizing quasi-experimental or experimental research designs to assess violence reduction. Results: Seventeen interventions were identified as producing a significant reduction in youth-perpetrated physical or sexual violence. The interventions were varied in terms of targeted age groups, content, strategies, and length of programming, and included programs focusing on individual, peer, school, family, and community factors. Conclusions: Widespread use of effective interventions such as these, across settings and development, and implemented with fidelity, is likely to substantially reduce youth violence. Continued evaluation of violence prevention programming is also needed to increase the number of options available for replication and establish effects on diverse populations.
Keywords
Introduction
Although decreasing somewhat in the last two decades, serious violent offending by adolescents continues to occur at high rates. The Federal Bureau of Investigation reported that approximately 86,000 adolescents were arrested for violent crimes in 2009 (Puzzanchera & Adams, 2011). Data from the national Youth Risk Behavior Survey indicated that in 2009, 32% of high school students reported past-year fighting and 18% reported carrying a weapon in the past month (Center for Disease Control, 2010). Involvement in violent crime by youth has serious negative consequences during adolescence and into adulthood, including low educational attainment, impaired social relationships, compromised mental health, and increased rates of victimization and physical injury (Fergusson, Lynskey, & Horwood, 1996; Moffitt, 1993; Patterson, DeBaryshe, & Ramsey, 1989; Sampson & Laub, 1993). Further, violence is the second leading cause of mortality for adolescents worldwide (Patton et al., 2009).
While much attention is paid to violent adult offending and its associated costs, youth are significantly more likely than adults to engage in violence (Puzzanchera & Adams, 2011), and violent offending by youth is associated with longer, more frequent and more serious involvement in crime over the life span (Elliott, 1994; Farrington, 2003; Moffitt, 1993). These findings emphasize the need to intervene early in the life course in order to reduce the likelihood that adolescents will engage in violence and to minimize the many harms that follow from this behavior. Fortunately, mechanisms that successfully prevent violence are increasingly being identified, largely through the efforts of prevention scientists from a range of disciplines, including medicine, psychology, public health, sociology, and social work. The goal of the current study is to identify effective violence prevention interventions that have been evaluated using high-quality research designs and which have targeted youth from birth through late adolescence. In doing so, we hope to provide a comprehensive view of “what works” in the field of violence prevention, encourage greater dissemination of these practices, and identify gaps in terms of research, evaluation, delivery, and widespread use of effective violence prevention strategies.
The Development and Precursors of Violent Behavior
It is widely acknowledged that violent behavior displays a developmental progression, in that the first acts of violence committed by an individual typically occur during middle to late adolescence, the prevalence and frequency of violence peak during late adolescence to early adulthood, and the likelihood of violence wanes as the individual moves into adulthood (Elliott, 1994; Farrington, 2003). Furthermore, it is recognized that there are multiple causes of or pathways to violence, including individual, peer, family, school, and community characteristics (Herrenkohl, Chung, & Catalano, 2004; Herrenkohl et al., 2000; Lipsey & Derzon, 1998). These factors, may increase (in the case of risk factors) or decrease (in the case of protective or promotive factors) the likelihood that youth will engage in violent activities.
Risk and protective/promotive factors also show a progression over the life course, emerging and/or becoming salient at particular developmental periods. During early childhood, for example, family risk factors such as child abuse and neglect and parenting practices such as monitoring, supervision, and reinforcement may be most important in affecting violence. During middle childhood and adolescence, school and peer experiences (e.g., poor academic performance, low commitment to school, and association with delinquent peers) become more influential. Community factors, such as availability of drugs or weapons and informal social control, tend to gain precedence during middle to late adolescence when children spend more time in the community (Catalano & Hawkins, 1996; Farrington, 2003; U.S. Department of Health and Human Services, 2001).
The Science of Prevention
The emerging field of prevention science (Catalano et al., 2012; Coie et al., 1993) seeks to synthesize and apply knowledge regarding these precursors of violence (and other problem behaviors) in order to reduce the prevalence of violence over the life course. A foundation of prevention science is the recognition that if violence is developmental, there are opportunities to intervene to reduce the progression of this behavioral trajectory. That is, interventions that seek to reduce the risk factors and increase the protective/promotive factors shown empirically to lead to violence can be utilized to reduce this behavior. Thus, prevention science advocates for the development and testing of interventions which seek to change risk and protective/promotive factors. To have maximal impact, such programs should focus on the factors that are most salient for the age group/groups targeted for services (Beardslee, Chien, & Bell, 2011; Coie et al., 1993). For example, a program for toddlers might focus on altering the family environment, while interventions for adolescents would seek changes in peer and community processes.
