Elsevier

Aggression and Violent Behavior

Volume 7, Issue 5, September–October 2002, Pages 423-451
Aggression and Violent Behavior

An ecological framework for understanding risk for exposure to community violence and the effects of exposure on children and adolescents

https://doi.org/10.1016/S1359-1789(01)00078-7Get rights and content

Abstract

This review covers the past decade's research on the risk for exposure to community violence and the effects of exposure on children's and adolescents' functioning. The studies are incorporated into a developmental-ecological framework that takes into account five domains of context—community and neighborhood, family and household, relationships with parents and caregivers, relationships with peers, and personal characteristics—for the purpose of identifying the risks for exposure and its effects on outcome and for suggesting the processes involved. Evidence from the literature is consistent with our proposed ecological model showing that variables in each of the five domains have both direct and indirect effects on risk for exposure and on its effects on internalizing and externalizing problems and academic functioning. Implications of adopting an ecological model for future risk research and for prevention and intervention are considered.

Introduction

Within the last 10 years, research on community violence, particularly with respect to children and youth, has grown exponentially. Now, after a decade of research on the phenomenon and studies of risk for exposure and studies of its effects, researchers are only beginning to summarize and conceptualize the field. A number of themes have emerged.

The first is that exposure to violence has pervasive and profound effects on development. Using the abuse and domestic violence literature to suggest implications for the effects of community violence, Osofsky's (1995) review of the effects of exposure on very young children clearly points to poor attachment, negative emotions, PTSD, and aggressive behavior. Kuther (1999), reviewing the effects of witnessing community violence (so-called covictimization) on emotional, cognitive, and behavioral domains, concludes that when these effects and their interactions are viewed in terms of a developmental-contextual approach, exposure has serious implications for important superordinate developmental concepts such as children's understanding of the social world, moral development, and health-related beliefs. Overstreet (2000), as well, concludes that the stress, anxiety, and fear generated by exposure to violence interfere with significant normal developmental tasks such as the development of trust, sense of safety, emotional regulation, explorations of the environment, and ability to form social relationships. Margolin and Gordis (2000), focusing more on traumatic events in the community and on the effects of abuse and domestic violence than on the many types of chronic exposure often addressed in studies of inner-city youth, comment on the need for a developmental perspective on the effects of exposure and for more research devoted to uncovering the processes and mechanisms leading to different outcomes. They point out that this will require examination of such developmental domains as children's school functioning, social competence, ability to regulate emotions, and physiological reactivity—all areas not ordinarily addressed in clinical assessments of the effects of trauma.

A second theme is the heuristic value of an ecological perspective to explain the effects of exposure. Cicchetti and Lynch Cicchetti & Lynch, 1993, Lynch & Cicchetti, 1998 were among the first to propose an ecological-transactional model to explain developmental findings in the combined literatures of child maltreatment, domestic violence, and exposure to community violence. Later, based explicitly on a classification of children's exposure in terms of the ecological levels of microsystem (intrafamilial violence), mesosystem (community violence), and exosystem (media and societal violence), Jonson-Reid (1998) offered a framework for understanding the effects of all forms of child violence exposure on youth violence. She evaluates the available studies for the extent to which they confirm the proposed associations among the risk and resilience factors operating between childhood exposure to violence and youth violence.

Influenced by the ecological approach, some researchers reviewing intervention and prevention efforts make a strong case for a shift from individual approaches to the inclusion of a family component (Reese, Vera, Simon, & Ikeda, 2000), especially parental support (Duncan, 1996). Reese et al. (2000), going even further, recommend inclusion of multisystemic therapeutic interventions in order to fully utilize the risk and protective factors that may influence youth violence while acknowledging the serious barriers to familial and parental inclusion that must be carefully dealt with if such interventions are to prove efficacious.

Relevant to the issue of barriers to successful intervention is the more general point, discussed by Lorion and Saltzman (1993), of the ethical implications of conducting social science research in violent inner-city neighborhoods. They and others call for involvement of subjects in these communities in the design and prosecution of the work, and they draw attention to the necessity for facilitating cooperatively designed interventions even as the research is proceeding.

The last theme emerging from earlier reviews is that violence is pervasive in American society and that its effects are not only widespread, affecting children in all types of communities, but also likely to give rise to a wide spectrum of dysfunctional effects from internalizing problems to aggressive and violent behavior. Researchers and clinicians working in the field warn that violent behavior conditioned as a coping response in the effort to be safe and cognitive strategies employed to adapt to or normalize the use of aggressive behavior only make the problem more persistent. Garbarino, Dubrow, Kostelny, and Pardo (1992) have long called attention to the toxicity of many environments in which we raise children, especially in the inner-city, and the feelings of futility engendered in children exposed to violence. Osofsky (1995), reflecting the feelings of many researchers and clinicians who have worked with children exposed to violence, has asked for a nationwide campaign to change attitudes toward violence and decrease the public's tolerance of violent behavior.

The present review covers research during the past decade. The general aim is to organize the available work in terms of its relevance for an overall developmental-ecological model intended to help identify likely paths between community context and the risk for children's and youths' exposure and paths leading from exposure to three broad dysfunctional behavioral outcomes: internalizing problems, externalizing problems, and poor academic functioning. Our intent is twofold: (1) to suggest some testable models for risk research that incorporate variables from various ecological levels and (2) to provide guidance for the selection and timing of intervention and prevention programs with respect to the outcomes examined.

The review is laid out as follows: Section 1.1.1 deals with risk for exposure and presents the literature detailing the direct effects of risk variables in five ecological domains ordered from most distal to most proximal. These are neighborhood and community contextual factors, family contextual factors, parenting and quality of relationship between parents (caregivers) and children, relationship with peers, and children's personal characteristics. This section concludes with a summary and a proposed model for describing the indirect effects, that is, the hypothetical trajectories along which the combined domains of risk might be expected to operate in contributing to risk for exposure to violence. The rest of the review deals with the effects of exposure on children's and youths' outcomes. Section 1.2.1 describes the literature concerned with the direct effects of exposure on internalizing problems and then the mediating or moderating effects on this relationship of variables in each ecological domain. Section 1.2.3 does the same for externalizing problems, and Section 1.2.5 deals with the direct and indirect effects of exposure on academic functioning.

We include studies in which exposure to community violence is defined as either victimization or witnessing or is left unspecified. Where possible, we indicate which type of exposure is associated with a given risk factor and the age or developmental stage (e.g., child or adolescent) at which it operates. Risk factors are classified in terms of the ecological domains listed above. For each risk factor, we first describe studies that report direct effects on exposure and, at the end of the section, offer a hypothetical model, consistent with the information reviewed or with current thinking, which specifies the putative indirect trajectories between each domain of risk and exposure.

We consider first the risk factors in the context most distal to the individual child, namely, the neighborhood or community. Among these are economic disadvantage, social disorder, lack of social control, and situation (times and places where exposure is most likely).

Although most studies of community violence focus solely on relatively poor neighborhoods, it seems apparent from the few comparative studies that economic disadvantage gives rise to increased exposure. Esbensen and Huizinga (1991), in a study of 877 11–15-year-olds in Denver, found that neighborhood type—specifically, neighborhoods characterized by poverty and unemployment and such variables as ethnic diversity, high density, and mobility—was related to rate of personal victimization. In another large study of 2248 Grades 6, 8, and 10 students in New Haven public schools, Schwab-Stone et al. (1995) reported that the poorest children (those enrolled in the free-lunch program) were more often witnesses of severe community violence. It seems more than likely that community economic resources protect against exposure in a variety of ways, such as providing attractive alternatives to hanging out on the street, after-school and summer programs, enriched school programs, better policing and neighborhood surveillance and law enforcement, and by reducing the incentives for underground illegal economic ventures, such as dealing drugs that carry increased risk of violence.

