Introduction

Gender differences in the prevalence rates for violent victimization and violent offending reveals boys are at an elevated risk. Official statistics show that male youth are the majority (70%) of the victims of violence and commit the majority (84%) of violent crimes (Finkelhor and Ormrod 2000; Snyder 2000). Over four decades of research using mostly male samples has linked childhood trauma to adolescent and adult aggression (Falshaw et al. 1996). Overall, these findings suggest that youth who experience a range of mild to severe types of traumatic life experiences in childhood or adolescence are at an increased risk of engaging in violence. Research in this area has largely focused on the connection between exposure to violence (i.e., being a victim or witness to violence) and violent offending. These studies have shown that between 32% and 79% of violent offenders have had histories of violent victimization, such as physical or sexual abuse (Lemmon 1999; Smith and Thornberry 1995). Additionally, less severe types of trauma also have been linked to violence among male youth, particularly stressful life events (SLE). Research has shown that youth who experienced SLE, such as parents divorcing, problems in school, losing a loved one, and living in a violent neighborhood are also at an elevated risk of engaging in delinquency and violence (Eitle and Turner 2002).

Gender specific reactions to trauma also place male youth at an elevated risk of violence (Mazerolle 1998). Research shows that adolescent males may respond to stress with maladaptive emotions and behaviors that include negative affect, often in the form of anger and rage, and physical aggression (Lui and Kaplan 1999). Thus, when confronted with stressful situations or coping with unresolved traumatic experiences, male youth have a higher tendency to express anger and act out aggressively (Hoffman and Su 1997).

The quality and quantity of peer relationships, especially for boys, may be a significant risk factor connecting trauma to violent delinquency. Problematic interactions with prosocial peers (Howe and Parke 2001) and affiliation with delinquent peers (Rodgers-Farmer 2000) have been found to be positively associated with antisocial behavior in youth. Male youth who affiliate with other male youth involved in crime also are at an increased risk for violence. Official statistics reveal that males are initiated into and commit the majority of their crimes in male peer groups (Snyder 2000). Moreover, unresolved emotions, such as anger, coupled with increased exposure to delinquent peers, can further exacerbate the risk that a youth who has experienced trauma will in turn react with violence (Agnew and Brezina 1997).

While the literature has found evidence that youth who experience trauma may respond with an array of adverse emotional, social, and behavioral responses, the socio-emotional pathways that link trauma to delinquency and violence for some youth and not others are not yet fully understood. For example, negative affect states (e.g., anger) and delinquent peer exposure have been shown to be both consequences of trauma and risk factors for violent offending (Aseltine et al. 2000; Benda and Corwyn 2002). However, the intricacies of how male youth respond to trauma, through unresolved emotional states and associations with delinquent peers, have yet to be fully explored.

Furthermore, this trauma-violence link may also be influenced by race/ethnicity among males. Official statistics reveal that a disproportionate number of minority youth are involved in the juvenile justice system. According to the Unified Crime Report (FBI 1999), 41% of juvenile arrests for violent crime involved African American youth compared to 57% involving Caucasian youth. However, this discrepancy may be indicative of other social processes not captured in official data, such as the influence of lower socioeconomic status, racial discrimination, geographic location, and/or level of involvement with peer groups or gangs (Youth Law Center 2000).

Therefore, the purpose of this study is to examine the influence of negative affect (i.e., anger and depression) and delinquent peer exposure on the relationship between trauma exposure and violence among male youth. Based on trauma theory and general strain theory, we examined how maladaptive emotions (i.e., negative affect or affect dysregulation) and maladaptive peer relations, such as delinquent peer exposure, in response to trauma place youth at increased risk for delinquency and violence (Agnew 2001). The following hypotheses were tested: (1) The relationship between trauma (i.e., exposure to violence and SLE) and violent offending will be moderated by delinquent peer exposure, positively; (2) The relationship between negative affect (anger and depression) and violent offending will be moderated by delinquent peer exposure, positively. Using a nationally representative sample of 2,065 adolescent males, it was shown that trauma, anger, and delinquent peer exposure exerted a direct effect on violence. Delinquent peer exposure also exerted a significant interaction effect on the relationship between anger and violent offending.