Another key objective of prevention science is to produce community-wide changes in targeted behaviors. In order to do so, communities need to offer a continuum of services for diverse populations of youth, including youth at all stages of development (Coie et al., 1993; Munoz, Mrazek, & Haggerty, 1996). Prevention services should include universal interventions intended to reach the general population of youth prior to their involvement in violence, selective interventions targeting youth with elevated levels of risk factors and low levels of protective factors, and indicated programs for those who have initiated violence but have not yet become serious or frequent violent offenders. By delivering services during every stage of development and for youth from all backgrounds and levels of risk, communities are best able to ensure reductions in violent behaviors among children. Because research has shown that the same risk and protective/promotive factors can affect multiple problem behaviors (Catalano, Haggerty, Hawkins, & Elgin, 2011; Coie et al., 1993; Jessor & Jessor, 1977), interventions that successfully decrease risk and enhance protection should reduce not only violence but also substance use, nonviolent offending, risky sexual behavior, and other related problems.
To date, many violence prevention interventions have been developed to target salient risk and protective/promotive factors, and many communities are implementing such interventions with their youth. Such efforts have been encouraged and/or mandated by federal agencies and initiatives, including the No Child Left Behind Act of 2001, the Institute of Medicine (National Research Council and Institute of Medicine, 2009), and the National Prevention Strategy released by the U.S. Surgeon General in 2011 (National Prevention Council, 2011). Despite an increasing emphasis on the creation, testing, and implementation of effective violence prevention programs, not all such interventions have demonstrated efficacy, not all have been evaluated using rigorous methodologies, not all have produced large or long-term effects, and not all have been equally effective across diverse populations. In addition, information about what works has not been effectively communicated to practitioners (Kerner & Hall, 2009; Saul et al., 2008). As a result, most community practice is not based on evidence of effectiveness (Glasgow, Lichtenstein, & Marcus, 2003; Ringwalt et al., 2009; Saul et al., 2008). That is, communities most often implement preventive interventions that have not been evaluated or which show no evidence of success in reducing problem behaviors, which hampers the ability to significantly reduce rates of youth violence community- and nation-wide.
The Current Study
The goal of the current study is to identify effective violence prevention interventions in order to encourage their widespread dissemination. Several comprehensive reviews have previously been undertaken to identify policies, practices, and programs that have been demonstrated as effective in reducing violence among youth (Hahn et al., 2007; Limbos et al., 2007; Mercy, Butchart, Rosenberg, Dahlberg, & Harvey, 2008; Mytton, DiGuiseppi, Gough, Taylor, & Logan, 2009; U.S. Department of Health and Human Services, 2001). The current review can be distinguished from past research on this topic in several ways. First, our review of the literature is guided by a developmental perspective of human behavior and by the principles of prevention science. As such, this study considers interventions targeting children aged 0–18 and which address risk and protective/promotive factors from all domains of children’s lives, including individual-, school-, family-, and community-focused strategies. Other reviews have taken a more limited approach in focusing on a particular age group (e.g., youth aged 12–17; Limbos et al., 2007) or on interventions targeting risk and protective/promotive factors from just one context (i.e., the school context; see Hahn et al., 2007; Mytton et al., 2009).
This review seeks to identify the effects of specific interventions on youth violence, which differs from the meta-analysis approach used in some other reviews (Hahn et al., 2007; Mercy et al., 2008; see also Wilson & Lipsey, 2007), in which the effects of discrete interventions are calculated and combined to produce an “average” effectiveness score across all evaluations. Meta-analyses examine whether or not a program type (e.g., school-based curricula or mentoring) is effective. The current review aims to foster widespread dissemination of effective, discrete programs. We are wary of advocating that a type of intervention be implemented if particular programs within that type are ineffective or iatrogenic, which is often the case when average effects are calculated. In addition, the identification of particular interventions which target risk and protective/promotive factors for specific age groups allows communities/practitioners to select the program that best fit the specific needs of their particular population/client/clients. Meta-analyses, which advocate for program types or strategies, not specific interventions, have less developmental and risk and protective factor specificity; as a result, practices based on recommendations from meta-analyses may have a diminished likelihood of effectiveness when locally implemented.