Communities characterized by social disorder, as indexed by high levels of crimes against persons and violence, provide a context in which children are more likely to be exposed to serious violence (Selner-O'Hagan, Kindlon, Buka, Raudenbush, & Earls, 1998). Bell and Jenkins (1993), in a survey of 536 inner-city African American 7–15-year-olds in Grades 2–8 and 1035 adolescents from four high schools and two middle schools, found that exposure to violence is cumulative. Those children who had witnessed a killing, the most severe form of violence, were likely to have witnessed less severe forms as well, such as robbery, shooting, and stabbing. Esbensen, Huizinga, and Menard (1999), reporting on data from the Denver Youth Survey of 1530 adolescents from high-risk areas, found that prior victimization was the best predictor of later exposure. Malik, Sorenson, and Aneshensel (1997), studying 707 high school students aged 14–17 years, found results consistent with these, namely, that exposure to weapons and violent injury as well as threats and physical assault predicted victimization in the community. Sheley, McGee, and Wright (1992), in a study of 1653 inner-city high school students, reported that gun-related victimization predicted increased exposure to violence elsewhere than in school.

Increased risk of exposure, given prior exposure, particularly victimization, raises the question of whether children, in an attempt to protect or defend themselves in socially disordered communities, actually place themselves in greater danger.

Neighborhoods that exhibit a breakdown in social processes, usually characterized by concentrated poverty and residential instability, are likely to be places where social control over behavior is reduced. As might be expected, such neighborhoods are generally, although not always, found to give rise to the highest rates of exposure to violence among youth (Sampson & Lauritsen, 1994). Ethnographic work with youth in inner-city neighborhoods in the Bronx in New York City suggests that rates of victimization are actually underreported in such neighborhoods because distrust of the police, a marker of reduced social control and breakdown in social processes, is so widespread (Freudenberg et al., 1999). The relation between intact social processes and exposure to violence may not be straightforward, however. Sheidow, Gorman-Smith, Tolan, and Henry (2001), in a study of 249 inner-city African American 13–17-year-old boys, unexpectedly found that a higher concentration of youth in struggling families in neighborhoods with reasonably intact social processes gives rise to higher rates of exposure.

Two situational parameters, location and time, have been found associated with increased risk for exposure to community violence for youth. Inner-city (Bell & Jenkins, 1993) and core metropolitan neighborhoods are high-risk areas for violence, including both homicide and potentially lethal violence such as assault. Gladstein, Rusonis, and Heald (1992), comparing 339 inner-city adolescents with 435 adolescents in a resort community, found that the former were more often victimized and witnesses to assaults, rapes, knifings, life-threatening events, and murders. Homicide by firearms was reported in national surveys to be more likely than homicide by other means in core metropolitan areas (O'Donnell, 1995). Within these areas, an ethnographic study by Lockwood (1997) showed that home and school were the locations where violence was most likely to be encountered, and Richters and Martinez (1993a), studying 165 6–10-year-olds in Grades 1, 2, 5, and 6 in a single school in a moderately violent neighborhood, reported that younger children who were both victimized by and witnesses of violence were more likely to live in houses than in apartments and that older children were more likely to encounter violence at the hands of strangers if they lived in houses than in apartments.

Time of day appears to be a consistent risk factor for victimization of juveniles. The hours from 12 noon to 6 PM (after school) are the most dangerous (Snyder & Sickmund, 1999). This indicates a lack of caregiver supervision (Reese et al., 2000) during these hours and strongly suggests that after-school programs would be heuristic preventive measures. At a more basic level, Reese et al. argue that violence prevention programs are ineffective largely because they have failed to involve caregivers and parents. They note that there are many obstacles to full-time and effective parenting at a period when families are pressured to devote more time to work and less time to caregiving.

Studies examining the effects of family risk factors, the next least distal domain of influence, have focussed on familial stress, family or household structure, family conflict, and high-risk behaviors of family members.

Economic stress in the family appears to be an important risk factor for exposure to violence. In telephone interviews with parents and adolescents in a national probability sample of 4023 12–17-year-olds, household income was found to be negatively related to the prevalence of witnessing violence, physical assault, physically abusive punishment, and sexual assault in Caucasian although not in African American or Hispanic youth (Crouch, Hanson, Saunders, Kilpatrick, & Resnick, 2000). It is unclear whether these differences are artifactual due to limited range or ceiling effects in the different groups. Other traditionally studied stressful life events have not yet been evaluated for their effects on exposure to community violence.

Familial or household structural characteristics, such as single parenthood, have often been found to be associated with a variety of youth problems. Stable secure living arrangements would be expected to protect against exposure to violence. Consistent with this expectation is the finding by Bell and Jenkins (1993) that children and adolescents between the ages of 7 and 15 living in Chicago homes in which the father was absent were more likely to be involved in fights. Likewise, Esbensen and Huizinga, studying a panel of 877 adolescents from the Denver Youth Survey, found that adolescents living in intact families reported the lowest rates of personal victimization, whereas those living in single-parent homes reported the highest rates Esbensen & Huizinga, 1991, Malik et al., 1997. Later, Esbensen et al. (1999) reported that adolescents who were socially isolated from their families were at increased risk for exposure to community violence. In another large study of 1653 inner-city high school students, number of siblings was positively correlated with gun-related victimization (Sheley et al., 1992), and although the effect size was small, a number of hypothetical processes, such as decreased parental monitoring involved in raising larger families in poor neighborhoods or modeling of maladaptive behavior in older peer networks, could account for such a finding. One study reported results inconsistent with the expectation that household structure was associated with violence exposure. Gorman-Smith and Tolan (1998), in their study of 245 11–15-year-old African American and Latino boys in Chicago, did not find that family organization and other family factors made an independent contribution to the prediction of exposure beyond that of other risk factors such as the breakdown of traditional social processes in the community.

Three studies support an association between family conflict and adolescents' exposure to community violence. Bell and Jenkins (1993) reported that family violence increased the risk for witnessing shootings or stabbings in 7–15-year-olds in Chicago, and Osofsky, Wewers, Hann, and Fick (1993) reported a strong relationship between family conflict and witnessing community violence in 9–12-year-old African American children. In addition, using 5 years of data from 1530 11–19-year-olds in the Denver Youth Survey, Esbensen et al. (1999) showed that parental involvement in violent behavior and negative family interaction quality were related to risk for adolescent criminal victimization. It is not immediately apparent what the mechanisms are that bring about the association, but the problem of identifying the possible processes is clearly one that requires attention. One possible mechanism might be that youth exposed to domestic violence and abuse are driven away from their households, as studies of runaway youth show. Insofar as stable homes are protective, such youth are placed at increased risk.

High-risk behavior among family members is associated with increased risk of exposure for adolescents. Two studies showed that substance use among parents was related to victimization among youth. In the Denver Youth Survey, parental substance use was related to adolescent criminal victimization (Esbensen et al., 1999), and in ethnographic work (Freudenberg et al., 1999), adolescents cited parental alcohol and drug use as a cause of violent behavior directed toward youth.

Child maltreatment, and in some cases even physical punishment short of maltreatment, have been shown by a number of investigators to be related to increased risk for exposure to extrafamilial violence. Bell and Jenkins (1993) found among 7–15-year-olds that being physically punished was associated with increased involvement in fights. Lynch and Cicchetti (1998), comparing 188 maltreated and 134 nonmaltreated 7–12-year-olds, found that maltreatment predicted both victimization and witnessing of community violence over 1 year. Malik et al. (1997) found among 707 high school students that child abuse increased the risk for both community and dating victimization. Also, finally, Whitbeck and Simons' (1990) work on runaways, whom they found to be greatly at risk for violent victimization, showed that being in an abusive family was what motivated the children to leave home in the first place.

