ADVERSITY AND PSYCHOSOCIAL COMPETENCE OF SOUTH AFRICAN CHILDREN

Abstract

Black children in South Africa commonly experience low socioeconomic status and community violence. Parents (N=625) in a longitudinal study of urbanization responded to structured questionnaires related to resilience, affability, maturity, and school readiness of their six-year olds. SES was found to have an inverse and linear relation to competence at age six; the relationship to violence was curvilinear, with children from moderately safe communities achieving better outcomes than those from very safe or very unsafe ones.

In the wake of important political changes and social transformations under way in South Africa, increasing attention is being paid to the residual effects of apartheid as expressed in economic inequality and community violence. Adverse conditions that were reinforced under the apartheid regime, such as household poverty and community violence, subject children to multiple daily stressful experiences that can seriously compromise psychological functioning and development (Barbarin, Richter, de Wet, & Wachtel, 1998; DeLongis, Folkman, & Lazarus, 1988). For example, Robertson & Berger (1994) observed that children growing up in dangerous and poor areas of South Africa displayed significantly more stress-related symptoms than did children living in communities that were socially and economically advantaged.

Until now, the psychological effects of community violence on children have been more consistently demonstrated in research than have the effects of low socioeconomic status (SES) (Barbarin & Richter, 1998). Moreover, the effects of poverty and violence are more pronounced for conduct problems and school achievement than they are for emotional difficulties (McLeod & Shanahan, 1993; McLoyd, 1998). Notwithstanding these differences, the accumulated evidence linking low economic status and community violence to cognitive impairments, externalizing disorders, and problems of emotional dysregulation is compelling (Cairns & Dawes, 1996; Duncan & Brooks-Gunn, 1997; McLoyd, 1998; Richters & Martinez, 1993).

In contrast, very little research has explored the relationship of adversity to the development of social competence, and claims about the nature of such relationships must consequently rest on uncertain grounds. In the absence of empirical data and a convincing theoretical rationale, at least three alternative relationships between adversity and competence can be argued: no significant effect, an inverse effect and a direct positive effect. Arguments that low SES and community danger have no relationship to social competence and academic adjustment rest on the assumption that adverse conditions are as likely as propitious conditions to give rise to social competence.

Arguments for an inverse relationship between adversity and competence hold that children's social and cognitive development are stunted as a result of the stress and adaptational patterns associated with household poverty and community violence. In propitious environments, characterized by safety and material adequacy, the balance of resources and needs engenders a basic trust in and reliance on others, freeing children from preoccupation with requirements of physical survival to invest their creative energies and emotional capital in the development of empathetic and cooperative social relations. In contrast, poor children face high adaptational demands with limited resources, receive less cognitive stimulation, exhibit premature self-care and self-reliance, and ultimately become unresponsive to direction from adults. Moreover, parents living in communities with high levels of violence may, in an effort to protect their children from danger, adopt practices that are physically restrictive and dependence-inducing (Jarrett, 1995; Richters & Martinez, 1993). In this pattern of social adaptation, children over-react to provocation from others, perceive threat where none exists, and approach social and academic situations with suspicion, fear, shyness, or social withdrawal (Steinberg & Dodge, 1983; Dodge & Price, 1994).

Claims for a positive relationship between adversity and competence assume that poverty and violence challenge children in ways that promote social competence and academic motivation (Chesler & Barbarin, 1987). In this view, a mastery orientation is just as likely to arise from hardship and community danger as demoralization; the strain of an adverse environment elicits resourcefulness that might not be required or developed in its absence. Accordingly, growing up in adversity presents children with opportunities for mastery and self-sufficiency not available to children in more advantaged circumstances. Early confrontations with adult challenges, can force them to be more resourceful and avoidant in the face of danger. Thus, they may mature more quickly than their more advantaged counterparts. Similarly, economic disadvantage may motivate concentration on academic tasks as a means of upward mobility, a principal way of escaping the hardships of material poverty. Evidence from studies of children in families challenged by life-threatening illness has provided strong support for this argument (Barbarin, 1990). Children challenged by such adversity evidenced high levels of maturity, an enhanced capacity for empathy, resourcefulness and long perspective, and a focused and purposeful engagement in academic pursuits (Barbarin et al., 1995).

The present study examines the relationship of economic status and community danger to early manifestations of social competence and academic adjustment in young children. The data are from a longitudinal study of the effects of urbanization on children's development in South Africa. The longitudinal study compared children in poor households and violent communities with children growing up in relatively propitious circumstances on such dimensions of social and academic functioning as resiliency, affability, maturity, and school readiness. Although modest empirical support can be marshaled for the competing claims of adversity's positive and negative effects on social competence in children, no clear pattern has emerged on which to base specific predictions or hypothesize outcomes.