The current review is most similar in scope to the Surgeon General’s report on youth violence (U.S. Department of Health and Human Services, 2001), which also reviewed a wide range of specific violence prevention programs targeting youth from the prenatal period through early adulthood. The current review provides an updated review of this literature, which is necessary, given the rapid changes in prevention science. In the past decade, and even the past few years, there has been significant growth in the number of program evaluations and in the sophistication of the research design methods and statistical procedures used to assess efficacy (Catalano et al., 2012; Fagan & Eisenberg, 2012). To illustrate, six of the programs identified as effective in this article were based on evaluations published in the last 5 years. In addition, prior evaluations of the Safe Dates program suggested that it was ineffective (e.g., see Chan et al., 2004), but a more recent evaluation (Foshee et al., 2005) utilizing longer follow-up data and more advanced statistical procedures demonstrated significant reductions in violence for intervention youth.
Method
Inclusion/Exclusion Criteria
The current study can also be distinguished from past research in the set of criteria used to determine eligibility for the review and when determining program effectiveness. Studies were eligible for inclusion in this review if they assessed the actual perpetration of physical or sexual violence by youth aged 0–18. Violence was defined here as interpersonal, aggressive behavior in which an individual attempted to and/or actually inflicted harm upon another person. This definition is similar to that used by Reiss and Roth (1994). It includes acts of aggression and the terms are used interchangeably in this review.
Other systematic reviews have classified violence using a broader set of behaviors, including, for example, general delinquency, verbal aggression, bullying, school disciplinary violations, and oppositional or externalizing behaviors. Likewise, other reviews have evaluated effects on violence based on measures assessing verbal (but not physical) aggressive behavior or which combined violent and nonviolent acts. Well-known examples of such measures include, for example, the aggression and externalizing scales of the Child Behavior Checklist (Achenbach & Edelbrock, 1983), the intensity subscale of the Eyberg Child Behavior Inventory (Eyberg & Robinson, 1983), the Teacher Observation of Classroom Adaptation–Revised (Werthamer-Larsson, Kellam, Everett-Reynolds, & Beaudry, 1994), and teacher ratings of aggression on the Behavioral Assessment System for Children (Reynolds & Kamphaus, 1992). Studies that relied solely on these measurement instruments to assess violence were excluded from the review, lest the findings be driven by the perpetration of the less serious and nonphysically aggressive behaviors included in these measures. Although we recognize that delinquent and oppositional behaviors may co-occur with violence and/or have similar causes, these acts are also fairly normative during adolescence (Farrington, 2003; Moffitt, 1993). We restricted our assessment to the perpetration of interpersonal, physically, and sexually violent behaviors because we consider these to be of greatest concern to the public and the most likely to lead to long-term, detrimental outcomes for youth.
We limited the review to evaluations utilizing a well-conducted quasi-experimental or true experimental design. Minimum criteria for inclusion were that evaluations involve one experimental and one comparison group and at least two waves of data collection (i.e., pretests and posttests). Demonstration of sustained effects beyond the end of the intervention was not a criterion in our review. Other aspects of the research design were evaluated to ensure that threats to internal validity were minimal. The lead author reviewed studies to determine whether or not there was appropriate assignment of participants to conditions, baseline comparability, minimal and nondifferential attrition, an intent-to-treat approach, valid and reliable instruments to assess outcomes, and appropriate statistical techniques. Where there were questions regarding the rigor of the research methodology, the two authors conferred and reached agreement on the eligibility of the study.
Universal, selective and indicated interventions were all eligible for inclusion. For the latter, we focused on interventions targeting youth who may have been involved with the juvenile justice system, but who were mandated to community-based services, not interventions carried out by juvenile justice officials, in order to retain a focus on preventive, not treatment, services.
Search Method
A comprehensive literature search was conducted to identify published evaluations of violence prevention interventions. We began by examining prior systematic reviews, including those cited in the literature review as well as reviews of interventions targeting reductions in related problem behaviors, such as delinquency, drug abuse, and mental health problems. We also reviewed program evaluations identified on web-based and published lists of best practices related to violence prevention, including the Blueprints for Violence Prevention database (Center for the Study and Prevention of Violence, 2010), the Campbell Collaboration (C2; http://www.campbellcollaboration.org/), the Communities That Care Prevention Strategies Guide (http://www.sdrg.org/ctcresource/), the Model Programs Guide (Office of Juvenile Justice and Delinquency Prevention, 2010), the Office of Justice Programs (CrimeSolutions.gov), and the National Registry of Evidence-based Programs and Practices (Center for Substance Abuse Prevention, 2010). A final step involved searching the Medline, PsycINFO, Social Sciences Citation Index, and Web of Sciences databases with key words including aggression, crime, delinquency, prevention, youth, and violence. The reference lists of all studies reviewed were used to identify additional evaluations until a saturation point was reached.