Research that uncovers the mediating and moderating pathways that link trauma to violent offending has important implications for practice, research, and policy involving at-risk youth. Understanding the social psychological pathways between trauma and violence will assist us in identifying those factors that ameliorate or exacerbate risk or resilience among youth. This knowledge can then be used to develop multi-level prevention and intervention efforts geared towards improving the health and wellbeing of at-risk youth and their families and communities. An overview of the methods to test the study hypotheses and their implications for social work practice and policy follows.

Methods

Research Design

The hypotheses were tested using the nationally representative male sample (N = 2, 018) from the 1995 Survey of Adolescents (NSA) (Kilpatrick and Saunders 1995). This sample size increased to 2,065 when weighted based on the general population of adolescents from the 1988 United States Census Bureau statistics for geographic location, age, gender, and race/ethnicity. A stratified random sample design was used and the male population consisted of two subsamples, a national probability household sample (n = 1, 599) and a probability oversample (n = 419) from urban areas in the United States. The age range of the youth was between 12 and 17 with a mean age of 14.47 (SD = .17). The race/ethnicity of the sample was mostly Caucasian (70%), followed by African-American (15%), and Hispanic (8%) youth.

Procedures

Self-report information was gathered in 1995 from adolescents and their caretakers during sixty minute structured telephone interviews. The interview was administered using computer assisted telephone interviewing (CATI), which is considered more advanced than standard interviewing for its ability to handle complex skip patterns and question ordering. In order to conduct the interviews, permission was first obtained first from a parent or guardian to interview the adolescent in the home. The response rate for the telephone interviews was high. Most (90%) of parents/caretakers participated and 80% also gave permission for an adolescent in their home to be interviewed. There was a 75% response rate of adolescents from eligible households of which 95% had parental permission and 83.2% had an accompanying parent interview. The adolescents interviewed discussed their prior experiences of direct and indirect violence, past year SLE, delinquent activity, mood states, and peer relationships. The adolescents’ self-report information was used in the analysis, except for the information on family sociodemographic variables, which was provided by the parents/caretakers.

Measures

Predictor Variables

Exposure to violence. The NSA’s 20-item multivictimization scale was used to measure the cumulative impact of lifetime exposure to violence for youth aged 12–17. Respondents were asked if they had experienced exposure to violence: physical abuse, physical assault, sexual assault, and witnessing violence. Physical abuse was defined as an act in which a youth received physical punishment from a caretaker, which consisted of spanking (resulting in bad marks, bruises, cuts, or welts), burning, cutting, or being tied up. Physical assault was defined as the youth being hit, attacked, or beat up either in the neighborhood, school, or home. Sexual assault included sexual acts in which a youth was forced to give or receive fondling, oral sex, and/or anal penetration. Witnessed violence was defined as having observed someone being shot, stabbed, sexually assaulted, mugged, robbed, or threatened with a weapon.

Stressful life events. The NSA SLE scale was used to measure SLE during the previous year. The SLE scale consisted of the following 14 items: parents separated or divorced, mother/father lost job, new stepmother or stepfather, death of a family member or close friend, loss of a close friend, move to a new home, change to a new school, and serious illness or injury of a family member, close friend, or self, failed a class, got left back a grade, and suspended from school. Each event that occurred during the past year equaled one point with a total of 14 possible points.

Delinquent peer exposure. Delinquent peer exposure was represented by the 13-item NSA Delinquent Peer Exposure Module. It was used to assess the proportion of friends who were reported as engaging in past year delinquency. These acts included: acts of theft or burglary, alcohol or drug use, alcohol or drug sales, and/or physical or sexual assault. Youth were first asked if they ever had friends who engaged in any of these activities. If the answer was yes, the youth were asked what proportion of his friends engaged in that type of activity in the past year. A 5-item Likert scale determined the proportion of delinquent friends who committed any of these delinquent acts, with responses ranging from 4 = all of them to 0 = none of them. Cronbach’s coefficient alpha was .88 for this sample.

Negative affect. Negative affect was represented by anger and depression. Anger was measured by the following two items: “Did you ever feel that little things bothered you a lot or could make you angry” and “When was the most recent time you felt that way?” It was coded as follows: 0 = never angry; 1 = angry more than 6 months ago; 2 = angry within the past 6 months; and 3 = angry within the past month. The alpha coefficient for this measure was .79. Depression was measured by the 12-item NSA Depression Module. This scale is based on the DSM-IV criteria for depression and includes items such as depressed most of the day for at least 2 weeks, felt worthless, had trouble/difficulty concentrating, and/or had thoughts of hurting oneself. Each affirmative response was equal to one point with 1 = yes and 0 = no with a total score of 12.