METHOD

Data Collection

The longitudinal study, from which selected results are presented here, is an extension of the Birth-to-Ten Project (BTT) (Richter, Yach, Cameron, Griesel, & de Wet, 1995), a cohort study of children born in 1990 in the Johannesburg-Soweto metropolitan area. The data were collected between January and April 1996, when the children were 5–6 years of age.

Interviews were conducted by five trained, multilingual, same-race community residents, who were selected because of their experience with earlier BTT data collection, their familiarity with BTT families, and the high quality of their prior work. The interviewers completed a two-week training course that focused on concepts of childhood, developmental psychopathology, and parent-child socialization. Training also included supervised practice in administration of the interview. Each interviewer was observed while administering the questionnaire, and corrective feedback was provided. Once trained, interviewers suggested ways to improve the clarity and precision of questions in the interview. Each was then assigned to several suburbs within the black townships.

Most of the interviews with mothers were conducted in their homes, except when they preferred an alternative venue, such as the local health clinic. In the initial stage of the project, the questionnaires were designed in English and then translated, back-translated, and checked in the three other major languages (Zulu, Sotho and Afrikaans) spoken in South Africa. Families speaking languages other than these were interviewed by a native speaker in, for example, Tswana. In these cases, the interviewer worked from a translation of the closest related language and translated the questionnaires for the families. However, because of the close proximity and frequent interactions among speakers of the 11 most common languages in South Africa, the colloquial speech used by most urban Africans is not a pure form of any of the major languages. For this reason, the language used in the interviews was adapted to conform to colloquial speech of day-to-day life. Once they had received training in the purpose of the interview and the concepts underlying each question, multi-language interviewers reached consensus about phrasings for the questions for each language group and used them consistently.

Interviewers read questions aloud to the mother and recorded her responses, along with notes, comments, and observations about the parents' behavior. Interviewers asked respondents to clarify any discrepancy. Items on economic status were placed at the end of the interview. On average, interviews took approximately 75 minutes to complete. Families were not reimbursed for the interviews; however, they were given a BTT calendar. Parents interviewed at health clinics were reimbursed for their travel costs and provided with refreshments.

Whenever serious problems were detected in the behavior rating scales, parents were offered the opportunity of a diagnostic interview, after which a decision about referral for services was made. In some instances, interviewers uncovered problems such as abuse or serious learning difficulties that were emerging in school. These were referred to social workers or clinics sponsored by the health service.

Sample

The sample was primarily African (90.5%), with Coloreds and Indians making up the remainder. As can be seen from the language divisions in table 2, the largest African ethnic group was Zulu, followed by Sotho. The typical children in the BTT longitudinal sample were born to relatively young, single, working mothers whose median age was 25 years. Of these mothers, 55.9% had never been married, 9% were widowed or divorced, and 35.1% were married. Education for 34.4% had ceased before high school, while 20.5% had completed high school. The majority (79.4%) were employed outside the home, mainly as domestic workers, at very low wages. Biological fathers lived with their children in about 34% of the cases. Although nearly three out of four children in the sample had only one biological sibling, or none, the typical child lived in a multigenerational household that held a median number of seven persons. About 51% of children in the low SES group regularly experienced hunger.

Table 2.

DEMOGRAPHIC VARIABLES BY SOCIAL AND PHYSICAL SECURITY GROUPS

	SES			COMMUNITY VIOLENCE
VARIABLE	LOW (N=210)	MIDDLE (N=214)	HIGH (N=201)	VIOLENT (N=197)	SOMEWHAT (N=289)	SAFE (N=139)
SES
Males (%)	48.6	52.3	47.8	48.2	50.9	48.9
Mother
Education
None	2.9	0.5	0.0	1.0	1.4	1.7
Elementary	24.2	9.0	4.8	11.8	14.1	10.3
Middle School	30.1	24.1	12.6	22.1	24.9	17.6
Some H.S.	36.8	48.2	42.4	46.7	40.4	41.1
H.S. Graduate	5.8	18.4	40.9	18.5	19.3	30.1
Language
Sisi Zulu	41.8	42.3	31.7	49.5	28.6	44.5
Se Sotho	25.0	19.7	24.6	24.0	18.5	31.4
Tswana	12.0	16.9	15.1	12.2	19.9	7.3
Xhosa	7.7	5.6	5.5	8.2	4.5	7.3
Tsongan	6.3	4.7	3.0	1.0	7.3	4.4
Afrikaans	3.8	5.6	9.0	0.5	12.9	—
English	—	0.9	9.5	1.0	5.2	2.9
Other African	3.4	4.2	1.0	3.6	3.4	1.4
Child Hunger	50.5	10.7	1.5	27.4	16.3	22.3
No Household Partner	58.6	49.5	38.8	54.3	46.7	46.8