Results
Table 1 lists the programs identified as effective based on one or more evaluations of the intervention indicating a statistically significant effect on violence using two-tailed tests and a significance level of at least p < .05. If the evaluation reported effects using more generous standards, we adjusted the estimates to reflect a p value of less than .05. Some of the interventions listed in Table 1 have been subject to multiple evaluations, but only those studies that assessed violent behaviors were eligible for review. For programs in which violence was assessed using multiple indicators and/or in multiple studies, significant reductions in violence are noted in Table 1, and any nonsignificant or iatrogenic effects reported in these studies are described in the following text.
Preventive Interventions With Significant Effects on Youth Violence.
Notes. QED = quasi-experimental research design; RCT = randomized control trial; OR = odds ratio; ES = effect size; RR = relative reduction; AA = African American; NS = nonsignificant; DSM = Diagnostic and Statistical Manual of Mental Disorder.
aCost estimates are per participant, based on 2003 dollars for the Big Brothers/Big Sisters program (Aos et al., 2004), 2004 dollars for Communities That Care (Kuklinski et al., 2012), 2007 dollars for the Chicago Child–Parent Centers (Lee et al., 2008), and 2010 U.S. dollars for all other interventions (Aos et al., 2011). bBased on official records collected from local, state, and/or federal corrections agencies. cBased on official records collected from local, state, and/or federal corrections agencies. dFour items derived from the National Youth Survey (Elliott, Huizinga, & Ageton, 1985): picking a fight with someone, hitting someone to seriously hurt them, beating someone up, and taking part in a group fight. eFour items: hitting others, shoving others, trying to hurt others, and getting into fist-fights (Deffenbacher & Swaim, 1999). fPhysical violence was based on 16 items (e.g., slapping, kicking, hitting a partner) at the 1-month follow-up and 12 items at the 3-year follow-up; sexual violence was based on two items: forced sex and forced to do sexual things. gThree items from the National Youth Survey (Elliott et al., 1985): hitting or threatening to hit a parent, school personnel, or another student. hTeachers and parents both reported on 3 items from the Preschool and Social Behavior Questionnaires (Behar & Stringfield, 1974): child fights, kicks/bites/hits, and intimidates other children. iAssessed using the Interpersonal Process Code (IPC; Rusby, Estes, & Dishion, 1991). jEight items from the National Youth Survey (Elliott et al., 1985); for example, picked a fight, beat someone up so badly they needed a doctor, kicked or hit a teacher. kFour items from the Behavioral Assessment System for Children Teacher Response Scale (Reynolds & Kamphaus, 1992) and 5 items from the Problem Behavior Frequency Scale (Farrell, Kung, White, & Valois, 2000). lFive items from the National Youth Survey (Elliott et al., 1985); for example, starting a fight, taking a handgun to school, hitting someone. mSelf-reported minor/major assault and robbery using items from the National Youth Survey (Elliott et al., 1985); official reports from state corrections agencies. n1-year follow-up relied on 3 felony assault items (aggravated assault, sexual assault, gang fights); 2-year follow-up on 10 items (simple and felony assault); both taken from the National Youth Survey (Elliott et al., 1985). oSeven items; for example, assault, mugging, robbery, arson, gang fighting, and shooting someone. pOne item, “hit someone,” from the Self-Perception Profile for Children (Harter, 1985). qBased on official records collected from state corrections agencies. rThree items from the National Youth Survey (Elliott et al., 1985): attacking someone with intent to harm, carrying a gun to school, beating somebody up.
The table is arranged according to program type, which largely corresponds to the type/types or domain/domains of risk and protective/promotive factors targeted for change. Within the five program types, interventions are ordered according to the targeted age of participants. The table columns, from left to right, provide (a) the intervention name (and evaluation citations); (b) a brief description of the intervention; (c) information on the sample (whether or not the program was universal, selective, or indicated; sample size and demographic characteristics at baseline; and sample size and attrition at follow-up, if applicable); (d) the study design (quasi-experimental or randomized controlled trial); (e) significant outcomes and the instruments used to assess outcomes; and (f) cost–benefit information, when available.
Regarding the significant intervention effects, when the information was provided by the authors, we report the prevalence rates and/or mean scores on outcomes for each program condition (E = experimental; C = control). For dichotomous outcomes (e.g., the prevalence of self-reported violence or arrest rates for violent crimes), we report odds ratios indicating the size of the difference between the intervention and control groups. For continuous outcomes, we report effect sizes, using Lipsey and Wilson’s (2001) calculation the difference between the two means (i.e., the intervention condition minus the control condition) divided by the pooled standard deviation of the two conditions. If not reported in the study, we calculated these results using the online calculator: http://gunston.gmu.edu/cebcp/EffectSizeCalculator/index.html. In the few cases in which outcomes were assessed using growth models, effect sizes were not calculated, as methods for doing so have not been agreed upon in the field (Feingold, 2009).