Conversely, attachment to parents reduced the risk for adolescent victimization in the 1530 youth studied for 5 years in the Denver Youth Survey (Esbensen et al., 1999), and attachment to family reduced the risk for assault, as shown in data from the national databases of the National Youth Survey and Monitoring the Future (Lauritsen, Laub, & Sampson, 1992). Also, protective against adolescent victimization are good parental discipline and monitoring practices (Esbensen et al., 1999). Although not a direct measure of the parent–child relationship, a study by Hill and Jones (1997) of 50 boys and 46 girls aged 9–12, half from high-violence and half from low-violence neighborhoods, showed that discrepancies between parent and child estimates of amount of children's exposure to violence seem to be associated with risk for further exposure. The discrepancy may reflect a lack of involvement of parents with children, which has been found in other studies to be a risk factor for exposure.

Finnegan, Hodges, and Perry (1998), reporting on the relationship between mother–child interaction and peer victimization in 184 preadolescents, found that perceived maternal overprotection was associated with peer victimization for boys, whereas perceived maternal rejection was associated for girls. The authors theorize that poor parenting “hinders children's development of gender-salient competencies”—in this instance, “autonomy for boys and communion for girls”—which places them at risk for peer victimization.

As noted above, one study of boys in high-crime Chicago neighborhoods (Gorman-Smith & Tolan, 1998) did not find that family factors, including parenting practices, predicted children's exposure to violence. The authors suggest that other neighborhood risk factors in these kinds of dangerous neighborhoods may operate on children of this age directly and that family factors are not strong enough to mediate or moderate their effects.

Supportive friendships, negative social status among peers, and association with deviant peers might be expected to influence children's exposure to violence.

Much of the victimization suffered by children and adolescents in their schools and neighborhoods is the result of perpetration of violence by peers. It is not surprising, therefore, to find that in a study of 229 boys and girls in Grades 3–7, children who lacked supportive friends or who were rejected by peers were especially likely to be victimized by peers (Hodges, Malone, & Perry, 1997). These investigators also reported that in 173 children followed over the course of 1 year, peer rejection predicted increases in victimization (Hodges & Perry, 1999). In addition, in a review of their own and others' work, Isaacs, Card, and Hodges (2001) concluded that peer rejection predicts increases in victimization over time. Consistent with that group's findings, Schwartz, McFadyen-Ketchum, Dodge, Pettit, and Bates (1999) found in a sample of 389 kindergarten and first-grade children of mixed low and middle socioeconomic status that a positive relationship between teacher-rated behavior problems and peer victimization 3 years later was moderated by both dyadic friendships and peer rejection.

Associating with deviant peers has been found to raise the risk for victimization. Whitbeck and Simons (1990), studying more than 80 male and female adolescents who had run away from abusive families, found that increased exposure to violence was due not only to the deviant behaviors the adolescents engaged in to subsist but also to their association with peers who engaged in deviant and illegal behavior. Similarly, insofar as a condition of membership in some gangs is engagement in delinquent activities, gang membership itself can be construed as associating with deviant peers, thereby increasing the likelihood of victimization (Thornberry, 1998).

Among the personal characteristics of children that might be expected to influence risk for exposure to community violence are their own behavior, cognition, academic functioning, and such demographic characteristics as gender, ethnicity/race, and age.

Delinquent, confrontational, and violent antisocial behavior have been reported in virtually all studies to raise the risk for victimization, much of it serious, and to a lesser extent the risk for witnessing violence as well. Delinquency was found by Esbensen and Huizinga (1991) in their study of 877 adolescents to be associated with victimization. Based on data from the National Youth Survey and Monitoring the Future, Lauritsen et al. (1992) found that delinquency predicted victimization by assault and robbery. Also, as mentioned earlier, deviant behaviors used for subsistence in the streets by runaway youth have been found to contribute to increased victimization among this high-risk group (Whitbeck & Simons, 1990).

Both qualitative and quantitative interviews with adolescents in the Bronx revealed that those who had been shot were more likely to have hit, punched, or hurt someone badly or threatened someone with a weapon (Freudenberg et al., 1999). Selner-O'Hagan et al. (1998) found that exposure to violence was more likely among violent offenders in a sample of 80 African American males living in a high-crime area. DuRant, Cadenhead, Pendergrast, Slavens, and Linder (1994), studying 225 Black male and female 11–19-year-olds in nine high-crime housing projects, found a correlation between self-reported use of violence and exposure to violence, especially victimization. Risk for violent victimization in or around school for 1090 elementary, 692 junior high, and 813 senior high school students was higher for students whose behavior was confrontational, i.e., who reported that it was impossible to walk away from a fight or confrontation (Fitzpatrick, 1999). Bell and Jenkins (1993) reported that fighting and weapon-carrying in 10–18-year-olds was associated with witnessing shooting and stabbing.

Carrying weapons, especially guns, seems invariably to place adolescents at risk for serious victimization. Further examination of a sample mentioned above (DuRant, Getts, Cadenhead, & Woods, 1995) showed that 35% of males and 16% of females carried weapons (often hidden) and that this was correlated with frequency of receiving a serious injury during a fight. Sexually abused youth have been found more likely to have carried a gun in the past 6 months (Freudenberg et al., 1999). Sheley et al. (1992), studying gun-related victimization in 1653 inner-city high school students, found that carrying guns accounted for a substantial amount of the effect.

Substance use was found in two studies to be related to victimization. Esbensen and Huizinga (1991) reported that alcohol and marijuana use were associated with victimization. Sheley et al. (1992) reported that both using and dealing drugs placed 1653 inner-city high school students at increased risk for gun-related victimization.

Both internalizing and externalizing behavior problems were found to be related to peer victimization among 229 boys and girls in Grades 3–7 (Hodges et al., 1997). The same group of investigators studied increases in peer victimization over the course of 1 year and found that internalizing problems and physical weakness were predictive (Hodges & Perry, 1999), as were teacher-reported internalizing and externalizing problems (Hodges, Boivin, Vitaro, & Bukowski, 1999). Isaacs et al. (2001) suggest that a cyclical process, involving internalizing and externalizing problems and the ensuing changes in self-perceived social competence, occurs to produce peer victimization over time. Results of work by Lynch and Cicchetti (1998) on 188 maltreated and 134 nonmaltreated (Aid for Dependent Children and Families) 7–12-year-olds are consistent with these findings, showing that externalizing behavior problems at time one predict both witnessing and victimization 1 year later. Studying younger children (389 kindergarten and first-grade children) of mixed and low SES, Schwartz et al. (1999) found that teacher-reported behavior problems predicted peer victimization 3 years later. Also, Kuther and Fisher (1998) found in 123 Grades 6–8 suburban students that those who reported self-controlled coping in response to hypothetical stressful situations, i.e., those who attempted to keep their emotions to themselves, were more likely to have been victimized by community violence, although assertiveness was found to be negatively correlated with victimization.

How children and youth cognitively process violence appears to influence their own involvement in violence. In an intensive ethnographic study (Lockwood, 1997) of 229 violent incidents among middle and high school students in schools with high violence rates, in-depth interviews revealed that the largest number of violent incidents occurred between students who knew each other. The incidents generally escalated from trivial affronts to more serious incidents where violence was excused as justifiable in the service of such goals as retribution. Use of violence was based on a well-developed value system in which violent behavior was considered acceptable. Malik et al. (1997) found among their sample of 707 high school students that justification for using violence was related to dating victimization but not to community victimization, a finding reminiscent of justifications given by children who have been maltreated by their parents.