Sociodemographic data, disaggregated by social status and community violence groups, are presented in table 2. The total sample was almost evenly male and female, nor did the groups differ significantly on gender. As might be expected, the high SES group contained the highest proportion of mothers who had completed high school and lived with partners, as well as the smallest proportion of children who experienced hunger. The two safer groups were not significantly different on these factors.

Measures

Socioeconomic status

Estimates of household SES were derived from the Household Economic and Social Status Index (HESSI) (Barbarin & Khomo, 1997), a self-report measure of material resources (e.g., food, housing, durable consumer goods, utility expenses, and assets) and social capital (e.g., occupation, education, and marital status). These SES estimates are significantly correlated with the Hollingshead (1975) four-factor SES score. Approximately 40% of the population of South Africa is assessed as poor according to the government's Minimum Living Standards index. The study sample was therefore divided into three roughly equal groups—low, middle, and high social status—based on their standardized SES scores; the low-status group was living at poverty level.

Community violence

Rankings of neighborhoods or suburbs were made by local community experts using a Q-sort procedure. They sorted communities into five groups, from the most violent (1) to the most safe (5). Intercoder agreement was in the acceptable range (Cohen's Kappa=.70) and mean coder ratings correlated significantly with objective archival data on the number of killings, injuries, and arrests gathered and maintained by the Center for the Study of Peace and Reconciliation (1996) in Johannesburg, South Africa (Pearson's r=.24, p≤.01). Persons from communities receiving an average ranking in the extremely and somewhat violent categories (1 and 2) were collapsed into a single violent group; those ranked in the middle (3) were classified as somewhat safe; and those with the two highest safety ratings (4 and 5) were combined as a safe/secure group. Violence and SES are mildly correlated (Pearson's r=.15, p≤.001).

Social competence

This was rated on the South African Child Assessment Schedule (SACAS) (Barbarin, 1996), a structured interview composed of maternal self-report items on a three-point scale that yields four competence scale scores: affable (11 items), an index of social attractiveness and acceptance that includes such qualities as humor and demonstrating affection; resilient (seven items), an index of personal flexibility and frustration tolerance; independent/mature (four items), an index of autonomy and capacity for self-care; and school ready (five items), consisting of indicators (e.g., curiosity, knowledge of numbers, and ability to attend, follow directions, and share) selected by teachers as suggestive of preparedness for classroom learning (National Center for Education Statistics, 1993). Construct validity of the resilient, affable, and independent/mature scales is supported by principal components factor analyses using Varimax method of factor rotation with Kaiser Normalization of factor scores. The final solution produced three factors that converged in five iterations and accounted for 48% of variance. The range of factor coefficients for items loading on the affable scale was 33–76, for the resilient scale it 30–89, and for the independent/mature scale 31–50. A similar procedure for factor analysis of the school-ready items yielded a single factor that converged in three iterations and accounted for 58% of the variance. The range of factor coefficients for items loading on the school-ready scale was 38–70. Estimates of internal consistency are in the acceptable range (Cronbach's alpha=.62, .64, .63, and .81 respectively). As evidence of discriminant validity the three psychological competence scales (affable, resilient, independent/mature) were found to have low to moderate correlations with the highest correlation occurring between affable and resilient. The school-ready scale is significantly but moderately correlated with the psychological competence scales. Zero-order correlations among the psychological and academic competence scales, as well as SES and community danger, are presented in table 1. As evidence of convergent validity, competence scales were correlated with the Health Resources Inventory (Gesten, 1976; Hightower, 1986), a well-validated measure of social competence. The affable scale is highly correlated with the sociable scale (r=.64), the resilient scale with the gutsy scale (r=.68), and the independent/mature and school-ready scales with the good student scale (r=.79 and r=.39, respectively).

Table 1.

ZERO-ORDER CORRELATIONS: SOCIAL COMPETENCE SCALES, SES, AND COMMUNITY DANGER

VARIABLE	AFFABLE	RESILIENT	INDEPENDENT	SCHOOL	SES
Affable
Resilient	0.39***	—
Independent/Mature	0.07	0.03	—
School Ready	0.44***	0.38***	0.08*	—
SES	0.06	0.20**	0.08	1.13**	—
Community Danger	−0.06	0.02	−0.01	0.05	0.15**

^*p<.05;

^**p<.01;

^***p<.001.

