J Child Fam Stud (2017) 26:1768–1779 DOI 10.1007/s10826-017-0708-6
ORIGINAL PAPER
Cumulative Risk, Emotion Dysregulation, and Adjustment in South African Youth Wendy Kliewer1,2 Basil J. Pillay2 Karl Swain2 Nishola Rawatlal2 Alicia Borre1,3 Thirusha Naidu2 Lingum Pillay2 Thiroshini Govender2 Cathy Geils2 Lena Jäggi1 Tess K. Drazdowski1 Anna W. Wright1 Naseema Vawda2 ●
●
●
●
●
●
●
●
●
●
●
●
Published online: 27 March 2017 © Springer Science+Business Media New York 2017
Abstract Research on cumulative risk is growing, however, little work has occurred in low- or middle-income countries, and few studies have focused on processes linking risk to outcomes. This study explored relations between components of cumulative risk and adjustment in a sample of 324 South African youth (M age = 13.11 years; SD = 1.54 years; 65% female; 56% Black/African; 14% Colored; 23% Indian; 7% White), and tested competing models of emotion dysregulation as a mediator or moderator of risk— adjustment links. Data was collected from youth and their female caregivers during home interviews. Structural equation models and regression analyses accounting for age and sex contributions revealed that emotion dysregulation mediated associations between sociodemographic risk and internalizing symptoms, externalizing problem behavior, and drug use severity, and moderated links between psychosocial risk and internalizing symptoms and externalizing problem behavior. For the mediator models, sociodemographic risk was associated with impaired emotion regulation, which in turn was linked with heightened adjustment difficulties. For the moderator models, psychosocial risk was linked with adjustment problems only when emotion dysregulation was high. These data indicate the importance of disentangling components of cumulative risk. Future research within the South African cultural context might build on these findings by adapting and testing
* Wendy Kliewer
[email protected] 1
Virginia Commonwealth University, Richmond, VA, USA
2
University of KwaZulu-Natal, Durban, South Africa
3
Hampton University, Hampton, VA, USA
school- or family-based prevention or intervention programs that include modules on emotion regulation. Keywords Adolescents Cumulative risk Emotion dysregulation South Africa Internalizing Externalizing ●
●
●
●
●
Introduction Over the last 15 years researchers have documented the toll that accumulated sociodemographic (e.g., poverty, low parental education, single-parent household, minority status), psychosocial (e.g., family stressors, parent-child separation, parental mental illness), and environmental (e.g., housing quality, exposure to violence) risks, collectively known as cumulative risk or multiple risk, have on youth’s health and well-being (for a review see Evans et al. 2013). Further, as noted by Evans et al. (2013), few studies have examined processes linking cumulative risk to adjustment outcomes. Processes are important to study because although it may or may not be possible to change risk, it is often possible to change the process variables that risk factors affect, which in turn can impact adjustment outcomes. Emotion regulatory capacity is a key skill that worsens in response to environmental stressors (e.g., Kelly et al. 2008; Kim and Cicchetti 2010; Lengua 2002; Maughan and Cicchetti 2002; McLaughlin et al. 2009; Schwartz and Proctor 2000), has been linked to adjustment outcomes (cf., Aldao et al. 2010; Berking and Wupperman 2012; Esbjørn et al. 2012), moderates associations between environmental stressors and adjustment (Kliewer 2016), and can be modified during adolescence (Frey et al. 2000; Greenberg et al.