Outcome Variable: Violent Delinquency

A subscale of NSA’s Juvenile Delinquency Module was used as a measure of violent delinquency for this study. A self-reporting delinquency measure, the subscale has four items: being involved in gang fights, using force or strong armed methods to get money or things from people, having or attempting to have sexual relations with someone against his or her will, and attacking someone with the idea of seriously hurting or killing them. It was coded as a binary variable (1 = past year violent delinquency, 0 = no past year violent delinquency).

Moderating Variable: Delinquent Peer Exposure

Control Variables

The variables: age, socioeconomic status (total family income and parental/head of household education), race/ethnicity, family structure, geographic location, and social support were included as covariates in the model because of their independent impact on trauma and delinquency (Maschi 2006a, b). Age was measured by the adolescent’s current age (between ages 12 and 17) at the time of the interview. Race/ethnicity was dichotomized into two groups based on perceived societal status and associated minority stress level (1 = minority status and 0 = majority status). The ethnic majority group consisted of white/non-Hispanic and the ethnic minority group consisted of African-American, Hispanic, Native-American, and Asian youth. Social class/socioeconomic status was measured by total family income and parental education. Income was dichotomized categorized as above and below the estimated 1995 poverty level (1 = $20,000 and above and 0 = under $20,000). Parental education was measured by whether the person designated as the head of household had obtained a high school diploma or GED (1 = yes and 0 = no). Family structure was defined as either a married couple (two-parent biological or adoptive family or step-family) or a non-intact family which consisted of parents/caretakers who were separated, living as an unmarried couple, divorced, widowed, or single/never married (1 = intact and 0 = non-intact). Geographic location consisted of urban (1 = city and city suburb) or rural (0 = small town or rural) location of the family residences. Social support was defined as the presence or absence of ‘anyone in childhood to count on or depend on to be there when you needed them throughout one’s whole childhood’ (yes = 1; no = 0).

Results

Descriptive Statistics

The majority of the youth, especially those who reported engaging in past year violence, reported histories of trauma exposure, feeling angry and depressed, and involvement with delinquent peers. Of the total sample, 1 out of every 10 adolescent males (10.6 percent, n = 219) reported that they engaged in a violent offense during the past year. While the majority of the youth 77.8% (n = 1,607) reported lifetime exposure to violence and past year SLE (88.3%, n = 1,823), it was much more common among the violent subsample. The violent group reported much higher levels of violence exposure (98.5%, n = 216) and SLE (98.5%, n = 216) compared to the nonviolent subsample (75.0%, n = 1,391; 87.1%; n = 1,607) and these differences were statistically significant (X 2 = 61.32, df = 1, P < .01; X 2 = 25.13, df = 1, P < .01).

Significant differences also were found between violent and nonviolent groups for anger (X 2 = 125.89, df = 1), depression (X 2 = 22.72, df = 1, P < .01), and delinquent peer exposure (X 2 = 140.89, df = 1, P < .01). Of the total number of youth who reported negative affect states of anger (14%, n = 290) and depression (30.6%, n = 244), many more of the youth who reported engaging in violence than those youth who did not reported being angry (61%; n = 134; 11.1, n = 204) and depressed (30.6%; n = 67; n = 204). Additionally, of the 55% (n = 1,136) of the sample of male youth who reported past year associations with delinquent peers, the majority of them also reported engaging in violence (92.6%, n = 203) compared to those youth who reported they did not (50.%, n = 932). Overall, the group reporting violence exhibited significantly higher levels of maladaptive socio-emotional reactions.

Moderation Analyses

Hierarchical Logistic Regression Analyses

The next analysis step entailed running hierarchical logistic regression analysis to examine the role of delinquent peer exposure as a moderator of trauma and anger on violent offending (See Table 1). The analysis methods used were consistent with established procedures for testing moderation (Cohen et al. 2003). Interaction terms were constructed for trauma and delinquent peer exposure and delinquent peer exposure and anger. First, delinquent peer exposure was dichotomized at its’ median value. This method is the recommended approach for testing interaction effects (Jaccard et al. 1990). Next, hierarchical logistic regression analyses were conducted using violent offending as the outcome variable. In the next step, the variables representing trauma and anger were centered on their mean scores. According to Jaccard et al. (1990), centering the variables on the mean reduces multicollinearity between the variables and their product terms. Interaction terms were then constructed for (1) exposure to violence X delinquent peer exposure, (2) SLE X delinquent peer exposure, (3) anger X delinquent peer exposure, and (4) depression X delinquent peer exposure.