Cost–benefit information was based on calculations from the Washington State Institute of Public Policy (Aos et al., 2011; Aos, Lieb, Mayfield, Miller, & Pennucci, 2004). This agency has conducted reviews of the evidence and cost/benefits of many violence prevention interventions (but not all of those listed in Table 1) using a similar methodology across programs, resulting in the ability to compare costs and benefits across programs.
Our review resulted in the identification of 17 interventions aimed at youth aged 0–18 that demonstrated significant reductions in violence. Effective strategies targeted a variety of risk and protective/promotive factors using diverse mechanisms and included individual-, school-, family- and community-focused strategies. The programs also varied in the age/ages at which the intervention took place, with the set of interventions including families and youth from early childhood through late adolescence. Across programs, results were demonstrated in both the short and long term.
Early Childhood Education Interventions
As shown in Section I of Table 1, two early childhood programs were found to be effective in reducing violence. The Chicago Child–Parent Centers and Perry Preschool interventions both seek to increase opportunities for academic success among children from economically disadvantaged families. These programs aim to enhance parent–child bonding and child learning by helping parents become more actively involved in their children’s education; for example, by providing opportunities for parents to volunteer in the classroom and to regularly communicate with teachers. They also seek to increase academic competence via individualized and interactive methods of instruction. By improving family and school promotive/protective factors, these programs seek to minimize the potential for negative consequences among children living in low-income families and neighborhoods. The Chicago Child–Parent Center intervention showed significant reductions in juvenile arrests for violent crimes for those receiving up to 2 years of preschool services and tracked to age 18 (Reynolds, Temple, Robertson, & Mann, 2001). However, no significant reductions in violent arrests were found at follow-up studies conducted when participants were aged 24 (Reynolds et al., 2007) or 27 (Reynolds, Temple, Ou, Arteaga, & White, 2011). For the Perry Preschool Program, long-term and relatively large effects on arrests for violent crimes were seen through age 40 (Schweinhart, 2005). The Chicago Child–Parent Center has been demonstrated as cost beneficial, with a return rate of $4.82 for every dollar spent on programming (Lee, Aos, & Miller, 2008).
School-Based Interventions
A number of elementary, middle, and high school programs have demonstrated effects in reducing violence when implemented as discrete classroom curricula and as part of multiple-component interventions. Although these programs have varied content, most seek to improve individual and peer risk and protective/promotive factors by improving social and emotional competence. For example, programs seek to enhance students’ ability to recognize and regulate their emotions, effectively communicate with others, make good decisions, positively cope with stressful situations, and resolve conflicts using assertive rather than aggressive behavior. Programs typically use cognitive/behavioral methods of instruction that incorporate didactic instruction and opportunities for discussion and role play; for example, by asking students to practice with their classmates skills to avoid violent and provocative behavior (e.g., Life Skills Training). Some also seek to counteract negative peer influences by using older youth as role models who deliver antiviolence messages (e.g., Resolve It/Solve It).
All three teacher-led, school-based programs listed in Section II of Table 1 are universal programs intended to be implemented with all students in the targeted grades. All were evaluated using student self-reports of violence. At posttest, the Life Skills Training (Botvin, Griffin, & Nichols, 2006) and Resolve It/Solve It (Swaim & Kelly, 2008) middle school programs each showed small to medium effect sizes on violence. Although the Resolve It/Solve It curriculum is not gender specific, its effects were significant only for females. The Safe Dates program showed both short- and long-term reductions in dating violence for high school youth, with significant effects on the perpetration of both physical and sexual violence 3 years following the end of the intervention (Foshee et al., 2005). However, an earlier evaluation of this program based on 1-year follow-up data did not find significant effects on violence (Foshee et al., 2000). Life Skills Training was found to have significant cost–benefits, saving $42.13 for every dollar spent on the curriculum (Aos et al., 2011).
School- and Family-Focused Interventions
The five programs listed in Section III of Table 1 combined the delivery of a school-based curriculum for youth with family-focused interventions, and, in some cases, teacher-focused components. All five programs focused services on high-risk youth living in low-income neighborhoods and/or demonstrating externalizing problems at a young age. The school curricula incorporated into these strategies were similar to those described in Section II targeting individual and peer influences via teacher-led sessions. In addition, to enhance academic performance, more actively engage children, and increase their commitment to school and positive behaviors in the classroom, the Coping Power intervention, Seattle Social Development Project, and Multisite Violence Prevention Program sought to increase teachers’ use of interactive teaching skills, cooperative learning strategies, and proactive classroom management techniques using small group training of teachers with subsequent consultation. Parent training was provided in all interventions to improve affective (e.g., warmth and bonding to children) and instrumental (e.g., consistent and appropriate monitoring, supervision, and reinforcement of behaviors) parenting practices and to increase parental involvement in school.