J Child Fam Stud (2017) 26:1768–1779
1995). However, it is unclear whether emotion dysregulation is best characterized as a mediator or a moderator of linkages between cumulative risk and adjustment outcomes in youth. That is, is emotion regulatory capacity best understood as a modifiable skill that changes in response to exposure to risk and accounts for associations between exposure to risk and adjustment, or as a resource that buffers the association between cumulative risk and adjustment? Thompson defined emotion regulation as “…the extrinsic and intrinsic processes responsible for monitoring, evaluating, and modifying emotional reactions, especially their intensive and temporal features, to accomplish one’s goals” (Thompson 1994, pp. 27–28). Emotional awareness, clarity, understanding, and acceptance, and the ability to act or stay in control and accomplish tasks in spite of an emotional experience, and to access regulation strategies in the belief that they will work, are proposed as additions to this multidimensional construct (Gratz and Roemer 2004). In contrast, there is less conceptual consensus regarding emotion dysregulation, although most scholars embrace the notion this also is a multidimensional construct (Weinberg and Klonsky 2009). Difficulties with impulse control and problems engaging in goal-directed behavior when emotionally aroused are two commonly cited qualities of emotion dysregulation. Zeman et al. (2002) have referred to emotional dysregulation as inappropriate expression of anger and sadness. Zeman et al. (2002) found that children’s selfreported dysregulation of anger and sadness predicted more self-reported anxious and depressive symptoms. A considerable body of research has demonstrated negative associations between environmental stressors and the capacity to regulate emotional responses (e.g., Kelly et al. 2008; Kim and Cicchetti 2010; Lengua 2002; Maughan and Cicchetti 2002; McLaughlin et al. 2009; Schwartz and Proctor 2000). For example, in a large sample of racially diverse adolescents, McLaughlin et al. (2009) found that emotion dysregulation worsened over a 4-month period in response to peer victimization. In a study of 199 low-income children, Kelly et al. (2008) demonstrated that violent victimization in the community was associated cross-sectionally and longitudinally with teacher-rated emotion dysregulation. In a study of 421 maltreated and non-maltreated children, Kim and Cicchetti (2010) showed that physical abuse, sexual abuse, and neglect each were associated with diminished emotion regulation capacities as rated by camp counselors. A key reason we are concerned with poor emotion regulation is that it has been implicated in a wide range of adjustment difficulties in adolescents, including internalizing and externalizing problems and substance use (Buckner et al. 2003; Shields and Cicchetti 2001). Reviews (Aldao et al. 2010; Berking and Wupperman 2012; Esbjørn et al.
1769
2012) highlight the role that deficits in emotion regulation play in the development and maintenance of many forms of psychopathology. For example, Esbjørn et al. (2012) noted the critical role that emotion dysregulation plays in the development of anxiety disorders, and Roberton et al. (2012) highlighted the contributions emotion regulation plays in aggressive behavior. Longitudinal work also indicates that emotion dysregulation predicts increases in anxiety symptoms, aggressive behavior, and eating pathology in adolescents (McLaughlin et al. 2011). With regard to emotion dysregulation and substance use, Weinberg and Klonsky (2009) found positive associations between emotion dysregulation and both alcohol and drug use in a sample of middle- and high-school students. Likewise, in a large sample of high school students, Wills et al. (2011) demonstrated links between poor emotion regulation and higher levels of substance use and substance use problems. With respect to sexual risk-taking, Hadley et al. (2015) found that poorer ability to regulate emotions was associated with a broader experience of sexual behavior by Grade 7. Whether emotion dysregulation is better characterized as a mediator of linkages between cumulative risk and adjustment difficulties—explaining why youth who experience greater cumulative risks also experience more adjustment difficulties—or a moderator of those linkages— characterizing youth with poor emotion regulatory capacity who experience high levels of cumulative risk as being particularly vulnerable to adjustment problems—is an empirical question. The evidence to date does not provide definitive answers. However, the answer to the question is important because it can help prevention scientists to know where to direct resources in a way that will do the most good. Evidence for moderation would suggest a selective approach targeting youth with high levels of risk and high levels of emotion dysregulation. Thus, evidence for moderation also would suggest the need for well-validated screening tools. Evidence for mediation would suggest the combination of a universal and a selective approach to reduce both risk and emotion dysregulation. Answering the mediator vs. moderator question will help to guide prevention scientists in designing approaches to facilitate optimal adjustment. In the same vein, disentangling components of cumulative risk is a strategy that can be used to more accurately inform prevention and intervention efforts. Cumulative risk indices may be parceled into sociodemographic risk factors, which on average are less malleable, and psychosocial risk factors, which may be more amenable to intervention, or may be more strongly linked to processes or factors that ultimately affect adjustment relative to sociodemographic risk (Kliewer and Robins 2017). This is important because to be most effective prevention and intervention programs need to target modifiable