There is not much work on the potential protective effect of school on involvement in violence. However, an analysis by Lauritsen et al. (1992) of data in the national databases of the National Youth Survey and Monitoring the Future shows that attachment to school reduces the risk for assault.

A number of demographic variables, including gender, ethnicity/race, and age, has been shown to affect the risk for exposure to community violence.

It is very clear over a wide range of studies of children and adolescents that boys are more likely than girls to be victimized by community violence. Bell and Jenkins (1993) found that among 7–15-year-olds, boys are more likely to be involved in fights. DuRant et al. (1995) reported that more frequent weapon-carrying among males than females leads to the receipt of serious injury. Freudenberg et al. (1999), in their ethnographic work in the Bronx, found males more likely than females to report having been beaten up, stabbed, or shot. Jenkins and Bell (1994), in their study of inner-city high school students, found that boys were more likely to have been robbed, shot, or beaten, whereas girls were more likely to have been raped. Malik et al. (1997), also studying high school students, found that girls reported less community victimization than boys. Sheley et al. (1992) found more gun-related victimization among inner-city high school males than females. Also, O'Donnell (1995) reported, on the basis of national survey data that for 15–19-year-olds, homicide and suicide by firearms are together six times higher for males than females.

Gender differences are also found for witnessing community violence. Schwab-Stone et al. (1995) found among 2248 students in Grades 6, 8, and 10 in urban public schools that more boys than girls reported witnessing violence. In developing the structured interview, My Exposure to Violence (My ETV), Selner-O'Hagan et al. (1998) found that among their 80 child and adolescent subjects, males reported more exposure than females.

Circumstances by which high-risk adolescents become victims of community violence are different for girls and boys. In their very interesting study of runaway adolescents, Whitbeck and Simons (1990) found three gender differences in the trajectories leading to victimization in the street. First, abuse within the family was directly related to the deviant subsistence strategies that increased risk for victimization for boys but indirectly related for girls. Second, affiliation with deviant peers directly affected victimization for boys but not for girls. Finally, frequency of running away was associated with victimization of girls but not of boys.

As for race/ethnicity, in almost every study, minority subjects, especially African American youth, were more exposed to community violence than Whites. Berton and Stabb (1996) reported that more Black than White adolescent high school juniors were exposed to violent crime in their neighborhoods and schools. Malik et al. (1997) reported that African American high school students were more victimized by community violence than Latino, Asian, and White students, and O'Donnell (1995) reported on the basis of national databases that African American youth have the highest rate of death by firearms, with their homicide rate over nine times that for Whites. O'Keefe and Sela-Amit (1997), studying 899 Latino, African American, White, and Asian urban and suburban 14–19-year-olds, found that race/ethnicity was an important risk factor for exposure to school and community violence even when SES was controlled for. Schwab-Stone et al. (1995) reported that White public school students witnessed less community violence than minority students. In Selner-O'Hagan et al.'s (1998) sample, African American youth were more at risk for exposure to community violence than other groups. Also, Crouch et al. (2000) reported more witnessing of violence for African American and Hispanic youth than for Whites.

As for the variable of age, Finkelhor and Dziuba-Leatherman (1994), in an important national survey, pointed out that childhood is a period during which more victimization occurred than during adulthood. Younger children, however, were more at risk from family than from strangers or peers. With respect to less serious victimization, Bell and Jenkins (1993) found that the younger children among their 7–14-year-olds were more involved in fights than the older children. For more serious forms of victimization by community violence, however, older youth are more at risk. O'Donnell (1995) showed that 16–19-year-olds have the highest rate of handgun victimization among all age groups, and Selner-O'Hagan et al. (1998) showed that among children and youth aged 9–24 years living in a high-crime area, older youth were most at risk for exposure to community violence.

Empirical evidence accumulated over the last decade shows that each domain of risk, from distal to proximal, influences children's and youths' exposure to community violence. The studies reviewed to this point all show a direct association between each risk factor studied and exposure to community violence in terms of both witnessing violence and being victimized by it. For the most part, studies leave unanswered the more important questions concerning the mechanisms that operate either to raise or to mitigate risk between each domain of risk and exposure. We propose an ecological model that lays out hypothesized indirect paths between each domain and exposure that represent plausible mediating and moderating processes (see Fig. 1).

We describe the proposed indirect paths of influence as we did the direct effects, beginning with the most distal, the neighborhood and community context. We posit that community context directly influences family and peer domains, which, in turn, shape the behavioral characteristics of the children. In tracing probable paths of influence between community and neighborhood variables and individual children's risk for exposure, we have assumed that children's behavior, the most proximal domain of risk, is most likely embedded within the domains of family and peer contextual influences. Consistent with this assumption, we do not include a direct path between community context and children's personal characteristics because the effects of community are most likely mediated through family and peer systems.

We begin with the trajectory involving family influences. The effects of family through which community factors work to influence children's functioning can be attributed to two levels of context: one general, consisting of the stresses and structural characteristics of the family as a whole and the high-risk behavior of family (or household) members, and the other more specific and proximal, consisting of the relationship between parents and children. Community factors are hypothesized to affect both levels of family context, but their effects on the more proximal level of parent–child relationships are expected to be partly mediated through the more distal general household and family characteristics. The other major trajectory through which community factors are posited to influence the children's behavior, and thereby exposure, is through their relationship with peers. We believe that both these indirect trajectories, combined with the direct effects of each domain of risk on exposure, provide a comprehensive picture of the ways in which children's risk for exposure in various kinds of communities is determined. A new generation of research is needed to operationalize and test these more comprehensive models. Knowledge of the indirect trajectories suggests specific mechanisms and processes that operate to raise risk for exposure and denote the intersections where interventions aimed at reducing risk for exposure might be most successful. Such interventions are needed even while research to uncover and confirm the basic processes that raise risk for exposure proceeds.

One additional note here is that many studies have found that the personal demographic variables of gender, ethnicity/race, and age have an effect on risk for exposure to community violence. It is probable that the proposed trajectories in the general model will need to be modified as empirical tests of the model with different subgroups of children are examined.

Studies that address the effect of exposure to community violence on children's and youths' functioning reveal a wide range of effects, with a good deal of agreement as to what many of those effects are. Our review classifies the effects generally as those affecting internalizing behavior, externalizing behavior, and academic functioning. We first present the work that shows direct associations between exposure and internalizing problems followed by a section on the variables found to influence (mediate or moderate) these effects. We then do the same for externalizing problems and for academic functioning.

An association between exposure to community violence and various forms of internalizing symptomatology (see Duncan, 1996 for an earlier overview) seems quite well supported by a substantial number of studies, some with very large samples. Schwab-Stone et al., 1995, Schwab-Stone et al., 1999 found that witnessing community violence was associated with depressed mood, anxiety, and somatization in multiethnic samples of over 2000 public school children. Singer, Anglin, Song, and Lunghofer (1995) found that exposure to physical violence was related to depression, anxiety, and PTSD in 3735 14–19-year-old high school students. Boney-McCoy & Finkelhor, 1995, Boney-McCoy & Finkelhor, 1996, in data from a national telephone survey of more than 1400 10–16-year-olds, mostly White, found victimization (assault) to be related to depressive symptoms over a 15-month period even after controlling for earlier psychopathology.

