|Home | About | Journals | Submit | Contact Us | Français|
The effects of exposure to direct and vicarious political, family, and community violence on the adjustment of 625 six-year-old black South African children was examined. Ambient community violence was most consistently related to children's psychosocial outcomes. Resources in the form of individual child resilience, maternal coping, and positive family relationships were found to mitigate the adverse impact in all the assessed domains of children's functioning.
In April, 1994 South Africa made a peaceful transition from an oppressive racist regime to a multiracial democracy. The path to this transition, however, was anything but peaceful. Under the prior Apartheid regime, politically motivated violence, particularly against blacks, was a salient feature of the social and political landscape. The wave of political violence had subsided by 1994, but the overall level of violence since then has not diminished. Consequently, concerns about threats to public safety from criminally and economically motivated violence continue to dominate public discourse. According to statistics reported by the South African police, sexual and property crimes, and crimes involving interpersonal violence have continued at levels at least as high, if not higher, than during the Apartheid era (Barbarin, Richter, de Wet, & Wachtel, 1998). In 1995, South Africa had the fifth highest murder rate among the countries tracked by Interpol, and more than half of all trauma cases admitted to hospitals in urban centers resulted from violent incidents involving weapons (Meumann & Peden, 1997). As Deputy President Thabo Mbeki underscored in an address to parliament, the threat posed by violence to the emotional well-being and the health of South African citizens: “Ongoing political and criminal violence are destroying the socioeconomic as well as psychological life of many communities. The statistics [on violence] represent a scale of human suffering and wretchedness which by any standard is impermissible.” Nowhere in South Africa is the burden of violence more onerous than among young children growing up in urban areas (Dawes, Tredoux, Feinstein, 1989).
The diverse sequelae associated with violence have been well documented. For example, in a study of urban African-American children, Osofsky, Wewers, Hann, and Fick (1993) reported that exposure to violence, witnessed or directly experienced, was related to both children's self-reports of distress and parents' observations of distress symptoms in their children. The observed symptoms included loneliness and sadness, loss of desire for amusement, daydreaming, inattention, disrupted sleep, nightmares, easy perturbation, intrusive disturbing and imagery, separation anxiety, and fear of death. Children may cope with the fear and loss associated with violence by restricting their activities, pretending not to care about anything, chronic worry about safety, anxious attachment to mothers, imitative aggressive play, and counter-phobic displays of bravado, (Osofsky, Wewers, Hann, & Fick, 1993). The effects of violence may also spill over into relationships with peers in the form of emotional withdrawal and aggressive behavior, both of which may contribute ultimately to rejection by those peers (Cooleyquille, Turner, & Beidel, 1995). The frequent report of fears, difficulty concentrating, reliving or re-enactment of distressing incidents, heightened arousal, irritability, anger, fear of being alone, nightmares about separation, and recurrent distress that is characterized by crying, tantrums, misery, emotional numbness, apathy, or social withdrawal have led some investigators to conceptualize children's response to violence in terms of post-traumatic stress disorder (Pynoos et al., 1987). However, the evidence suggests a broad range of dysfunction that includes depression, conduct problems, attention deficits, social maladjustment, and academic difficulties (Attar, Guerra, & Tolan, 1994; Gorman-Smith & Tolan, 1998).
Much of the research on the effects of violence has compared forms of exposure. These can differ in mode (e.g., direct victimization, indirect—or vicarious—exposure, witnessing, residence in dangerous communities) and in locus (e.g., family versus community) (Horn & Trickett, 1998). Direct experience of violence occurs when an individual or household is victimized or placed in immediate danger. Indirect, or vicarious, experience occurs when an individual is a witness to such incidents of violence as assault or abuse, or is subject to a general ambience of danger or residue of fear created by high levels of conflict and violence in a community. Exposure to ambient violence may not be tied to specific incidents or encounters, it can result from a generalized conviction that the probability of harm is high. Surprisingly, the available evidence suggests that the impact of violence on children does not depend on mode of exposure. Direct and vicarious exposure, as indexed by victimization and witnessing violence, have virtually the same correlation (.28 and .30, respectively) to children's distress (Richters & Martinez, 1993) Why the presumably weaker and less threatening vicarious exposure results in the same level of distress as direct victimization is puzzling. Perhaps the physical harm, trauma, or loss of direct victimization has less impact than do other features of violence exposure; for example, cognitive processing that involves altered expectations about degree of control and vulnerability may arise just as easily from vicarious as from direct violence exposure.