1770
risks (National Institutes of Health [NIH] Prevention Research Coordinating Committee [PRCC], 2007). In a study of cumulative risk and physiological stress responses in 205 low-income, urban, African American adolescents, Kliewer and Robins 2017 found that in analyses predicting salivary alpha amylase responses to a stress task, time (e.g., pre, mid, post task) interacted with psychosocial but not sociodemographic risk, illustrating the value of disentangling the components of risk. In most studies of cumulative risk, sociodemographic risk factors include indictors of race/ethnicity, parental education, household income, and family structure. Indexes of psychosocial risk are more variable but often include assessments of family risk or stress, parental mental health, environmental living conditions (housing quality), and exposure to neighborhood violence or discord. In our view psychosocial risk, or exposure to it, is more malleable than sociodemographic risk because of evidence of indicating the positive effects of intervention on increasing family bonding and reducing family stress (Mason et al. 2010) and depression (Firth et al. 2015), as well as the negative associations between parental monitoring and youth externalizing behaviors (LopezTamayo et al. 2016). Finally, despite advances in our understanding of the impact of cumulative risk on youth outcomes, a relatively small portion of this work has taken place in low- and middle-income countries. Each year the World Bank divides economies across the globe into four income groupings: low, lower-middle, upper-middle, and high. Income is calculated using gross national income (GNI) per capita, in U.S. dollars, and estimates are revised each July. For context, in the 2017 fiscal year low-income economies are defined as those with a GNI of $1025 or less per capita; lower middle-income economies are those with a GNI per capita between $1026 and $4035; and upper-middle economies are those with a GNI per capita between $4036 and $12,475 (http://www.worldbank.org/). South Africa is a middle-income country with both historical violence and high levels of current violence and poverty (Lockhat and Niekerk 2000; Seedat et al. 2009; United Nations Office on Drugs and Crime 2013), making it an important context in which to examine the impact of cumulative risk. To investigate relations between components of cumulative risk, emotion regulatory capacities, and South African youth adjustment and behavior we used data from Project CARE: Community Assessment of Risk and Resilience, a collaborative project between a US and a South African University. We examined the following questions: (1) Is there more empirical support for emotion dysregulation as a mediator or a moderator of relations between cumulative risk and adjustment? (2) Does psychosocial risk or sociodemographic risk have stronger associations with emotion
J Child Fam Stud (2017) 26:1768–1779
dysregulation? (3) Does psychosocial risk or sociodemographic risk have stronger associations with adjustment? We conceptualized sociodemographic risk factors (e.g., non-white racial status, low maternal education, poverty, single parent household) as less modifiable sources of risk, whereas psychosocial risk factors, or exposure to them (e.g., housing quality, family stress levels, maternal mental health, community and peer violence exposure) were considered more malleable (Kliewer and Robins 2017).
Methods Participants Participants included one adolescent and their female caregiver from 324 South African families in the greater Durban, South Africa area. There were two cohorts of adolescents: 256 7th grade learners, the South African term for students (M = 12.45 years, SD = .80) and 68 10th grade learners (M = 15.60 years, SD = .96). Approximately twothirds (65%) of the sample was female (210 girls, 114 boys). The sample was racially and ethnically diverse, with 56% of the sample identifying as Black/African, 14% identifying as Colored, 23% identifying as Indian, and 7% identifying as White. Though largely poor, given the aims of the study, there was some economic diversity in the sample. Approximately one-third (33%) of the families had a monthly household income of 2500 Rand or less [equivalent to US $250]; 31 percent earned between 2501 and 5000 Rand per month [equivalent to US $250 to $500]; and 36 percent earned more than 5000 Rand per month. The median household income was 3501–4500 Rand per month [equivalent to US $350 to $450], which is considered lowincome. Households ranged in size from 3 to 16 (M = 6.30, SD = 2.41). Primary maternal caregivers included the adolescent’s biological mother (n = 237, 73.2%), grandmother (n = 39, 12.0%), aunt (n = 23, 7.1%), stepmother (n = 8, 2.5%), sister (n = 7, 2.2%), adopted mother (n = 5, 1.5%), and other female relatives (n = 5; 1.5%). Marital status of the caregivers in the sample varied. Half of the sample (51.9%) was married, and another 3.8% were cohabitating. A third of the sample (31.4%) had never married. Smaller percentages were separated (1.6%), divorced (3.8%), or widowed (7.5%). Caregiver education also was diverse. One in seven caregivers (15.1%) had completed an 8th grade education or less; 29.7% completed some high school or technical school; 42.3% graduated from high school; 5.7% completed 1 to 3 years of College or technical school; and 7.3% completed 4 or more years of College.