Table 1 Summary of hierarchical linear regression examining the interaction effects of delinquent peer on anger and violent offending
Full size table

Variables were entered into the equation in three steps. In step one, the control variables (i.e., age, income, parental education, race/ethnicity, family structure, geographic location, and social support) were entered into the model simultaneously. In step two, the main effects for exposure to violence, SLE, anger, depression, and delinquent peer exposure were entered into the equation as a block. In step three, a test of the moderating effect of delinquent peer exposure was conducted. Each of the two-way interaction terms (delinquent peer exposure X exposure to violence, delinquent peer exposure X SLE, delinquent peer exposure X anger, and delinquent peer exposure X depression) were entered into the model independently. The presence of significant interaction terms signified the presence of hypothesized moderator effects between variables. Next, significant interactions were calculated by using the standard regression equation, DV = b0 + b1 (VAR1) + b2 (VAR2) + b3 (VAR1*VAR2), to plot the slopes of the significant interaction effects (Aiken and West 1991).

Table 1 presents the results of the moderation analysis for the impact of delinquent peer exposure on the relationship between trauma and violent delinquency and negative and violent delinquency. After entering the control variables in the model (Table 1, Model 1), the entry of the main effects variables significantly improved the fit of the model (X 2 = 533.98, P < .01) (Table 1, Model 2). Among the central variables of interest, positive and significant main effects were found for exposure to violence (b = 0.35, Expb = 1.41, P < .01), SLE (b = 0.14, Expb = 1.25, P < .01), anger (b = 0.24, Expb = 1.28, P < .01) and delinquent peer exposure (b = 0.12, Expb = 1.13, P < .01).

For brevity’s sake, only significant interaction terms were included in Table 1, Model 3. As shown in Model 3, significant main effects were found for exposure to violence (b = 0.35, Expb = 1.42, P < .01), SLE (b = 0.14, Expb = 1.15, P < .01), anger (b = 0.74, Expb = 2.09, P < .01), and delinquent peers (b = 0.12, Expb = 1.13, P < .01). Additionally, control variables that remained significant in the final model included race/ethnicity (b = 0.81, Expb = 2.24, P < .01), family structure (b = −0.50, Expb = 0.60, P < .05) and social support (b = −0.60, Expb = 0.55, P < .05). One significant interaction effect was shown between delinquent peer exposure and anger that impacted violent behavior among the youth. In Model 3, the entry of delinquent peer exposure X anger, showed a chi-square change of 7.833 (p < .01), and significantly improved the overall fit of the model. The interaction effect for delinquent peer exposure was negative and significant (b = −0.57, Expb = 0.57, P < .01). Figure 1 illustrates how delinquent peer exposure exerted a moderating effect on the relationship between anger and the predicted log odds for violent offending. Although the slope for low and high delinquent peer exposure had a moderate increase in their slope, the youth who reported low peer interaction were at slightly greater odds of violent offending.

Fig. 1
figure 1

Plot for the interaction effect of delinquent peer exposure x anger on the predicted log odds for violent offending (Table 1, Model 3)

Full size image

In summary, moderation analysis results showed direct effects for male youths’ exposure to violence, SLE, anger, and delinquent peer exposure on violent offending. A significant interaction effect for delinquent peer exposure and anger was found to increase the logs odds that male youth reported violent offending. A discussion of these findings follows.

Discussion

This purpose of this study was to explore how the socio-emotional factors of negative affect and delinquent peer exposure influence the link between trauma and violent behavior among male youth. Results of chi-square analyses indicated that youth who reported engaging in past year violent offending also reported a higher rate of exposure to violence, SLE, anger, depression, and delinquent peer exposure than those youth who did not report engaging in past year violent offending. Results of a hierarchical linear regression revealed that youth who reported histories of exposure to violence, SLE, anger, and delinquent peer exposure, had a greater likelihood of reporting violence compared to youth who did not report these experiences. Yet, negative affect was only partially related to violence. That is, although anger was found to have a significant and direct effect on juvenile delinquency, depression did not.