While direct and vicarious modes of exposure to violence do not seem to differ in impact, its effects are known to vary with how close the target of violence is to the child, being greater when the violence is against a family member, friend, or acquaintance than when it is against a stranger. For example, Osofsky and colleagues (1993) reported that adolescents exhibited psychological distress only when the incident of community violence they witnessed involved someone they knew. Similarly, Pynoos, Nader, Frederick, Gonda, and Stuber (1987) reported that the most severe reactions to a kindergarten sniping incident were among children emotionally close to those killed, and that physical proximity also contributed to severity of emotional distress, independent of the closeness of the social relationship. These findings suggest a principle of social propinquity: expectations of directly experiencing violence increase when violence happens to someone with whom a child has a relationship or identifies. The closer the connection with the victim, the greater the adverse impact of the incident (Boney-McCoy & Finkelhor, 1995). Thus, it is not surprising that violence in the family has more detrimental effects on children than does violence in the community (Osofsky et al., 1993). In fact, when it is severe, family violence has more negative sequelae for children than any other form (Horn & Trickett, 1998).
Differences in the effect of violence may depend on information gleaned or interpretations made by the children about the likelihood of future harm. The process of identification and empathy, may also explain why an acquaintance or family member's exposure to violence can have as powerful and emotionally disruptive an effect as direct victimization (Farrell & Bruce, 1997; Taylor, Zuckerman, Harik, & Groves, 1994). In this way vicarious and direct exposure would be equally capable of engendering an expectation of future harm.
If the effects of exposure to violence emanate from expectations about future loss and heightened vulnerability to violence, then they should be responsive to resources that are known to enhance coping with other forms of stress (Berman, Kurtines, Silverman, & Serafini, 1996). If so, the sense of threat and danger from divers experiences of violence should be modifiable by personal resilience, emotion regulation, spirituality, and a supportive family environment (Wallen & Rubin, 1997). Such resources can help children manage fears and sadness in the face of life disruptions and personal loss by enabling cognitive reappraisal and re-interpretation, and transcendent ideology (Garbarino & Kostelny, 1996). Just as family stress can be a risk factor, family harmony and support can be protective factors for children exposed to violence (Paschall & Hubbard, 1998). Moreover, because children often emulate and take their cues from their mothers, maternal capacity to manage distress may facilitate children's coping.
In South Africa, children have been exposed to many different forms of violence: political, familial, and community. They have been exposed both directly, through victimization, and vicariously, through ambient community danger. Consistent with previous research, the study reported here hypothesized that exposure to violence would adversely affect children's psychosocial adjustment (behavioral, emotional, social, and academic) but that these affects would be moderated by coping resources (spirituality, family support, child resilience, and maternal coping) available to the child.
Children and their parents were recruited as part of the Birth to Ten (BTT) project, a longitudinal study of the effects of urbanization on physical and psychological development in South Africa. Although the larger study from which this sample was drawn consisted of both black and white children, the data reported here are from a subsample of 625 black children whose families were interviewed extensively when the sample children were five years old. With respect to population group, 91% of the respondents were classified as African, and the remainder as colored or Asian. A detailed demographic description of this 1996 sample and of the procedures used to gather data can be found elsewhere (Barbarin & Khomo, 1997; Barbarin & Richter, 1999).
By ordinance, local health authorities in South Africa maintain official registers of all births in their area. The study sample consisted of children born during a seven-week period in 1990 to women with a permanent address in Soweto-Johannesburg. Enrollment took place over a 15-month period, beginning prenatally (at the clinic visit coinciding with a gestational age of 26 weeks) and extending up to 12 months after delivery. Mothers (or other primary caregivers) were contacted in hospitals, at clinics, and in their homes. They were informed about the longitudinal birth cohort study, and their consent to participation in interviews and assessment of their children and their family life was requested. Nonenrollments occurred because of still births or early deaths of babies, adoption out of the home, outright refusal (which was seldom), mother's death, infants being sent away to be cared for by relatives in rural areas, untraceable addresses, communication problems, and unavailability of the mother for interview because of the nature of her employment.
Follow-up of children and their families was assisted by such procedures as marking BTT children's health records to facilitate tracing; regular publicity of the study through the media, including a one-year birthday party hosted by the mayor of Johannesburg; gifts (including BTT tee shirts, rubber balls, and Polaroid photographs) to the children at particular times; annual BTT calendars and a biannual BTT newsletter given to families, the latter containing news and appropriate results of the study and advice on the health of children in the age range of the BTT sample. In addition, participants could call a toll-free number at the BTT office for information about the study; to enquire about children, child health, and child care; and to get referrals to local health and social services.