J Child Fam Stud (2017) 26:1768–1779
Procedure The research study was approved by Virginia Commonwealth University (VCU) in the United States, the Biomedical Research Committee at the University of KwaZuluNatal (UKZN), Durban, South Africa, and the Department of Education in the Province of KwaZulu-Natal, South Africa. Participants were recruited from primary and secondary schools in low-income areas in the greater Durban, South Africa area that served all of the four major racial/ ethnic groups in the country. Principals were contacted and were asked to allow the research team to recruit youth and families for a study on “risk and resilience in South African youth.” Once the principals understood the study objectives and design and agreed to participate recruitment packets were sent to parents or guardians of all Grade 7 or 10 students in the school; these packets included a letter from the principal. Parents who agreed to participate in the study were contacted and a time was arranged for the parent and youth to be interviewed separately in their home. Parental consent and youth assent was obtained prior to conducting all interviews. Interviews were conducted by research assistants trained by the study team in the Department of Behavioral Medicine, and were completed either in English or isiZulu, a Bantu language spoken by many in the region, depending on the preference of the participant. All recruitment materials, consent forms, and interview questions that were part of the project had been translated and back translated into isiZulu by an accredited translator. Youth received a shopping voucher for R50 (~ $15) and parents received a voucher worth R150 (~ $50) at the end of the home interview. Measures Prior to initiating Project CARE, the research team, which consisted of native South Africans as well as the US researcher, met weekly for 4 months. During this process the team co-designed the study, and the study measures and procedures were reviewed and pilot tested for cultural fit and comprehensibility. Risk Two risk measures were created based on the work of Evans (see Evans et al. 2013 for a review of cumulative risk measures): a sociodemographic risk index and an environmental/psychosocial risk index, hereafter referred to as the psychosocial risk index. The 4-item sociodemographic risk index included variables that were theoretically less amenable to change: (1) single parent household; (2) low maternal education level (less than high school); (3) low household income (<5000 Rand per month [equivalent to
1771
US $500/month]); and (4) non-white racial or ethnic status. Each family received a score of 0 if the risk factor was absent or 1 if the risk factor was present. Thus, the possible range on the index was 0–4. This index was based on maternal caregiver report. The 7-item psychosocial risk index, representing variables that were theoretically more amenable to change or to which exposure could be limited included: (1) large household size; (2) poor housing quality; (3) high levels of family stress; (4) high levels of maternal mental health problems; (5) familial risk; (6) high levels of community violence exposure; and (7) high levels of peer victimization. Each family received a score of 0 or 1, indicating the risk factor was absent or present, using the formulas described below. Thus, the possible range on the index was 0–7. This index was based on both maternal caregiver and youth report. The psychosocial risk index was calculated as follows: (1) household size: household size was reported by maternal caregivers and households with eight or more individuals received a score of 1, which was 1SD above the sample mean; (2) housing quality: youth whose maternal caregivers indicated they resided in “informal settlements” or “RDP/ low cost housing” received a score of 1. Based on consultation with South African researchers familiar with the types of housing, these housing types, vs. “outbuildings,” “flats,” or “semi-detached dwellings” were considered high risk; (3) family stressors: The family stress measure was adapted from a measure developed for the Fast Track Project and included a list of 20 items that maternal caregivers rated as not occurring, occurring and causing minor stress, or occurring and causing major stress. This measure has been used extensively with urban families and appears to have excellent validity (Miller-Johnson et al. 2004). Cronbach alpha for the current study was 0.80.Youth whose families who were 1SD above the mean (i.e., had eight or more life events in the past year) were considered at risk and received a score of 1; (4) maternal mental health problems: The measure of maternal health problems was created by standardizing and combining the anxiety (Cronbach alpha = .87), depression (Cronbach alpha = .88), hostility (Cronbach alpha = .80), and somatic complaints (Cronbach alpha = 0.91) subscales from the Brief Symptom Inventory (Derogatis and Melisaratos 1983), reported by maternal caregivers. Caregivers rated symptoms on a scale from 1 (not at all) to 5 (extremely). The measure is a valid indicator of mental health problems in urban samples (Borre and Kliewer 2014). The subscales were inter-correlated from 0.73 to 0.87 (ps < 0.001). Youth with caregivers scoring at or 1SD above the mean received a score of 1; (5) familial risk: any endorsement of maternal or paternal mental illness, suicidal ideation, psychiatric hospitalization, or incarceration, reported by maternal caregivers, received a score of 1;