Freeman, Mokros, and Poznanski (1993), in 223 6–11-year-old urban schoolchildren, found that exposure was related to low self-esteem and violence to self, relatives, or friends was related to depression. Gorman-Smith and Tolan (1998) found that exposure was associated with an increase in depression over a 1-year period in 243 African American and Latino inner-city boys ages 11–15. Kliewer, Lepore, Oskin, and Johnson (1998) found in 99 8–12-year-olds that exposure was related to intrusive thinking, anxiety, and depression, controlling for demographics and concurrent stressors. Lynch and Cicchetti (1998) found in 188 maltreated and 134 nonmaltreated 7–12-year-olds that exposure to high levels of community violence was related to self-reported traumatic stress, depression, and lower self-esteem. Hill, Levermore, Twaite, and Jones (1996) found in 97 African American Grades 4–6 students from low- and high-violence neighborhoods that anxiety correlated with exposure even when social support and income were controlled. Osofsky et al. (1993) found that children's stress symptoms and exposure to community violence, both based on mothers' reports, were correlated. Shahinfar, Fox, and Leavitt (2000), in a relatively infrequently studied preschool sample, found that witnessing community violence related to parent reports and self-reports of distress in 155 Head Start 3.5–4.5-year-olds. Juvonen, Nishina, and Graham (2000) found in 223 6–11-year-old schoolchildren that self-perceived harassment by peers predicted depression, loneliness, and self-worth.

Suicidality was examined by Elze, Stiffman, and Dore (1999), who found it related to both victimization and witnessing. Pastore, Fisher, and Friedman (1996) found in 630 urban teenagers that likelihood of suicidal ideation and attempts was increased by knowing someone who was murdered and by witnessing a stabbing.

Not surprisingly, in addition to studies already cited in connection with the internalizing problems of depression, anxiety, and distress, a number of studies have assessed traumatic effects of exposure to community violence, as indexed by PTSD symptomatology. Berman, Kurtines, Silverman, and Serafini (1996) found in 96 14–18-year-old urban high school students that 34.5% of those exposed to violence met DSM-III-R criteria for PTSD. An association between exposure and PTSD symptoms was also reported by Berton and Stabb (1996) who found among 97 high school juniors that 29% exhibited PTSD-like symptoms, by Fitzpatrick and Boldizar (1993) among 221 7–18-year-old low-income African Americans, by Jenkins and Bell (1994) among 203 African American high school students in a high-violence urban area, by Mazza and Reynolds (1999) among 94 11–15-year-old inner-city parochial school students even after controlling for depression and suicidal ideation, and by Overstreet, Dempsey, Graham, and Moely (1999) among 75 10–15-year-old inner-city African Americans.

Pynoos et al. have focused on the sequelae of single, particularly intense or catastrophic experiences in whose wake PTSD symptomatology would seem especially likely. Following a sniper attack on a school playground, reactions among 159 5–13-year-olds increased with physical proximity to the event (Pynoos et al., 1987). The best predictor of persisting PTSD symptomatology 14 months later was the extent of exposure as indexed by proximity to the event and knowing a child who was killed (Nader, Pynoos, Fairbanks, & Frederick, 1990).

In a rare example of a violence exposure study including direct physiological measures, Cooley-Quille and Lorion (1999) measured blood pressure and pulse rate and sleep disturbance in a sample of 64 16–18-year-old mostly African American urban youth. As might be expected, exposure was positively related to sleep disturbance. In addition, however, the highest levels of exposure were associated with the lowest resting pulse rates. The investigators conjecture that in pervasively violent communities, physiological adaptation as well as emotional desensitization may occur, an idea well worth further investigation. A related idea, framed in more cognitive terms by Garbarino (e.g., Garbarino et al., 1992), is that living in what he has referred to as urban war zones brings about fundamental developmental changes, altering children's views of the world and expectations for the future and leading to so-called pathological adaptation in which children escape depression and other internalizing problems but engage in high-risk and aggressive behavior (Ng-Mak, Salzinger, Feldman, & Stueve, in press).

Some inconsistencies among findings point to unresolved issues in how various dimensions of exposure (e.g., witnessing vs. victimization; intensity and duration of exposure) may be differentially related to outcome. Martinez and Richters (1993) and Richters and Martinez (1993b) found in 165 6–10-year-olds in a moderately violent low-income neighborhood that children's distress is related to both victimization and witnessing community violence, and Elze et al. (1999), in 792 14–18-year-olds, found the same for depression. However, Fitzpatrick (1993), in 221 7–18-year-old low-income African Americans, found depression related to victimization but not to chronic witnessing of violence, and Farrell and Bruce (1997), in 436 mostly African American urban sixth graders, similarly found witnessing to be unrelated to changes in emotional distress over time. Furthermore, not all studies measure both victimization and witnessing or report them separately.

There are also issues with regard to the assessment of symptoms. Some studies use diagnostic instruments and some various symptom checklists. For example, Fitzpatrick (1993) assesses depression with the Children's Depression Inventory (CDI; Kovacs, 1985). Kliewer et al. (1998) use both the CDI and the depression/anxiety subscale of the Child Behavior Checklist (Achenbach, 1991), while Freeman et al. (1993) use the Children's Depression Rating Scale-Revised (Poznansky, Freeman, & Mokros, 1985). Aside from questions that routinely arise when comparing studies using different assessment procedures, not all studies assess the persistence of any identified symptomatology or any consequent functional impairment. While it is certainly an important first step to show that anxiety, for instance, increases as a consequence of exposure to community violence, we then need to know whether or under what conditions it causes significant distress or interferes with normal development. Only some of the work cited addresses these issues.

Many factors have been shown to affect the relationship between exposure and internalizing symptomatology. In line with our overall schema, we begin with the more distal contexts of neighborhood or community. Overstreet and Braun (2000) found in 70 inner-city African American 10–15-year-olds living in or near public housing that impact of community violence exposure on PTSD symptoms was mediated by the youths' perceptions of decreased neighborhood safety. Shumow, Vandell, and Posner (1998), in 168 African American and White urban fifth graders and their parents, found evidence that children's and parents' perceptions of neighborhood danger mediated the relationship between neighborhood risk and their reports of psychological distress in the children.

Two studies, in particular, underscore the importance of the perspective from which exposure is measured for understanding exposure–outcome relationships. In Graham and Juvonen's (1998) study of 418 ethnically diverse, public school sixth and seventh graders, self-perceived and peer-perceived victimization predicted differently, the former predicting intrapersonal, internalizing consequences (e.g., anxiety), and the latter interpersonal consequences (e.g., peer rejection). Also, in Howard, Cross, Li, and Huang's (1999) study of 333 parent–youth dyads in urban housing developments, parent–youth concordance in reports of the youths' exposure appeared to act as a moderator of the association between exposure to community violence and the youths' distress, in that low concordance was associated with distress symptomatology and lower self-esteem, while high concordance was associated with positive aspects of youths' and families' functioning. Low concordance arose from parents underestimating both the exposure and the distress of their children, a point to keep in mind when evaluating studies relying on parents as informants.

Several already cited studies found evidence for the mediating role of family contextual variables. Overstreet and Braun (2000) found that increased family conflict mediated the impact of exposure on children's PTSD symptoms. Gorman-Smith and Tolan (1998) found that boys in less cohesive families suffered more depression and anxiety following exposure. Presence of mother in the home was found to protect against depression in the face of exposure (Fitzpatrick, 1993) and absence of mother increased the risk for depression, although not for PTSD, as exposure to community violence increased (Overstreet et al., 1999). Absence of fathers or brothers in the home exacerbated the relationship between exposure and PTSD symptoms (Fitzpatrick & Boldizar, 1993).