For each of the scheduled interviews (antenatal, delivery, six months postnatal, and at yearly intervals thereafter—five years, to date), participating families were first notified by mail, then telephoned for scheduling of the mother and child for personal interviews at one of the local health facilities. When parents could not be reached by post or telephone, or come to the clinic, interviewers visited their homes in the evening or at weekends. Two hours were allotted for completion of each interview and measurement the child's physical and psychological development. Caregivers' cost of public transportation to the interview site was defrayed, and food and beverages provided for parent and child. More often than not, mothers accompanied their child, sometimes both parents came. When a mother could not get time off from work, interviews were conducted with the father or other primary caregiver.
At the interview, the child was measured anthropometrically and cognitively. Parents were asked to provide information about the family's social and economic status, child-rearing expectations and practices, and health experiences (illnesses and injuries). Ratings of child behavior were also obtained through parents' reports.
Interviews were conducted by seven trained, multilingual, community residents. The language used in the interviews was adapted to the colloquial speech of day-to-day life. Once interviewers had received training in the purpose of the interview and the concepts underlying each question, they worked toward consensus on the best phrasing for each question, and this was then used consistently by all interviewers.
Interviews, questionnaires, archival data, and key informant ratings were used to assess violence.
Seven community experts employed a Q-sort procedure to categorize each of the communities from which children were drawn for the study. These experts included people with a broad comparative knowledge of neighborhoods (e.g., taxi drivers and community health workers) with knowledge based on long-term residence and on the kind of work that required them to move around the selected communities. They ranked communities on the spectrum of danger-safety. In effect, the ratings were subjective appraisals of the likelihood of being a victim or indirectly experiencing such violent events as physical threat or intimidation, shooting, stabbing, injury, sexual assault, murder, mugging, car hi-jacking, forced entry into a home, or robbery. They sorted communities independently into five categories, from 1 (most safe) to 5 (most dangerous). Interrater agreement was acceptable (Cohen's Kappa=.70). Kappa provides a very conservative estimate of intercoder agreement on multilevel ratings by correcting for chance agreement. Disagreements were resolved by averaging the ratings. Rating scores correlated significantly with documented incidents of political violence (Pearson's r=.24, p≤.01) and with mothers' personal experiences of violence as reported in the parent interviews (Pearson's r=.04; P<05).
Archival data on political violence, gathered from police reports, monitoring agencies, and press accounts, were obtained from the Human Rights Commission (1996), an independent watchdog group located in Johannesburg. These data are similar to those reported by the South Africa Institute of Race Relations (1996) and must be interpreted with the caveats that a) the data are incomplete because many incidents go unreported and unmonitored, and b) objectivity and adequate corroboration cannot be claimed for the reports. However, the data are useful because they portrayed trends in the nature, locus, and prevalence of violence, and encompass divers situations: conflict between political groups; among shack or hostel dwellers and other township residents; between blacks and whites; in trains, taxis, and schools; and in mass actions such as strikes and boycotts. Data files on incidents of violence were obtained for the years 1990 through 1994. They included the number of incidents and persons critically injured, killed, and arrested, disaggregated by community. Proximal violence scores represent incidents occurring within a suburb or community, while distal violence represents those occurring in surrounding areas.
The experience of violence by members of a child's household was assessed within the six-month period preceding or following the child's birth in 1990, and in 1995, when the child was five years old.
To interview questions about violence of all three types, respondents indicated Yes or No. For community violence, the questions were: “Have you or a family member witnessed a violent crime?” and “Has any household member been in danger of being killed?” For political violence, the questions related to injury or death as the result of political violence. For family violence, they concerned being hit or beaten by a domestic partner.
The Family Relations Scale (FRS) is a brief self-report questionnaire assessing three dimensions of family life: family conflict, satisfaction with family life (particularly family support), and spirituality (Barbarin, 1994). Family conflict items assess the extent to which family members engage in violence and physical conflict with one another (α=.78). Family satisfaction items reflect favorable appraisals of the quality of family life and relations, particularly with respect to the need for love, acceptance, and support (α=.87). Family spirituality items assess quality of family life as characterized by religious faith and use of prayer in family rituals and in dealing with stressful situations (α=.79). FRS scales utilize a four-point rating (Not, Sometimes, Often, or Always true). Support for the construct validity of the FRS has been obtained with several independent samples of families of African descent (Barbarin, 1998). Using principal components analysis, the factor structure has been replicated with African-American, South African, and Ugandan families. Data on family violence, since they rely on self-reports, are likely to be underestimates because of the effects of social desirability and possible fear of retribution.