1772
(6) lifetime community violence exposure: Youth reported on their lifetime experiences with community violence using the Survey of exposure to community violence: Selfreport version (Richters and Saltzman 1990), a widely used and valid measure of exposure (Fowler et al. 2009). Youth responded on a scale from 0 (never) to 5 (20 or more times). The total indirect (witnessing and hearing about violence) (Cronbach alpha = 0.94) and direct (victimization; Cronbach alpha = 0.71) exposure subscales were standardized and combined. These subscales were correlated 0.63 (p < 0.001). Youth 1SD above the mean received a score of 1; and (7) peer victimization:Youth reported on their victimization experiences in the past month using the peer physical (Cronbach alpha = 0.78) and relational (Cronbach alpha = 0.84) victimization subscales of the Problem Behavior Frequency Scales (PBFS; Farrell et al. 2000). Response options ranged from 0 (never) to 5 (20 or more times).These subscales were correlated 0.69 (p < 0.001). Youth 1SD above the mean received a score of 1. Emotion dysregulation The 5-item anger dysregulation and 5-item sadness dysregulation subscales of the Children’s Emotion Management Scales (CEMS; Zeman et al. 2002) were used to index emotion dysregulation. Adolescents rated the frequency with which they engaged in specific behaviors, ranging from 1 (hardly ever) to 3 (often). The CEMS has good testretest reliability and predictive validity (Zeman et al. 2002). Cronbachα for the combined scales in the present study was 0.81. Outcomes: behavior and adjustment Four domains of behavior and adjustment, all reported by youth, were assessed in the study including internalizing symptoms, externalizing behavior problems, substance use severity, and sexual behavior. Internalizing symptoms Internalizing symptoms were indexed by scores on the 27item Children’s Depression Inventory (CDI) (Kovacs 1981), the 28-item Revised Children’s Manifest Anxiety Scale (RCMAS) (Reynolds and Richmond 1978), the 10-item Post-Traumatic Stress Scale of the Trauma Symptom Checklist for Children (TSCC) (Briere and Lanktree 1995), and the 18-item Children’s Somatization Inventory—Short Form (CSIs) (Walker et al. 2009). The CDI is a widely used measure of depressive symptomatology with excellent validity (Kovacs 1985). Youth rated their level of depressive symptomatology (e.g., felt sad, felt like crying, felt lonely) over the prior 2-week period by identifying which of
J Child Fam Stud (2017) 26:1768–1779
three statements was most like them. Total scores can range from 0 to 54, with higher scores indicating greater depressive symptomatology. The RCMAS assesses youth’s emotional and physical symptoms of anxiety using a yes/no rating scale. This scale has good reliability and validity (Reynolds 1980). Higher scores indicate greater anxiety. For the TSCC, youth rated symptom occurrence in the prior 2 weeks on a scale from (0) never to (3) almost all of the time. Sample items include “bad dreams or nightmares” and “remembering things that happened that I didn’t like.” The TSCC was normed on urban and suburban samples with a large proportion of ethnic minority adolescents and correlates strongly with children’s report of behavioral problems as measured by the Child Behavior Checklist (Achenbach 1991; Briere and Lanktree 1995). Higher scores indicate greater symptom levels. For the CSIs youth rated their frequency of physical symptoms (e.g., headaches, abdominal pain) over the prior 2-week period using a 5-point scale from 0 (not at all) to 4 (a whole lot).The CSI has good internal consistency and validity for both clinical and community samples (Walker et al. 2009). Higher scores indicate greater symptom levels. The latent construct of internalizing symptoms—youth report was formed from these four indicators. Externalizing behavior problems The 7-item Physical Aggression, 5-item Non-Physical Aggression, 6-item Relational Aggression, and 8-item Delinquency subscales of the PBFS (Farrell et al. 2000) were used as youth reports of externalizing behavior problems and served as indicators for the latent contrast of this outcome. For all items, students were asked how frequently they engaged in each behavior during the past 30 days. Responses were based on a 6-point scale: (0) never, (1) 1–2 times, (2) 3–5 times, (3) 6–9 times, (4) 10–19 times, and (5) 20 times or more. This measure was used in the Multisite Violence Prevention Project (Miller-Johnson et al. 2004) and has excellent reliability and validity. Higher scores indicate more aggression or delinquency. Substance use severity The 18-item drug use severity subscale of the Personal Experiences Inventory (PEI) (Winters and Henly 1989b) was used to assess severity of use. Adolescents indicate how often they have used alcohol or drugs in different contexts on a scale from (1) never to (4) often. Higher scores indicate greater severity of drug and alcohol use. Studies examining the psychometric properties of the PEI revealed favorable internal consistency with alphas of 0.75 and higher (Winters and Henly 1989a). Cronbachα for the present study was 0.85.