Kliewer et al. (1998) found that exposure to community violence had its strongest internalizing effects in children having low social support from mothers or caretakers and in those having high social strain (the major component of which was again low maternal support). Kuther and Fisher (1998), in a mostly White upper-class sample of 123 Grades 6–8 private parochial school students, found that the victimization–distress relationship was moderated by perceived family support.

For all the extensive attention peer relations have received in the child development literature, peer support seems to have been largely overlooked with respect to its possible specific role in mediating the association between community violence exposure and internalizing problems. One exception is the study by Hill and Madhere (1996), who found some evidence that peer support moderates the impact of exposure on anxiety. Berman et al. (1996), in their study of 96 ethnically diverse 14–18-year-old urban high school students, found that perceived social support from the students' personal networks mediated the effects of exposure on PTSD symptomatology. Unfortunately, the investigators do not report data on the separate components of the support networks, and so the specific role of peers cannot be evaluated.

In the most proximal domain—the child's personal characteristics—several variables have been shown to play a moderating role in the face of exposure to community violence. Dempsey, Overstreet, and Moely (2000) studied the use of both avoidant and approach coping strategies in 70 African American 11–14-year-olds. Low use of behavioral avoidance under conditions of high exposure was associated with increased behavioral arousal (e.g., feeling jumpy or scared). Cognitive distraction under the same conditions was associated with increased cognitive arousal (i.e., trauma-related symptoms such as flashbacks). The two coping strategies characterized as approach (problem-solving and social support) showed no significant effects. Dempsey et al. discuss their results in terms of the importance contextual factors have for the effect of any coping strategy. Hill and Madhere (1996) found in 150 African American inner-city Grades 4–6 students that in the face of chronic exposure, children with a higher willingness to retaliate showed lower levels of state anxiety. This suggests, the authors remark, that some intervening behavior serves to reduce anxiety.

Intrusive thinking and PTSD symptoms, previously noted as possible direct consequences of exposure to community violence, may also affect the association between exposure and outcome. Kliewer et al. (1998) found intrusive thinking to partially mediate the exposure-internalizing association. Mazza and Reynolds (1999) found that PTSD mediated between exposure and both depression and suicidal ideation.

Fitzpatrick (1993) found that victimized females reported more depressive symptoms than males. Fitzpatrick and Boldizar (1993) and Jenkins and Bell (1994) found PTSD symptoms to be more severe for victimized females. In fact, Jenkins and Bell make a point of commenting on the weak association between exposure and distress for boys in contrast to that for girls. White, Bruce, Farrell, and Kliewer (1998), in 385 mostly African American 11–14-year-olds, found exposure to be significantly correlated with anxiety in girls but not in boys. Paquette and Underwood (1999), in 76 11.8–14.7-year-olds, found more distress among girls than among boys in response to peer victimization.

Martinez and Richters (1993) found that age mediated between the nature of the exposure and distress such that the older children were more affected by violence involving persons they knew, while the younger children were more affected by seeing guns or drugs at home. Fitzpatrick (1993) found that younger exposed children reported more depressive symptomatology and hypothesized that older youth who chronically witnessed violence but showed few or no depressive symptoms may be coping with their exposure in a way that is somehow protective in the short run but may undermine long-term healthy adaptation. Schwab-Stone et al. (1999), like Fitzpatrick, found that exposure to community violence and internalizing symptoms were more strongly related for younger than for older children, although it should be noted that the former study includes adolescents only, while the latter includes ages 7–18 years.

Cooley-Quille, Turner, and Beidel (1995), in 37 7–12-year-olds, found high exposure to be associated with externalizing, but not with internalizing, problems. DuRant, Pendergrast, and Cadenhead (1994) found among 225 African American adolescents that fighting in the previous year was predicted by exposure and victimization and that self-reported violent behavior was correlated with exposure (Durant, Pendergrast, et al., 1994). Farrell and Bruce (1997) found in 436 mostly African American urban sixth graders that witnessing community violence was associated with the frequency of violent behavior in both boys and girls, although its association with changes in the frequency of violent behavior held for girls only. Other studies reporting an association between exposure to community violence and externalizing problems for both boys and girls include Hill et al. (1996) for parent-reported behavior problems and Song, Singer, and Anglin (1998) for self-reported violent behavior. Shahinfar et al. (2000) found that victimization was related to externalizing problems in their preschool sample. In contrast, O'Keefe (1997) found that exposure was related to aggression only for the boys among the 935 urban and suburban high school students in her sample and the analysis controlled for several demographic variables, including family violence.

In a stratified national sample of 2595 Grades 3–12 students, Fitzpatrick (1997a) found that threats and victimization predicted fighting across the whole age range, and in a sample of 150 low-income African American 8–14-year-olds, exposure to community violence discriminated between aggressive and nonaggressive youth (Fitzpatrick, 1997b). Gorman-Smith and Tolan (1998) found exposure to community violence related to aggression over a year's time even with previous status controlled. Sussman, Simon, Dent, Steinberg, and Stacy (1999) found that self-identification with a high-risk peer group and not avoiding dangerous places (both of which can be assumed to increase risk for exposure) predicted perpetration 1 year later in a multiethnic sample of 962 high-risk boys and girls in continuation high schools.

Singer et al. (1999) found among 2245 public school African American and White boys and girls in Grades 3–8 that exposure to community violence predicted violent behavior. Jenkins and Bell (1994) found in 203 African American 13–18-year-olds in a single urban school that exposure was related to both offensive and defensive fighting, and DuRant, Pendergrast, et al. (1994) found that exposure predicted gang fighting. Schwartz, McFadyen-Ketchum, Dodge, Pettit, and Bates (1998) found that victimization by peers, assessed sociometrically, was associated with externalizing problems both concurrently and prospectively 2 years later among 330 third and fourth graders.

Given the preponderance of evidence, the finding by Osofsky et al. (1993) that parent-rated externalizing behavior was not related to exposure to community violence appears anomalous. The authors suggest two reasons: limited variability in the behavior problem scores, which were in general high, and the possibility, given that exposure and behavior were both related to family conflict resolution, that family variables mediate the exposure–behavior relationship. In a study raising a particularly interesting methodological as well as substantive issue, Hill and Madhere (1996) found, in addition to a modest correlation between exposure to community violence and willingness to retaliate, that the children's perceptions of their exposure were more strongly associated with their adjustment than were the actual counts of incidents they had been exposed to.

Bowen and Bowen (1999) found in a national probability sample of 2099 middle and high school students that students' perceptions of school danger and of problem behavior by youth in their neighborhoods as well as self-reports of exposure in the preceding 30 days predicted school-related problems (e.g., suspensions and school complaints to parents). Elze et al. (1999) found high rates of neighborhood and school violence to be associated with conduct disorder. Miller, Wasserman, Neugebauer, Gorman-Smith, and Kamboukos (1999) found in 97 6–10-year-old urban boys at familial risk for antisocial behavior that witnessing community violence was associated with changes in antisocial behavior (CBCL Delinquency scale; Achenbach, 1991) over 15 and 30 months even when several parenting variables were controlled. Schwab-Stone et al. (1999), in their follow-up of over 1000 children, reported that witnessing community violence was associated with antisocial behavior initially and 2 years later.

Carrying weapons has been found associated with exposure to community violence in several studies. DuRant et al. (1995) found in 225 African American adolescents living in or near public housing projects that previous exposure—in particular, fighting and being seriously injured—was associated with weapon-carrying. Fitzpatrick (1997b) and Jenkins and Bell (1994) have reported similar findings.