RESULTS

On the whole, parental ratings on the SACAS competence ratings were overwhelmingly favorable. Parents described over 90% of their children as frequently affectionate, happy, relaxed, loving, well-liked by other children, having a good sense of humor, well-behaved in school, and appropriate in expressing their needs. They were also rated as frequently interested in people around them, interested in school work, feeling good about themselves, and able to adjust to changes.

However, interesting if predictable gender differences were found on several items. For example, girls were more frequently rated as fragile than were boys (14.6% vs. 8.4%), but as less often requiring restrictions to control them (24% vs. 33.8%). In general, data from these items suggested that boys were viewed as more robust, energetic, and difficult to manage, while girls were considered more emotionally expressive and socially engaged—a view consistent with portrayals of young boys and girls growing up in Western societies. Thus these data provide cross-national support from an African perspective of these well-established gender patterns.

The results of a Multiple Analysis of Variance (MANOVA) performed on the competence scales are presented in table 2. This analysis used community violence (violent, somewhat safe, safe/secure communities) and SES (low, middle, and high) as independent variables. The main effects for each of these factors and the interaction between violence and social status were tested. The multivariate F-test for violence, F(8)=10.30, p<.001, Eta squared=.063, and social status, F(8)=2.23, p<.023, Eta squared=.014, were both significant, but the interaction was not, F(16)=1.14, p=NS. As an alternative to the test of interaction effect for assessing the independence of the effects of SES and violence, a MANCOVA was calculated with community violence as a single independent variable, SES as a covariate, and the four social competence scales as dependent variables. This approach did not alter the relationships between violence and social competence found in the two-factor MANOVA.

Thus, at age six, significant differences were found in the social and academic functioning of children on the basis of economic status and community safety. The relationship between SES and competence was linear and direct but the relationship between community violence and competence was curvilinear. Children in somewhat safe communities had higher mean scores on the competence scales than did children from the least safe and the most safe communities. Means and standard deviations of the raw scores on the four competence dimensions for the nine groups are presented in table 3, along with the univariate F-tests of significance. The groups were formed by combinations of community violence and SES. For community violence, significant univariate differences were found for all the competence dimensions except affable; for SES, resilient and independent/mature were significant.

Table 3.

MANCOVA: INDEPENDENT VARIABLES=SES AND COMMUNITY SAFETY, DEPENDENT VARIABLES=SOCIAL AND ACADEMIC COMPETENCE

	LOW SES			MIDDLE SES			HIGH SES			UNIVARIATE F-VALUE
VARIABLE	V	SV	NV	V	SV	NV	V	SV	NV	Va	SESa	V × SEa
No Cases	73	97	40	79	89	46	45	103	53
Res./Adapt	8.53 (2.5)	9.68 (3.1)	7.85 (2.6)	8.49 (2.3)	10.30 (2.6)	7.91 (2.9)	8.24 (2.5)	10.83 (2.30)	9.89 (2.5)	2.82*	6.32**	1.50
Affable	19.19 (2.5)	19.80 (2.3)	18.93 (2.1)	18.83 (2.1)	19.14 (3.1)	19.24 (2.3)	19.07 (2.3)	19.50 (2.1)	19.04 (2.3)	15.73***	0.19	0.69
Ind./Mature	5.45 (2.3)	5.62 (2.2)	4.90 (2.2)	5.71 (2.2)	6.15 (2.0)	4.85 (1.8)	5.40 (2.3)	6.53 (1.7)	6.23 (1.9)	5.60**	5.51**	1.68
Schl.-Ready	8.86 (2.1)	9.21 (1.6)	8.62 (1.9)	8.97 (1.9)	9.36 (1.3)	8.67 (1.7)	8.84 (2.2)	9.47 (1.5)	9.11 (1.9)	5.49**	0.62	0.68

Note. V=violent community, SV=somewhat violent community, NV=safe (nonviolent) community.

^adf=2,599.

^*p<.05;

^**p<.01;

^***p<.001.