J Child Fam Stud (2017) 26:1768–1779
1773
severity, and sexual behavior as the outcomes in a second model. Each mediator model included sociodemographic risk and psychosocial risk as exogenous predictor variables, paths from the risk variables to emotion dysregulation and to the outcome variable(s), and controls for youth age and sex (see Figs 1 and 2). Models were run using Mplus 7.3 (Muthén and Muthén 2015), which allowed missing data to be handled with full information maximum likelihood (FIML). FIML uses all information in the data for analyses, allows for less biased estimates, and is an efficient missing data technique. Maximum Likelihood estimation was used in the models. The fit of the models were assessed using the χ2 value, the Comparative fit index (CFI), and the root mean square error of approximation (RMSEA). Values of 0.90 or above for the CFI (Bentler 1990) and 0.08 or below for the
Sexual behavior Seven items from the CDC Youth Risk Behavior Survey was used to assess sexual behavior. Response options on the risk survey ranged from 1 (not at all) to 7 (more than 20 times). Higher values indicate more sexual activity. The scale has excellent reliability and validity (Basen-Engquist et al. 1996). Cronbachα for the present study was 0.56. Data Analyses Structural equation modeling was used to test our mediator hypotheses, and hierarchical regression analyses were used to test our moderator hypotheses. Mediator models were run with internalizing symptoms as the outcome in one model, and externalizing behavior problems, substance use Fig. 1 Test of emotion dysregulation as a mediator of associations between cumulative risk and internalizing symptoms. N = 324. Chi square (23) = 55.85, p < 0.001; RMSEA = 0.066 90% CI [0.044, 0.089]; CFI = 0.941; SRMR = 0.048. *p < 0.05; **p < 0.01; ***p < 0.001
Sex .08
.52***
Age Sexual Behavior
.13*
-.04
Sex .08
Age .12*
Sociodemographic Risk
.11
Non-Physical Aggression
.02 .11*
Emotion Dysregulation
0.20***
.74** .26***
Externalizing Problem Behavior
Physical Aggression
.69***
Relational Aggression 0.68***
.08
Psychosocial Risk
Delinquency
.83***
.17** .19*** .11
-.02
Drug Use Severity
.31***
Age
.03
Sex
Fig. 2 Test of emotion dysregulation as a mediator of associations between cumulative risk and externalizing symptoms N = 324. Chi square (29) = 104.36, p < 0.001; RMSEA = 0.09 90% CI [0.071, 0.108]; CFI = 0.901; SRMR = 0.051. Externalizing problem behavior,
drug use severity, and sexual behavior were significantly associated but beta weights are not displayed for model clarity. *p < 0.05; **p < 0.01; ***p < 0.001
1774
J Child Fam Stud (2017) 26:1768–1779
RMSEA (Browne et al. 1993) indicated that the model adequately fit the data. In order to evaluate the extent to which patterns of association between cumulative risk and adjustment varied as a function of emotion dysregulation and to utilize the full range of the emotion dysregulation variable we employed hierarchical linear regression analyses. The risk and emotion dysregulation variables were centered and interaction terms were computed from the centered variables (Aiken and West 1991). For the model with internalizing symptoms as the outcome, a composite internalizing symptoms score was computed by standardizing and combining the four indicators of internalizing problems. Likewise, for the model with externalizing problems as the outcome, a composite externalizing problems score was computed by standardizing and combining the four indicators of externalizing problems. Additional models with drug use severity and sexual behavior as outcomes also were calculated. Age and sex were controlled in these analyses. When interaction terms were significant they were plotted using conventions outlined in Aiken and West (1991) to determine the nature of the effect.
Results Descriptive Information on and Correlations among Study Variables Table 1 presents descriptive information on the study variables, including data on the proportion of the sample Table 1 Descriptive information and proportion of the sample with specific sociodemographic and psychosocial risk factors
with specific sociodemographic and psychosocial risk factors present. Emotion dysregulation (not shown on Table 1) had a fairly wide distribution (M = 0, SD = 1.77, range = −2.39–5.88). Table 2 presents reliability and descriptive information on, as well as correlations among, the study variables. With the exception of drug use severity, psychosocial risk was more strongly and consistently associated with adjustment than was sociodemographic risk.
Results of Mediator Models Figure 1 presents the results of the mediator model with internalizing symptoms as the outcome. As seen in the figure, the model was a good fit to the data (N = 324; Chi square (23) = 55.85, p < 0.001; RMSEA = 0.066 90% CI [0.044, 0.089]; CFI = 0.941; SRMR = 0.048). Sociodemographic risk was positively associated with emotion dysregulation, and psychosocial risk and emotion dysregulation were positively associated with internalizing symptoms after accounting for age (which was significant) and sex (which was not). Thus, psychosocial risk had direct effects on internalizing symptoms while sociodemographic risk impacted internalizing symptoms through influencing emotion dysregulation. Just over thirteen percent of the variance in internalizing symptoms (p < 0.001) was explained by this model. Figure 2 presents the results of the mediator model with externalizing symptoms as the outcome. As seen in the figure, the model as an adequate fit to the data (N = 324; Chi square (29) = 104.36, p < 0.001; RMSEA = 0.09 90%