Elze et al. (1999) found high rates of neighborhood and school violence to be associated with substance abuse or dependence. Fick and Thomas (1995), in a national sample of disadvantaged 10–13-year-olds participating in smoking and drug abuse prevention classes, found exposure related to use or intention to use cigarettes and alcohol. Also, Jenkins and Bell (1994) found exposure related to alcohol use by boys and girls and to drug use by girls.

Attar, Guerra, and Tolan (1994) found in 384 African American and Hispanic urban first, second, and fourth graders that total stressful events (including exposure to violence, transitions such as change of school or residence, and family life events) and exposure by itself predicted aggression only when neighborhood disadvantage was high. Shumow et al. (1998) found that parents' perceptions of danger mediated between actual neighborhood crime rates and their reports of misconduct in their children.

Sampson et al. have devoted extensive attention to measurement of neighborhood-level social and physical variables and their relationship to crime in general and adolescent delinquency in particular (e.g., Sampson, 2000, Sampson & Raudenbush, 1999, Sampson et al., 1997). Sampson's construct of collective neighborhood efficacy—“social cohesion among neighbors combined with their willingness to intervene on behalf of the common good” (Sampson et al., 1997)—results in what he terms informal social control found to significantly inhibit delinquency even when prior neighborhood crime was taken into account Sampson, 1997, Sampson, 2000.

Gorman-Smith and Tolan (1998) examined the relationship of Family Structure, which included measures of support and the extent to which the family did not endorse deviant beliefs about the acceptability of such behaviors as cutting school, to the behavioral outcome of aggression. Exposure to community violence was related to aggression in High Structure but not in Low Structure families. The authors argue that while supportive families are in general protective, even High Structure cannot entirely protect against severe exposure's effects, while Low Structure does not add to risk.

Miller et al. (1999) found that the family environment, particularly parent–child conflict, affected the relationship between witnessing violence and antisocial behavior. With low conflict, witnessing was associated with increases in antisocial behavior, but in high-conflict families, witnessing had no independent effect. Howard et al. (1999) found, as was the case with internalizing symptomatology, that low parent–youth concordance in reporting the youths' exposure was related to their perpetration of violence.

As in the case of peers as mediators of the association between exposure and internalizing problems, there is a dearth of studies of peers as mediators of the exposure-externalizing association. Clearly, research is needed here. There are, however, some clues from the work on gangs and delinquency. Thornberry (1998) summarizes work showing, among other things, that associating with deviant or delinquent peers is a strong risk factor for gang membership, that adolescents with deviant or problem behavior are more likely to become gang members, and that gang members commit a disproportionately large number of serious offenses. While not a direct test of the mediation hypothesis, it seems entirely plausible to view association with delinquent peers, and certainly gang membership, as increasing both the likelihood of exposure to violence and subsequent externalizing problem behavior, with peer association mediating between them. Whether this conceptualization will prove heuristic in the long run, and whether it will apply more generally beyond gangs and the most serious offenses, remains to be seen.

Malik et al. (1997), in a multiethnic sample of over 700 high school students, found that personal norms concerning justification of violence constituted an important mediator between exposure to community violence and perpetration of violence in the community and both perpetration and victimization in dating relationships. Webster, Gainer, and Champion (1993) found in 294 inner-city, predominantly African American seventh and eighth graders that cognitive variables mediated between exposure and weapon carrying, although there were important gender differences. For boys, one risk factor for carrying a knife was not believing that weapon-carrying increased personal risk for injury or death and that for carrying a gun was willingness to justify using it. For girls, carrying a knife was again not associated with believing that weapon-carrying increased personal risk, but it was predicted by willingness to justify shooting someone. Schwartz and Proctor (2000), in a multiethnic sample of 285 inner-city Grades 4–6 students, found that the association between witnessing violence and aggressive behavior was mediated by social information processing, indexed here by the children's ratings of the value, expected outcome, and efficacy of aggressive responses to hypothetical situations. In contrast, emotional dysregulation, measured by teachers' ratings, did not mediate the foregoing association but rather mediated the association between being a victim of violence and such negative social outcomes as peer rejection.

Gender, as already noted, has been found to affect the relationship between exposure and externalizing problem behavior. Farrell and Bruce (1997) found exposure to be related to changes in frequency of violent behavior only for girls. Jenkins and Bell (1994) found that for boys, victimization was most strongly related to such outcomes as weapon-carrying (in contrast to the victimization–distress relationship for girls, described earlier). O'Keefe (1997) found that exposure significantly predicted aggression for boys. Webster et al. (1993) reported, as detailed in the immediately preceding Section 1.2.4.5.1, that the relationships between beliefs and weapon-carrying differed by gender.

There is sufficient evidence to conclude that gender should be taken into account in considering the exposure–outcome relationship. However, there are important methodological differences among studies. Sometimes, only one gender is studied (e.g., Miller et al, 1999). Even within studies, gender may relate to only some outcomes (e.g., Farrell & Bruce, 1997). Further work on gender's mediating role is needed.

Several studies have shown that exposure to violence has a direct negative relationship with children's academic functioning. A national telephone sample of 10–16-year-olds found that self-reported victimized youth reported more school difficulties (Boney-McCoy & Finklehor, 1995). Schwab-Stone et al. (1995) found that as witnessing of violence increased, so did poor academic performance. Bowen and Bowen (1999) found in a national probability sample of over 2000 middle and high school students that exposure to neighborhood and school violence (especially the former) affected school attendance, behavior, and grades. A 1-year prospective study of 120 African American junior high school students showed that perceptions or rates of neighborhood risk, including violence and crime, were negatively related to self-reported grades 1 year later (Gonzales, Cauce, Friedman, & Mason, 1996). Shumow, Vandell, and Posner (1999) found in 168 low-SES children followed from Grades 3–5 that neighborhood risk (including but not limited to violent crime), while independent of academic performance in third grade, was negatively associated with academic performance 2 years later.

Of the small number of studies here, several have focused on parenting. Richters and Martinez (1993c) found in 72 first and second graders that stability and safety of the home mediated the effects of exposure to violence on adaptational failure or success, including academic performance. Overstreet and Braun (1999) found in 45 African American 11–14-year-olds that children who perceived high achievement expectations and moral-religious emphasis in their families were most at risk for poor academic functioning with increasing exposure to violence. In contrast, at the lowest levels of exposure, the same children had the highest levels of academic performance. The Gonzales et al. (1996) study, reported above, showed that among children who rated neighborhood crime and violence as high, self-reported GPA increased with restrictive maternal control. In low-risk neighborhoods, such control decreased self-reported GPA. These last two studies should probably not be seen as contradictory but more likely seen to reflect different aspects of parental control or monitoring having a greater or lesser effect depending on the level of neighborhood violence the families and children are exposed to.

Bowen and Bowen (1999) suggest, based on the fact that exposure had a smaller effect on grades than on attendance and behavior, that the exposure–grades relationship may be mediated through psychological factors (e.g., school motivation) or that attendance and trouble-avoiding behaviors in response to danger in turn affect grades.

The Gonzales et al. (1996) study is the only one to have examined mediating effects of peer support. For children living in self-rated violent neighborhoods, GPA remains the same regardless of peer support, but in low-risk neighborhoods, GPA increases with peer support.

Juvonen et al. (2000), in 244 ethnically diverse middle school students, found empirical support for a model proposing that psychological adjustment (indexed by measures of loneliness, depression, and self-worth) mediates between perceived harassment by peers and the school outcomes of grades and attendance.