Evaluations of these interventions were based on a variety of sources, including students, teachers, parents, and observers. Each program demonstrated small to medium reductions in physical violence. Three of the interventions showed long-term positive effects, at 2 years postintervention for the Preventive Treatment Program (Tremblay et al., 1992) and the Multisite Violence Prevention Program (2009), and at 6 years postintervention for the Seattle Social Development Project (Hawkins, Catalano, Kosterman, Abbott, & Hill, 1999). The Seattle Project was also found to be cost effective, saving $2.11 for each dollar spent (Aos et al., 2011). Two programs in this category of programming evidenced iatrogenic effects in addition to their positive outcomes. In the Preventive Treatment Program (Tremblay et al., 1992), although teacher reports indicated significant reductions in fighting, parent reports showed significant increases in fighting. In the evaluation of the Multisite Violence Prevention Program (2009), significant reductions in school-level rates of violence were found for schools assigned to the selective intervention condition versus the control condition at the 2-year follow-up, but when considering reports from females only, schools participating in the universal condition had higher rates of violence at posttest compared to schools in the control condition.
Parent Training/Family Therapy Interventions
The three programs listed in Section IV of Table 1 focused their interventions on the delivery of parent training and family therapy services, although each also provided direct instruction for youth. Targeted family risk and protective/promotive factors included increasing parents’ skills in setting guidelines for children’s behavior, monitoring children’s activities, and providing children with positive and negative consequences for behavior as well as the promotion of parent/child communication and bonding. Services for youth typically involved reducing individual and peer risk factors, such as peer pressure to engage in problem behaviors. Staying Connected with Your Teen is a universal program in which parents and youth participate in weekly group-based sessions led by trained prevention specialists. Multi-Systemic Therapy (MST) and Multidimensional Treatment Foster Care (MTFC) are indicated prevention programs for families of youth who have already come to the attention of the juvenile justice system. Trained clinicians provide intensive therapy in community and home settings to help reduce the likelihood of recidivism and increase the likelihood that families remain intact.
Staying Connected with Your Teen, which used an ethnically balanced research design involving White and African American families, reduced violence only among African American youth 2 years following program delivery (Haggerty, Skinner, MacKenzie, & Catalano, 2007). MTFC and MST each demonstrated robust effects in reducing self-reported violence and arrests for violent offenses, 2 years following the intervention for MTFC (Eddy, Bridges Whaley, & Chamberlain, 2004) and 4 (Henggeler, Clingempeel, Brondino, & Pickrel, 2002) to 14 years (Schaeffer & Borduin, 2005) following the delivery of MST services. Both of these indicated interventions were rated as cost effective by Aos and colleagues (2011), $4.07 for MST and $5.28 for MTFC per dollar invested in the programs.
Community-Based Interventions
Section V of Table 1 provides information on four effective community-based interventions. The Baltimore City Youth Bureaus and Big Brothers/Big Sisters program both provide mentoring services from adult volunteers to high-risk youth in order to increase social support and bonding to prosocial adults and organizations. The Baltimore program demonstrated reductions in self-reported violent behaviors at posttest (Hanlon, Bateman, Simon, O'Grady, & Carswell, 2002), and the Big Brothers/Big Sisters program showed decreases at the 1-year follow-up in youth reports of “hitting someone,” although a similar item asking about “being involved in a fight” did not demonstrate reductions (Tierney, Grossman, & Resch, 1995). The Big Brothers/Big Sisters program has been demonstrated as cost effective, with a savings of $3.28 for each dollar spent (Aos et al., 2004).
The Moving to Opportunities program provided monetary incentives for families to relocate from highly disadvantaged neighborhoods to safer and higher income areas. Two years following the distribution of housing vouchers, this intervention demonstrated reductions in arrests for violent offenses among male but not female participants; after 6 years, the effects were significant only for female participants (Kling, Ludwig, & Katz, 2005). This intervention also, unexpectedly, led to increased arrests for property offenses among males only at the 6-year follow-up. Communities That Care (CTC) is a universal prevention approach that relies on broad-based community coalitions to collect local epidemiologic data regarding levels of risk and protective factors experienced by youth, prioritize needs based on this assessment and fill gaps in service with effective individual-, school-, family- or community-focused interventions that target the particular needs of the community. This intervention also seeks to enhance community risk and protective factors by reducing norms favorable to problem behaviors, increasing communication and collaboration between adult residents, and enhancing youth attachment to the community. A randomized trial involving 24 communities in seven states demonstrated reductions in self-reported violent behaviors at the 1-year follow-up (Hawkins et al., 2012). A cost–benefit analysis using the Washington State Institute of Public Policy methodology found a return of $5.30 for every dollar invested (Kuklinski, Briney, Hawkins, & Catalano, 2012).