These findings contribute to the extant literature in the impact of socio-emotional factors on juvenile delinquency. We found that male youth who spend a disproportionate amount of time with delinquent friends are at increased odds of violent offending. On the interpersonal or relational level, an association with delinquent peers was found to increase the odds that youth would engage in delinquency. These results are consistent with other empirical studies that have found support for a direct effect of delinquent peer exposure on delinquency (Aseltine et al. 2000; Mazerolle and Maahs 2000; Paternoster and Mazerolle 1994).

This study builds upon the extant literature by exploring how social relationships with delinquent peers may moderate the link between negative affect and violent offending among males. Support was found for the moderating influence of delinquent peer exposure on the relationship between anger and violent offending. Yet, the results were not exactly as expected. Results of the moderation analyses revealed that youth who reported being angry and low levels of exposure to delinquent peers were at increased odds of engaging in violence compared to youth with low levels of anger who reported high levels of delinquent peer exposure. As illustrated in Fig. 1, high delinquent peer exposure has an influence, although weak, on the relationship between anger and violent delinquency. In contrast, when low delinquent peer exposure is present, anger exerted an even stronger influence on violent delinquency. This finding is consistent with Aseltine et al. (2000) in which they found support for a buffering impact of delinquent peers. Perhaps the support of the peer group, despite involvement in minor to severe act of delinquency, has some positive impact on male youth. These findings emphasize how the role of the delinquent peer groups may act as a source of support that decreases the likelihood that youth will engage in violence. However, when anger increases despite the level of peer involvement, violence is heightened.

Additionally, the significant influence of the control variables, race/ethnicity, family structure, and social support, on violent offending is noteworthy. Perhaps additional forms of stress experienced by youth related to minority status, living in non-intact family, and having low levels of social support may impact youths’ socioemotional development. As suggest earlier, the significance of racial/ethnic influences may be indicative of other social processes for minority youth, such as racial discrimination, living in a violent neighborhood, and/or level of involvement with delinquent peer groups (Youth Law Center 2000). The significant buffering impact that social support had on violent offending further support the notion that having someone to count on, including adults and peers, may significantly reduce the likelihood that youth will choose delinquency as a coping mechanism.

The findings related to anger and delinquent peer exposure on violence among youth has important implications for prevention and intervention efforts with at-risk youth. Targeting prevention and intervention efforts in schools, social service, and juvenile justice settings would assist in enhancing the socioemotional functioning of youth, especially for those youth impacted by trauma. Social workers and other human service workers who interact with youth should assess for the objective and subjective impact of minor to severe SLE that impact youth. Currently, being a victim of child maltreatment (i.e., physical and sexual abuse and neglect) remains the sole type of trauma that is illegal and mandated to report. However, even reporting child maltreatment does not guarantee a child and his or her family will received adequate services to address their ongoing needs. Other types of trauma, such as moving to a new home, death of a significant other, may go unidentified and untreated or even if identified, left untreated. Not attending to these minor to severe stressors may results in an increase in maladaptive emotional, social, and behavioral consequences for youth.

Schools provide the most logical setting for the provision of such prevention and intervention services as school based personnel have easy and direct access to youth. However, school intervention services staff often are impeded in their efforts to work with students due to the current emphasis to view academic success as the only goal for students. Throughout the United States, the effect of the Federal No Child Left Behind (NCLB) legislation negatively impacts school districts’ abilities to focus on the socioemotional concerns of students due to the emphasis on the achievement of passing scores on mandated testing. Ironically as shown through the results of this study, the focus on academic achievement to the neglect of the recognition of the needs of the whole child may have the effect of lowering scores in some districts. Male youths whose socioemotional needs are overlooked or neglected may exhibit less attachment to school as it requires that are able to regulate emotion and exhibit impulse control.

Currently in states, such as New Jersey, there is discussion of raising the high school graduation requirements so as to ensure a more work ready population. Again the emphasis is on strictly academic areas. Concerns regarding school funding formulas coupled with the demands of NCLB, increase the likelihood of the elimination of services provided by student personnel staff. It is the daily focus of just such staff (school social workers, student assistance counselors, guidance counselors and school nurses) on the provision of prevention services and on the identification of and intervention with students exhibiting at-risk behaviors.