Measure
M
SD
Range
Proportion of the sample with the risk factor
Sociodemographic risk Single parent household
45.4
Low maternal education (
46.0
Low household income (≤5000 R/month)a
61.4
Non-white racial or ethnic status
93.2
Total sociodemographic risk
2.46
1.09
0–4
6.30
2.41
3–16
19.8
4.77
3.67
0–20
19.1
Psychosocial risk Large household size (≥8) Poor housing quality
21.3
High past year family stressors (≥8)
a
High maternal mental health problems
0
4.08
−3.37–14.06
16.0
Any endorsement of parental mental illness, suicidal ideation or incarceration
0.23
0.54
0–3
17.6
High lifetime community violence exposure
−0.02
2.64
−2.80–10.14
16.7
High peer victimization
0
1.84
−1.44–8.59
17.0
Total psychosocial risk
1.22
1.10
0–5
Equivalent to approximately US $500 per month
J Child Fam Stud (2017) 26:1768–1779 Table 2 Descriptive informative on and correlations among study variables
Outcome measure
1775 Sociodemo graphic risk
Psychosocial risk
Emotion dysreg
M
SD
α
Post-traumatic stress symptoms
0.03
0.24***
0.26***
5.70
5.69
.86
Depression
0.01
0.13*
0.28***
8.46
6.73
0.85
Anxiety Somatic complaints Physical aggression
−0.03
0.11*
0.15**
8.83
6.31
0.91
0.09
0.10
0.23***
11.53
11.17
0.91
0.08
0.20***
0.21***
1.27
2.27
0.69
−0.04
0.16**
0.17**
0.90
1.62
0.64
Relational aggression
0.05
0.15**
0.16**
0.58
1.51
0.69
Delinquency
0.12*
0.15**
0.29***
0.46
1.14
0.60
Drug use severity
0.20***
0.08
0.24***
0.90
2.69
0.85
Sexual behavior
0.03
0.22***
0.16*
9.02
3.50
0.56
Non-physical aggression
Note: Sociodemographic risk was correlated r = 0.13, p = 0.02 with emotion dysregulation; Psychosocial risk was correlated r = 0.11, p = 0.07 with emotion dysregulation. Sociodemographic and psychosocial risk were correlated r = 0.20, p < 0.001
Results of Moderator Models The overall model with internalizing symptoms as the outcome was significant, F(7, 287) = 7.31, p < 0.001, explaining 15% of the variance in symptoms. At the final step of the regression, age (b = 0.15, p < 0.01), psychosocial risk (b = 0.11, p < 0.05), and emotional dysregulation (b = 0.28, p < 0.001) were significant, but not sociodemographic risk, sex, or the sociodemographic risk X emotional dysregulation interaction. The main effect of psychosocial risk was qualified by a psychosocial risk X emotional dysregulation interaction (b = 0.12, p < 0.05). The plot of this interaction (see Fig. 3, top panel) revealed a
(a)
high med low low
Externalizing Symptoms Composite
CI [0.071, 0.108]; CFI = 0.901; SRMR = 0.051). As with the model with internalizing symptoms as the outcome, sociodemographic risk was positively associated with emotion dysregulation. Psychosocial risk was directly and positively associated with externalizing problem behavior and with sexual behavior; Emotion dysregulation was positively associated with drug use severity. Youth age and sex were accounted for in the model. Age was a significant covariate for drug use severity and sexual behavior; sex (being male) was significantly associated with externalizing problem behavior. Thus, psychosocial risk had direct effects on externalizing problem behavior and sexual behavior, while sociodemographic risk impacted drug use severity indirectly through influencing emotion dysregulation. Just over thirteen percent (13.1%) of the variance in externalizing behaviors (p < 0.001), 15.0% of the variance in drug use severity (p < 0.001), and 30.2% of the variance in sexual risk-taking behavior (p < 0.001), was explained by this model.
Internalizing Symptoms Composite
*p < 0.05; **p < 0.01; ***p < 0.001
PSYCHOSOC med
RI SK
high
(b)
high med low low
PSYCHOSOC
med
RISK
high
Fig. 3 Interaction of psychosocial risk and emotion dysregulation predicting composite symptoms. Internalizing symptoms are displayed on the top panel; externalizing symptoms are displayed on the bottom panel
significant association between psychosocial risk and internalizing symptoms under conditions of high emotional dysregulation but not under conditions of low emotional dysregulation. The overall model with externalizing problems as the outcome also was significant, F(7, 287) = 6.64, p < 0.001, explaining 13.9% of the variance in symptoms. At the final step of the regression, psychosocial risk (b = 0.17, p < 0.01) and emotional dysregulation (b = 0.23, p < 0.001) were
1776
significant, but not age, sex, sociodemographic risk, or the sociodemographic risk X emotional dysregulation interaction. The main effect of psychosocial risk was qualified by a psychosocial risk X emotional dysregulation interaction (b = 0.13, p < 0.05). The plot of this interaction (see Fig. 3, bottom panel) revealed a significant association between psychosocial risk and externalizing symptoms under conditions of high emotional dysregulation but not under conditions of low emotional dysregulation. The overall model with sexual behavior as the outcome was significant, F(7, 287) = 13.34, p < 0.001, explaining 24.5% of the variance in behavior. At the final step of the regression only age was significant (b = 0.45, p < 0.001), although the interaction of psychosocial risk and emotional dysregulation approached significance (b = 0.10, p = 0.054). The final model with drug use severity as the outcome was significant, F(7, 287) = 9.18, p < 0.001, explaining 18.3% of the variance in behavior. At the final step of the regression only age (b = 0.37, p < 0.001) and emotional dysregulation (b = 0.16, p < 0.01) were significant.