Discussion and Applications to Social Work
This review of the literature identified 17 interventions as effective in reducing youth-perpetrated violence. These programs represent great diversity in terms of their targeted age groups, strategies implemented to reduce or prevent violence, and length of programming. While such breadth makes it somewhat difficult to draw specific conclusions about the methods and components most likely to produce significant reductions in violent behavior, the results are consistent with developmental views regarding violent behavior and the tenets of prevention science. Developmental theories of delinquency and violence indicate that violence typically begins during adolescence, peaks during early adulthood, then wanes over the life course (Elliott, 1994; Farrington, 2003), which suggest that taking preventive actions prior to the emergence and escalation of this trajectory (e.g., prior to age 18) can have a significant payoff in reducing rates of violence. Furthermore, there is evidence that a variety of risk and protective factors, including individual, peer, family, school, and community experiences, influence the likelihood of youth violence (Herrenkohl et al., 2000; Lipsey & Derzon, 1998). Violence prevention may be best achieved by targeting one or more of these influences, ideally during the developmental period/periods in which they are most salient (Beardslee et al., 2011; Coie et al., 1993; U.S. Department of Health and Human Services, 2001).
Taken as a whole, the interventions listed in Table 1 indicate that strategies designed to reduce risk factors and enhance promotive and protective factors across all areas of children’s lives can significantly reduce violent behaviors. Further, more widespread adoption of these programs in communities could have a significant impact on violent behavior.
As seen in Table 1, the programs listed in Table 1 vary considerably in effect size and duration of effects. Some interventions have produced modest and immediate reductions in violence, while others have produced larger effects and demonstrated positive outcomes multiple years postintervention. Ideally, we would use these results to identify the specific aspects of programming responsible for producing the largest effects on violence. The breadth of programming and diversity in targeted populations, evaluation methodologies, and follow-up periods across the studies included in this review make this challenging, however. Although the field of prevention science has advanced significantly in recent years (Catalano et al., 2012; Fagan & Eisenberg, 2012), much more research is needed to expand the number of interventions being created, tested, and demonstrated effective. This includes replication of current programs to ensure their effectiveness across diverse populations; more nuanced evaluations that can identify the content, mechanisms, and risk and promotive/protective factors responsible for producing change; and evaluations that compare interventions in order to pinpoint elements of programming most likely to lead to desired changes.
Although we recommend some caution in drawing conclusions from the results of this review, the findings do suggest that some commonly asserted beliefs of researchers and practitioners may need to be reconsidered. Prevention science tends to promote multicomponent interventions, such as CTC, as most effective because they can simultaneously address multiple causes of problem behaviors (Coie et al., 1993; Dahlberg & Potter, 2001), but some findings suggest that significant and meaningfully large decreases in violence can also be evidenced when focusing on a narrow set of risk and protective/promotive factors in just one domain (e.g., the family environment). Likewise, programs that are longer in duration are often touted as best able to produce significant and long-lasting effects on problem behaviors (Nation et al., 2003). However, this review indicated that some relatively short interventions (e.g., school curricula and parent training programs lasting only a few months) can reduce violence among a universal audience of youth and families. In addition, short-term but intensive therapeutic services (e.g., MST and MTFC) have been shown to reduce violence among adolescents already involved with the juvenile justice and child welfare systems, with effects lasting for a number of years postintervention.
Whether or not it is better to focus services on high-risk individuals or the more general population of youth has been debated (Coie et al., 1993). Some contend that benefits to society can best be achieved through universal programs, because even if the amount of change is small for each individual, the number of individuals reached will be greater than if only higher risk youth (who comprise a smaller segment of the population) are targeted (Rose, 1985). Others contend that serving the most vulnerable groups of individuals (e.g., those with low socioeconomic status) provides the best way to reduce health disparities in outcomes across social groups (Frohlich & Potvin, 2008). This review demonstrates that universal services can be effective in reducing violent behaviors, as can interventions targeting selective and indicated youth. The best strategy, then, may be for communities to assess their own levels of violence, risk, and protection and, guided by these profiles, offer a range of services targeting youth of different ages and levels of risk and protection.