Attention, aggression, anxious-depressed, and somatic complaints are subscales of the Child Behavior Checklist (CBCL) (Achenbach & Edelbrock 1983), a widely used and well-standardized parent-report measure of behavioral and emotional problems in children aged 4–18 years. Evidence for the validity and reliability of the scales are re-ported in Achenbach (1991).
Oppositional behavior is a subscale of the Behavior Problem Index. (BP1) (Zill, 1985) developed for the National Health Interview Survey (Center for Health Statistics, 1981). Items assess the extent to which the child is noncompliant and highly strung. Evidence for construct validity and reliability of the BPI scales is substantial (Bussing, Halfon, Benjamin, & Wells, 1995; Peterson & Zill, 1986).
Resilience is a subscale of the South African Child Assessment Schedule (SACAS) (Barbarin & Richter, 2001) a structured maternal-report questionnaire based on a three-point frequency scale. Items assess personal adaptability and frustration tolerance. Construct validity is supported by principal components factor analyses. Estimates of internal consistency are in the acceptable range (α=.64).
Academic motivation is adapted from the good student subscale of the Health Resources Inventory (Gesten, 1976), a previously validated measure of social competence in young children involving level of independence, performance, motivation, and persistence in learning tasks. Items assess how well the child performs under less than optimal conditions, such as distraction and minimal adult supervision.
Maternal distress was assessed using the Pitt Depression Inventory (PDI) (Pitt, 1968), a self-report questionnaire patterned after the Hamilton Depression scales and used widely in Great Britain and its former colonies. Its 24 items capture the primary cognitive, affective, behavioral, and somatic features of depression and anxiety through ratings of dysphoric mood, irritability, and loss of interest in previously pleasurable activities; disturbance of appetite and sleep; and diminished energy, activity, and life satisfaction. For this South African sample, the range of scores was 0–42 (M=4.1, SD=7.4). The PDI discriminates between clinical and nonclinical cases, and estimates of reliability are in the acceptable range (α=.74).
A series of multiple regression analyses was used to test the model explaining the effects of violence on children and their mediation by maternal coping. Missing data were handled by list-wise deletion of cases for each analysis. To avoid violating the regression assumptions, outlier cases were removed, and variables with skewed distributions (political and family violence, and oppositional behavior) were transformed. Standard scores were computed for oppositional behavior, and log transformations were used for political violence. An initial series of regression analyses tested the relationship of demographic variables to the child outcomes employed in the study. Gender was entered as a dummy variable (1=male; 0=female). Education (number of years) and occupational status ratings were entered as continuous variables. When child's gender, mother's education and occupation, and household socioeconomic status (SES) were used as independent variables with child outcomes as dependent variables, none of the regressions produced significant results.
Next, the relationship of types of violence to child outcomes were tested. To facilitate testing of the mediating effect of maternal distress, four factorial interaction terms were computed by multiplying the maternal distress score by the four types of violence. A hierarchical or stepwise method was used for the regression analyses. Variables were entered in two stages. The four indicators of violence (political, community, and family violence, and victimization) were entered in step 1. In step 2, the measures of coping resources (individual resilience, family satisfaction and family spirituality) were included, along with an interaction term between maternal distress and each of the violence indicators that were significant hi step 1. For each stage, the criterion for inclusion in the model was a p-level of .05. List-wise deletion of cases from all analyses was again used if any one of the variables used in the regression analyses was missing from a case.
The relationship of violence to maternal distress was tested using simultaneous multiple regression in which expert ratings of community danger, incidents of political violence reports of family violence and direct victimization of family members were regressed onto mother's scores on the Pitt depression scale. The variables for violence were entered in a single step. The effect was small (2% of total variance) but statistically significant, F(4, 621)=4.0, p<.01. Maternal distress was significantly associated with higher levels of community danger and family violence (see table 1).
Pearson product-moment correlations were computed for the six child outcome measures (see table 2). Behavior problems (attention, opposition, and aggression) were highly correlated with one another and moderately correlated with emotional and somatic complaints. Anxious-depressed was not related to somatic complaints but was directly related to academic motivation. Academic motivation was inversely correlated with somatic complaints and oppositional behavior, but positively correlated with anxiety-depression and aggression.