It is apparent from the sparse literature here that research is badly needed, although the available evidence points consistently to the negative effects of exposure on school functioning. Given the central role of school in children's lives, the lack of research is unfortunate. To begin with, good academic functioning is a marker of healthy developmental adaptation throughout childhood and adolescence. In the face of exposure, it may well serve as a powerful protective factor. Further, school is a place where exposure to violence often occurs. It is therefore a heuristic setting in which intervention or preventive programs should be situated to increase their likelihood for success.

Section snippets

Conclusions, future directions, and implications for intervention

The present review, like those cited earlier, shows that beyond the general conclusion that exposure to community violence can have serious deleterious effects on children's functioning and development, much remains to be learned. In this regard, the community violence literature has many similarities to the domestic violence and child maltreatment literature. At the most basic methodological level, differences in sampling, instruments, and analytic strategies often preclude conclusive

Acknowledgements

We wish to acknowledge support from a grant awarded to the first two authors from the National Institute of Mental Health, MH56198.

References (115)

  • M Jonson-Reid

    Youth violence and exposure to violence in childhood: an ecological review

    Aggression and Violent Behavior

    (1998)
  • S Malik et al.

    Community and dating violence among adolescents: perpetration and victimization

    Journal of Adolescent Health

    (1997)
  • M O'Keefe

    Adolescents' exposure to community and school violence: prevalence and behavioral correlates

    Journal of Adolescent Health

    (1997)
  • D.R Pastore et al.

    Violence and mental health problems among urban high school students

    Journal of Adolescent Health

    (1996)
  • M Schwab-Stone et al.

    No safe haven: a study of violence exposure in an urban community

    Journal of the American Academy of Child and Adolescent Psychiatry

    (1995)
  • M Schwab-Stone et al.

    No safe haven II: the effects of violence exposure on urban youth

    Journal of the American Academy of Child and Adolescent Psychiatry

    (1999)
  • T.M Achenbach

    Manual for the child behavior checklist and 1991 profile

    (1991)
  • B.K Attar et al.

    Neighborhood disadvantage, stressful life events, and adjustment in urban elementary-school children

    Journal of Clinical Child Psychology

    (1994)
  • C.C Bell et al.

    Community violence and children on Chicago's southside

    Psychiatry: Interpersonal and Biological Processes

    (1993)
  • S.L Berman et al.

    The impact of exposure to crime and violence on urban youth

    American Journal of Orthopsychiatry

    (1996)
  • M.W Berton et al.

    Exposure to violence and post-traumatic stress disorder in urban adolescents

    Adolescence

    (1996)
  • E.A Blechman et al.

    Prosocial coping by youth exposed to violence

    Journal of Child and Adolescent Group Therapy

    (1994)
  • S Boney-McCoy et al.

    Psychological sequelae of violent victimization in a national youth sample

    Journal of Consulting and Clinical Psychology

    (1995)
  • S Boney-McCoy et al.

    Is youth victimization related to trauma symptoms and depression after controlling for prior symptoms and family relationships?. A longitudinal prospective study

    Journal of Consulting and Clinical Psychology

    (1996)
  • N.K Bowen et al.

    Effects of crime and violence in neighborhoods and schools on the school behavior and performance of adolescents

    Journal of Adolescent Research

    (1999)
  • D Cicchetti et al.

    Toward an ecological/transaction model of community violence and child maltreatment: consequences for children's development

    Psychiatry

    (1993)
  • M Cooley-Quille et al.

    Adolescents' exposure to community violence: sleep and psychophysiological functioning

    Journal of Community Psychology

    (1999)
  • J.L Crouch et al.

    Income, race/ethnicity and exposure to violence in youth: results from the national survey of adolescents

    Journal of Community Psychology

    (2000)
  • M Dempsey et al.

    “Approach” and “avoidance” coping and PTSD symptoms in inner-city youth

    Current Psychology: Developmental, Learning, Personality, Social

    (2000)
  • D.F Duncan

    Growing up under the gun: children and adolescents coping with violent neighborhoods

    The Journal of Primary Prevention

    (1996)
  • R.H DuRant et al.

    Factors associated with the use of violence among Black adolescents

    American Journal of Public Health

    (1994)
  • D.E Elze et al.

    The association between types of violence exposure and youths' mental health problems

    International Journal of Adolescent Medicine and Health

    (1999)
  • F Esbensen et al.

    Juvenile victimization and delinquency

    Youth and Society

    (1991)
  • F Esbensen et al.

    Family context and criminal victimization in adolescence

    Youth and Society

    (1999)
  • A.D Farrell et al.

    Impact of exposure to community violence on violent behavior and emotional distress among urban adolescents

    Journal of Clinical Child Psychology

    (1997)
  • A.C Fick et al.

    Growing up in a violent environment: relationship to health-related beliefs and behaviors

    Youth and Society

    (1995)
  • D Finkelhor et al.

    Victimization of children

    American Psychologist

    (1994)
  • R.A Finnegan et al.

    Victimization by peers: associations with children's reports of mother–child interaction

    Journal of Personality and Social Psychology

    (1998)
  • K.M Fitzpatrick

    Exposure to violence and presence of depression among low-income, African-American youth

    Journal of Consulting and Clinical Psychology

    (1993)
  • K.M Fitzpatrick

    Fighting among America's youth: a risk and protective factors approach

    Journal of Health and Social Behavior

    (1997)
  • K.M Fitzpatrick

    Violent victimization among America's school children

    Journal of Interpersonal Violence

    (1999)
  • D.J Flannery et al.

    Effective youth violence prevention

  • N Freudenberg et al.

    Coming up in the boogie down: the role of violence in the lives of adolescents in the South Bronx

    Health Education and Behavior

    (1999)
  • J Garbarino et al.

    Children in danger: coping with the consequences of community violence

    (1992)
  • J Garbarino et al.

    Coping with the consequences of community violence

  • N.A Gonzales et al.

    Family, peer, and neighborhood influences on academic achievement among African-American adolescents: one-year prospective effects

    American Journal of Community Psychology

    (1996)
  • D Gorman-Smith et al.

    The role of exposure to community violence and developmental problems among inner-city youth

    Development and Psychopathology

    (1998)
  • S Graham et al.

    Self-blame and peer victimization in middle school: An attributional analysis

    Developmental Psychology

    (1998)
  • B.M Groves et al.

    Interventions with parents and caregivers of children exposed to violence

  • N.G Guerra et al.

    A cognitive-ecological approach to the prevention and mitigation of violence and aggression in inner-city youth

  • Cited by (128)

    • The social norms approach as a strategy to prevent violence perpetrated by men and boys: A review of the literature

      2022, Engaging Boys and Men in Sexual Assault Prevention: Theory, Research, and Practice
    • Hyperarousal and hypervigilance in African American male adolescents exposed to community violence

      2020, Journal of Applied Developmental Psychology
      Citation Excerpt :

      Within some urban communities, reported prevalence of exposure to violence is as high as 97% (McDonald, Deatrick, Kassam-Adams, & Richmond, 2011). Among African American youth in urban communities, violence exposure is significantly higher for boys (e.g., Lambert et al., 2012; Salzinger, Feldman, Stockhammer, & Hood, 2002; So, Gaylord-Harden, Voisin, & Scott, 2018; Thomas, Caldwell, Jagers, & Flay, 2016), as boys are often granted greater autonomy than girls (Cunningham, Mars, & Burns, 2012). Across studies, estimates of exposure range from 50% to 96% of African American boys, with many exposed more than once during adolescence (Fehon, Grilo, & Lipschitz, 2001; Gaylord-Harden, Cunningham, & Zelencik, 2011; Gorman-Smith, Henry, & Tolan, 2004).

    View all citing articles on Scopus
    View full text