Although a variety of prevention programs were found to be effective in this comprehensive review, ideally, we would like to recommend far more than 17 efficacious interventions. Like many literature reviews, we limited our search to high-quality evaluations utilizing quasi-experimental and experimental designs because these are considered to have the greatest potential to identify valid effects on participants; however, doing so limited the number of interventions that could be demonstrated effective. Unlike some prior reviews, we utilized a somewhat strict definition of “violence”: attempted and/or actual perpetration of interpersonal, physical, or sexual aggression. We recognize that this definition also limited the number of interventions that could be identified as effective and possibly decreased the likelihood of identifying programs for very young children, given that evaluations of early childhood interventions without long-term follow-up usually measure other types of problem behaviors (e.g., oppositional and defiant behaviors, conduct disorder, etc.). Nonetheless, the results do include some interventions for young children. Moreover, taken as a whole, the identified interventions have all demonstrated the ability to reduce behaviors of great concern to the public: serious, interpersonal aggressive actions. We would also note that our search criteria did not require, as some reviews do, that evaluations involve follow-up periods; requiring a follow-up period of 1 year, a requirement of some reviews, would have reduced the number of effective programs from 17 to 12.
It is also important to note that this review focused on identifying discrete prevention programs, not types of programming or prevention practices or policies. Some contend that specific, “named” programs may be too costly and/or difficult for communities to implement, given that they tend to be highly structured and often involve multiple components; and that it may be more feasible to foster implementation of more general strategies, practices, and/or policies (Greenwood & Welsh, 2012; Lipsey, 2009; Weitzman, Nelson, Lee, & Wechsler, 2004). A few well-conducted systematic reviews and/or meta-analyses have identified effective violence prevention program types and/or policies (Hahn et al., 2007; Lipsey, 2009; Mercy et al., 2008; Wilson & Lipsey, 2007). However, the implementation of general strategies can be fraught with difficulties. Without specific manuals or training, implementers may not ensure that all important elements are delivered and they may not have the skills needed to effectively deliver prevention messages. Further, few strategies have been evaluated for financial costs and benefits. It would be helpful to compare the effects and costs/benefits of discrete programs versus strategies in experimental trials in order to provide community agencies and practitioners with the best possible information needed to make informed choices when considering new programming.
Many challenges will undoubtedly be faced when seeking to increase community use of interventions shown to be effective in scientific studies. While a full discussion of these issues is beyond the scope of this review, it is clear that much community practice is not based on scientific evidence regarding what works to prevent violence and other problem behaviors (Glasgow et al., 2003; Ringwalt et al., 2009; Saul et al., 2008; Woolf, 2008). Important first steps in expanding dissemination, then, are to develop strategies to increase local knowledge regarding effective violence prevention interventions, build capacity and local infrastructure to support such interventions, including the development of a skilled and motivated workforce, and ensure adequate support to ensure that effective practices can be sustained over time (Meade & Steiner, 2010; Saul et al., 2008; Task Force on Community Prevention Services, 2007). Local agencies and practitioners may also benefit from focused technical assistance to help them research, plan for, and carefully monitor implementation practices. Prior to implementation, potential adopters will want to consider not only the effectiveness of new programs but also the degree to which they are appropriate for the particular needs, resources, and philosophical orientations of implementing agencies. Given research that implementation fidelity is associated with more positive participant outcomes across diverse types of interventions (Durlak & DuPre, 2008; Fixsen, Blase, Naoom, & Wallace, 2009; Henggeler, Melton, Brondino, Scherer, & Hanley, 1997), it is important that agencies and practitioners buy into the importance of monitoring service delivery, ensure the routine collection of implementation data, and utilize these data to improve practices as needed, regardless of the implementation strategy they select.
While these challenges are important but perhaps daunting to overcome, it is encouraging that policy makers, scientists, and local administrators and practitioners are increasingly discussing ways to foster the dissemination of effective practices (Beardslee et al., 2011; Elliott & Mihalic, 2004; Saul et al., 2008). The increasing promotion of science-based practices among federal government agencies—such as the focus on prevention, violence-free communities, and increased use of evidence-based services promoted in 2011 National Prevention Strategy—should help increase funding and support for the dissemination of effective preventive interventions and, in turn, the potential for widespread reductions in youth violence across the United States.
Footnotes
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. Catalano is a board member of the Channing Bete Company, publisher of Staying Connected with Your Teen (a program included in this review), Guiding Good Choices, and Supporting School Success (two of the family components of the Seattle Social Development Project that is included in this review).
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.