Pearson product-moment correlations were computed to determine relationships among the intended independent variables—resilience, maternal distress, spirituality, and satisfaction with family life—in the multiple regression analyses. They were not significantly related to one another or to the violence indicators. Zero-order correlations were computed to examine how closely types of violence were related to child outcomes; the correlations among types of violence, resources, and psychological adjustment in children are presented in table 3. Ambient community violence was significantly and directly correlated with attention problems, aggression, and anxious-depression. Victimization was significantly and directly correlated with oppositional behavior. Political violence was not related to any of the measured child outcomes. Family violence was correlated directly with attention and aggression, and inversely with academic motivation. Family satisfaction was significantly and directly correlated with academic motivation and inversely correlated with oppositional behavior. Spirituality was correlated inversely with aggression and directly with academic motivation. Children's individual resilience was correlated inversely with attention, oppositional behavior, and aggression.
Table 4 presents the results of the hierarchical linear regression testing the differential effects of the four forms of exposure to violence on children's psychological adjustment and the extent to which these were mediated by coping resources such as resilience and family satisfaction. Significant regression models were computed for each of the child outcomes. However, with the exception of academic motivation, the total variance explained by the final regression model for each outcome was under 10%. For attention problems, a final model was produced in three steps: in step 1, ambient community violence was the best predictor of such problems; in step 2, victimization was added, and contributed to significantly to the pre-diction; in step 3, community violence and victimization were retained but the interaction term of maternal distress and community violence also contributed uniquely to the prediction of attention difficulties.
Only one step was required to produce a model for oppositional behavior, the level of which was predicted by resilience. Victimization and satisfaction with family life had significant zero-order correlations with oppositional behavior until the effect of resilience was accounted for, when they no longer had a significant relationship. The regression model for aggression was produced in three steps: in step 1, community violence was the lone significant predictor; in step 2, family violence added significantly to the prediction of aggression; in step 3, community and family violence continued to account for aggression but the term for the interaction between maternal distress and community violence was significant. As in the case of attention problems, maternal adjustment moderated the effects of community violence on children.
With respect to emotional functioning, child scores on the anxious-depressed scale of the CBCL were significantly predicted by community violence. However, in step 2, the significant interaction term suggested that effect of community violence on anxious-depressed scores was again moderated by the level of maternal distress. One step was required to reach a final model for children's somatic complaints, for which resilience was the single significant predictor. Academic motivation was modeled in two steps: resilience emerged in step 1, suggesting that the higher the level of resilience the greater the academic motivation; in step 2, family spirituality contributed, along with resilience, to prediction of academic motivation. To test for possible gender and SES differences in these relationships, a final set of regression models was computed that included gender, maternal education, household SES, and violence measures. This attempt to test the unique contribution of demographic variables to child outcomes resulted in models identical to those from which demographic variables were omitted; i.e., none of the demographic variables was significant.
The results of the present study of South African children confirm previous findings (Richters & Martinez, 1993) that exposure to ambient and vicarious violence produces effects parallel to those observed when the violence involves direct victimization. It is noteworthy that the effects of violence on psychological and academic functioning were found to be independent of gender and socioeconomic status. Males and females, the economically advantaged and disadvantaged, all evidenced similar difficulties in the face of violence.
Research on American children and adolescents has suggested that both direct and vicarious exposure to violence produce an array of psychological effects that intensify as the social and relational distance between the child and the victim decreases. However, the effects of social propinquity to the victim were not as strong in the current study findings. Surprisingly, community violence was more strongly predictive of child outcomes than family violence. The divergence between the BTT and the U.S. findings is probably attributable to differences in the relative intensity of family and community violence, and in the age range of the samples. Many U.S. studies showing strong effects for family violence involved very serious physical injuries, such as those from stabbing, shooting, and severe beating, whereas milder forms of violence, such as hitting, made up the index of family violence in the BTT study. Conversely, the community violence to which the current sample was exposed was arguably more intense, given the high levels of political conflict and social upheaval in South Africa at the time of the study. In addition, much of the research on violence in the U.S. included children aged seven and older, while the South African study was limited to children under age seven.