Student personnel staff design and implement the prevention and intervention programming within the public schools. The development or improvement of resiliency theory based programs, such as Emotional Literacy and Mentoring Programs and Asset Building In-School Support Groups, may provide mechanisms, by which at-risk male youths may learn to manage anger and stress, develop positive social support and make meaningful connections to caring adult figures. Groups such as those described above emphasize strengths exhibited by the youth rather than utilizing pathology based approach. School districts with student assistance programs offer such services. Programs provided by schools should also be evaluated for their effectiveness, particularly since school populations vary across national and international regions. Some evidence-based practices are outlined below.

Violence prevention or emotional literacy programs have been shown to be effective in reducing adverse emotions and behaviors among youth. Programs, such as Second Step: A Violence Prevention Curriculum, is an emotional literacy program that was developed to increase the social and emotional skills of youth and includes modules on empathy, anger management, and emotional learning. Research on this program has shown that youth participants increased their social and emotional skills and decreased their use of physical and verbal aggression and disruptive behavior (McMahon et al. 2000).

Other prevention and intervention measures that have demonstrated effectiveness include mentoring programs, especially in schools and neighborhoods with high levels of youth gang affiliation. Mentoring programs provide at-risk youth the opportunity to bond with prosocial adults or peers. Evidence suggests that at risk youth who participate in mentoring programs compared to youth who do not are less likely to engage in antisocial activities, such as substance use and violence as well as increase their academic performance. Mentoring programs, such as the Juvenile Mentoring Program (JUMP) and Big Brothers/Big Sisters (BB/SS), are examples of evidence-based mentoring programs (Grossman and Garry 1997).

Other helpful intervention strategies include multisystem approaches. For example, Multisystemic Therapy (MST) has been shown to be effective in reducing antisocial behavior (i.e., disobedience, running away, drug use, arson, vandalism, theft, and violence against persons) in at-risk youth. Individual treatment may include cognitive behavioral treatment geared towards changing relations with peers and building academic competence in school. It may include interventions with the family, including marital interventions and family linkages with community support. Community interventions may include the development of needed community based health care services (Swenson et al. 2005).

Additional prevention and intervention strategies that can be more widely adopted in schools and communities include the addition of the use of expressive arts, music, and sports in order to harness positive youth development (Delgado 2000). According to Delgado (2000), the use of the arts and sports for micro-macro level interventions serve to engage youth in the process and stress their assets as well as serve to empower youth and enhance their skills, knowledge, and attitudes, and transform environments.

Limitations of the research design also warrant discussion. This study was a secondary data analysis using self report methods to gather information from adolescents and their parents/guardians. Although self-report studies often are more reliable than official reports for gathering undocumented report of trauma and delinquency, they are limited by the use of retrospective reports of past events and thus reliability and validity may be comprised (Brown 1999). Adolescent male self-reports via the telephone, especially over stigmatizing situations, such as exposure to trauma and delinquency, is questionable, (Huizinga and Elliott 1986). The use of secondary data is a limitation because the only measures available to measure a variable of interest may not be the most valid one. The temporal sequencing of the variables limits the ability to establish causal links. The findings of a causal pathway between trauma and delinquency were based on a cross-sectional design; therefore, any interpretations of the data are tentative. Additionally, the study did not fully explore the impact of race/ethnicity beyond minority versus majority status or gender (i.e., female youth). However, despite these limitations, the findings can serve as a building block for future research in this area.

In conclusion, this investigation found a link between differing types of trauma and violent delinquent behavior among male youth. Directions for future research include intervention research that target identifying and coping with anger and harnessing the power of the peer group for the purposes of positive outcomes for youth should be further explored. The influence of race/ethnicity, culture, and gender should be further explored in order to develop and improve gender and culture specific prevention and intervention efforts.

Future research also should include longitudinal studies that a broad measure of the types of trauma and SLE linked to delinquency and violence. Additionally, more comprehensive measures of anger and the nature of youth social networks of delinquent and non delinquent peers would provide useful information for the development of interventions targeting these risk factors. Gathering information related to the state-trait characteristics of youth’s own anger as well as their peers, will help to discern how much of youth violence may be a part of direct (i.e., one’s own anger) and/or ‘vicarious anger’(i.e., collective peer group anger) that may influence violent behavior. Future research in this area could greatly contribute to uncovering the factors to target for prevention and intervention efforts that will harness the positively directed power of anger and peer affiliations for positive youth development.