To facilitate comparisons across groups and competence dimensions, standardized scores were developed for each scale, using data from the study sample. The mean was set to 500 and the standard deviation to 100. These data are depicted in figure 1. Tukey post-hoc comparisons with Bonferroni correction for multiple comparisons were computed to explore further the significance of differences among the SES and violence groups. As can be seen in figures 1 and 2, children in families from the highest SES groups showed significantly higher levels of social competence than did the low and middle SES groups on resilient/adaptive (p<.001, p<.02) and independent/mature (p≤.002, p≤.05). The tests on community violence revealed higher levels of social competence in children from moderately safe areas than in those from violent and safe/secure areas: children in moderately safe communities were significantly more resilient/adaptive (p≤.001, p≤.001), independent/mature (p≤.008, p≤.001), and academically ready (p≤ .02, p≤.01) than children in the most violent and the most secure areas. The latter two groups did not differ significantly from one another.

Figure 1.

STANDARDIZED SCALE SCORES ON COMPETENCE DIMENSIONS BY SOCIOECONOMIC STATUS

Figure 2.

STANDARDIZED SCALE SCORES ON COMPETENCE DIMENSIONS BY COMMUNITY VIOLENCE

DISCUSSION

This study reports the psychological and academic competence levels of young South African children whose families differed as to available social and economic resources and the physical safety of the communities in which they lived. Across all economic and community groups, their parents rated children positively on such characteristics as respect for authority, independence, maturity, affability, and interpersonal flexibility—all highly valued and specifically socialized in many African societies (Welch, 1978). Development in most areas of social competence proved significantly influenced by either community danger or economic hardship. The only assessed dimension of children's social development that appeared impervious to differences in family or community environment was affability. Perhaps an agreeable disposition is strongly influenced by temperament and congenital aspects of personality. Although expression of this trait can undoubtedly be shaped and directed by the social environment, it is widely believed to arise from genetic and biological determinants, as well (Zeanah, Boris, & Larrieu, 1997).

Results on the relationship of children's competence and SES were consistent with prior research findings that, in most cases, availability of social and economic resources facilitate development (Richter, 1994). In all domains of competence save affability, children from the most economically advantaged households scored higher than those from less advantaged backgrounds. Children from high SES households appeared more resilient, adaptive, mature, and independent than those from lower SES groups. This appears to support the argument that children from high SES groups encounter fewer hardships and are thus protected from the more taxing demands of life. Moreover, their higher scores on the school-ready scale suggest a greater freedom to devote their energies to the preparatory tasks of childhood—learning and acquisition of skills. The absence of pressures and responsibilities that are integral to the lives of poor children may lead these more advantaged children to view life as orderly, strengthen their sense of autonomy, and enhance their capacity for adaptation.

On the other hand, life is often more complex and challenging for lower SES children. Financial insecurity in the family often gives them first-hand knowledge of their parents' worries. They may be called upon to help meet basic family needs, and may suffer hunger, malnutrition, and homelessness when the adults in their families are unable to provide for them adequately.

With respect to community violence, the best outcomes for children were associated with moderate levels of safety. Children living in moderately safe neighborhoods functioned better socially and academically than those living in the least safe and the most safe communities. Moreover, the developmental advantages of a moderately safe environment applied to children across the entire socioeconomic spectrum. For all SES groups, children living in communities that were moderately safe were most academically motivated and ready for school. An obvious analogy with the inverted U-shaped function depicting response to stress (Selye, 1978) is suggested. Accordingly, favorable social development occurs best under community conditions in which children are neither oblivious to nor preoccupied with issues of safety. This is a reassuring possibility because, even when the most effective public policy initiatives are combined with concerted action, they are, at best, more likely to reduce threats to children's safety than eliminate them entirely. The data suggest that even modest reductions of violence in dangerous neighborhoods will be of considerable benefit to children.

Focus on the adverse effects of community danger on children's development may unintentionally minimize the significance and beneficial effects of sociocultural resources found to varying degrees in all communities. These resources can constitute important protective and supportive mechanisms that wield a powerful influence in the lives of children and their families. For example, by sharing material resources and providing personally affirming ideologies that serve to normalize children's perception of their socioeconomic disadvantage, communities can buffer the potentially deleterious effects of material hardship and dangerous environments on social development. When children perceive that others live in similar circumstances of economic hardship and physical danger, and experience neighbors as protective and nurturing, then poverty and danger become less salient features of their self-image. Thus, access to emotionally meaningful relationships with children and adults in their community is an important source of resilience in young children (Garmezy, 1994) and can work in concert with child nurturing and socialization processes within the family. A similar insight about the significance of community is to be found in the ecological analyses of Garbarino and his co-workers, who argued that neighborhoods contain a highly influential set of life conditions that uniquely amplify the strength and weakness of caregiving characteristics at the family level (Garbarino, 1985). Efforts to understand and promote social competence in children must, then, analyze the synergies created in the relationship of a family to its community.