The principal effects of violence in these young children occurred in the domains of attention, aggression, and anxiety-depression. Community violence emerged as the most consistent predictor of adverse child outcomes (and was also predictive of maternal distress). The propinquity principal received modest support, in that family violence predicted child aggression, while victimization had an adverse effect on attention. Surprisingly, political violence showed no relationship to children's psychological and academic functioning. It is important to note that these results confirm the significant role played by individual and family coping resources in mediating the problems of violence. On three of the six child outcomes measured in this study (oppositional behavior, somatic complaints, and academic motivation), coping resources, particularly child resilience, mitigated ad-verse effects of violence. These relationships do not appear to depend on the child's gender or the SES of the household.
Mothers themselves were not immune to the effects of violence. Community and family violence were associated with high levels of emotional distress in mothers. However, their ability to regulate their distress had important implications for their children's response to community violence. When mothers exhibited low levels of distress, community violence was less likely to be associated with attention, aggression, and anxiety-depression problems in children. This finding hints at another way in which family life may protect children from the effects of adversity. Maternal distress may constitute an index of the broader emotional climate in the family unit. This interpretation avoids the possibility of inadvertently (and inaccurately) blaming mothers for the consequences of violence for children. The impact of such negative life events as violence is largely dependent on what the family conveys about these events, its reactions to and interpretations of them. Moreover, the child's experiences are dependent on what the family allows to enter or permeate the child's psychological space. Accordingly, the child's experience of violence is largely a function of the family's ability to serve as a barrier to disturbances, and this may depend on the quality of family relationships and the amount of support and other social resources available to the child through the family. Of course, this buffering effect becomes less and less potent as the child matures and expands the range of social contacts.
The total variance in children's psychosocial adjustment accounted for by violence was small. This suggests that the impact of violence on children is not inevitable and may be attenuated by a host of factors not measured in the BTT study. Although this research focused on the detrimental effects of adverse conditions associated with violence, it is important to note that many children showed no such effects, and that many thrived in spite of the violence. “Like research conducted on child abuse, poverty, and other forms of adversity, dangerous and violent conditions offer important lessons about subjects such as risk and resilience and can teach us a great deal about psychological mechanisms such as stress, coping, and support” (Ladd & Cairns, 1996, p. 17). An important implication of the current results is that the negative consequences of exposure to violence can be moderated by family functioning and social relations (Richters & Martinez, 1993). Specifically, individual child resilience, and family religiosity and support can be construed as coping resources related to such favorable adjustments as high academic motivation and low levels of oppositional behavior and somatic complaints.
Will the adverse effects of violence and the mediating role of coping resources observed at age five prevail when the children in this study are older? Can maternal coping mitigate the effects of violence as powerfully at later ages? What protective role might be played by a supportive, nurturing school environment and by the experience of success and competence in school? The present data do not directly address these issues, but our knowledge of the children and the setting provide a partial basis for speculation. A great deal of research already exists on the effects of violence on older children; it suggests that if the violence continues the effects will persist, and if it ceases they will diminish. As this study suggests, family coping resources provide a degree of inoculation against the effects of violence. As children mature and expand their social contacts through the neighborhood and school, then peer groups and adults (such as teachers) might play an increasingly important role in the children's ability to resist the effects of violence.
This gives room for optimism about the prospects for successful intervention with high-risk groups. The range of possible interventions is broad. Most important is the effort to rebuild support and reciprocal caring within community life. Community life in South Africa needs an infusion of the traditional value of Ubuntu, the obligation to show mutual concern and provide for the needs of others. The effectiveness of family-level support—delivered, for example, though home visiting—has been demonstrated (Garbarino, Kostelny, & Barry, 1998). The efficacy of community-based, family-focused interventions underscores the value to children of expanding those programs that promote positive adaptation and mental health in communities where children are likely to be exposed to violence.
Recently, schools have become a locus of efforts to reduce violence using programs that attempt to promote school environments in which children can feel safe, prosocial methods are used for resolving conflict, and staff are able reduce reliance on power-assertive control strategies. Schools can play an especially important role by incorporating into their curriculum modules that promote favorable self-appraisal and enhance children's capacity for self-reflection and coping skills. Such programs challenge extant norms about aggression as the principal means of dealing with problems (Taylor, 1994).
In developing programs to assist children, we must be mindful of a key point demonstrated in the present study, i.e., the critical mediating role of parental figures. The lesson here is that programs of clinical outreach providing emotional support and guidance to parents can contribute significantly to the well-being of children in risky and dangerous situations.
Because the economic inequality fostered by Apartheid continues to be at the heart of the problems of violence in South Africa, any successful effort to reduce violence must also attack economic inequality. As promising as violence prevention programs may be, they cannot be successful in the absence of widespread economic transformation.