Discussion Our study of cumulative risk, emotion dysregulation, and adjustment in South African youth revealed that emotion dysregulation had significant direct associations with sociodemographic risk, but not with psychosocial risk, and both mediated and moderated associations between cumulative risk and adjustment. Emotion dysregulation as mediator was supported in analyses linking sociodemographic risk with internalizing and externalizing adjustment difficulties and drug use severity. Emotion dysregulation as moderator was supported in analyses linking psychosocial risk with internalizing and externalizing adjustment difficulties. There were significant direct associations between psychosocial risk and all outcomes except drug use severity, but no direct associations between sociodemographic risk and outcomes. In the sections below we discuss these findings, their relevance for South African society today, and directions for future research. Sociodemographic risk, which was operationalized in this study as non-white race or ethnicity, low household income, low maternal education, and single parent household status, is less malleable in our view than psychosocial risk, and was positively and uniquely associated with emotion dysregulation, after accounting for other variables in the model. This finding is consistent with research linking poverty to deficits in self-regulation in children (Evans and Kim 2013), and suggests that the deleterious effects of sociodemographic risk transcend cultural and geographical borders.
J Child Fam Stud (2017) 26:1768–1779
Emotion dysregulation partially accounted for observed associations between sociodemographic risk and internalizing symptoms, externalizing behavior problems, and drug use severity. This pattern of findings is consistent with reviews highlighting the role of emotional deficits in psychopathology (Aldao et al. 2010; Berking and Wupperman 2012; Esbjørn et al. 2012). Importantly, in our data, there were no significant direct associations between sociodemographic risk and adjustment when other variables in the model were included. The effects of sociodemographic risk in our sample were observed because of the dysregulated emotion, which then had cascading effects on adjustment. Psychosocial risk, or exposure to it, in contrast to sociodemographic risk, was considered more malleable (Kliewer and Robins 2017). Psychosocial risk was not significantly associated with emotion dysregulation. However, emotion dysregulation interacted with psychosocial risk such that significant associations between psychosocial risk and internalizing symptoms and externalizing behavior problems only were observed when emotion dysregulation was high. Conversely, when emotion dysregulation was low, there was no observed association between psychosocial risk and internalizing symptoms and externalizing behavior problems. This finding is consistent with recent work by Kliewer (2016) who found that emotion regulation skill, as rated by caregivers, protected low-income African American youth from deleterious physiological consequences of victimization experiences. Strengths and Limitations Strengths of the study include a focus on multiple adjustment outcomes, disentangling the concept of cumulative risk, and examination of both mediating and moderating processes related to the role of emotion dysregulation. Study limitations include a lower than desired Cronbach alpha for the measure of sexual behavior, which may have accounted for the non-significant interaction of psychosocial risk and emotion dysregulation predicting sexual behavior. Another study limitation was our sole focus on female caregivers. Mothers and fathers do provide different perspectives on children and are important and unique contributors to child well-being (Adamsons and Johnson 2013; Goncy and van Dulmen 2010). However, given the large proportion of female-headed households in the sample, enrolling fathers in the study would have been challenging and would have taxed our resources beyond what we were able to accomplish. Our cross-sectional design, which did not allow us to examine change in adjustment, also was a limitation. However, based on the definition and construction of our risk index, the risk index arguably preceded adjustment in time.
J Child Fam Stud (2017) 26:1768–1779
Directions for Future Research One direction for future research suggested by these results is that within the South African cultural context, schoolbased prevention and intervention programs that include modules on emotion regulation could be adapted and tested. For example, beginning with programs that have shown efficacy in other cultural contexts (e.g., the PATHS curriculum, Greenberg et al. 1995; the Resolving Conflict Creatively Program, Brown et al. 2004; and Second Step, Frey et al. 2000), researchers might work with local experts using a systematic approach to determine cultural appropriateness in the adaptation process (cf., Borsa et al. 2012; Solomon et al. 2006). Another direction for future research suggested by our findings is to focus on implementing and evaluating familyfocused interventions. Many psychosocial risks are located within the family, or are risks over which caregivers have some control. In a recent nationally representative survey conducted by the South African Human Sciences Research Council, the majority of respondents indicated they were family oriented (Roberts et al. 2013). Using this strong value as leverage, community-based researchers might begin by implementing interventions that emphasize the role of the family. Culturally adapting and testing interventions with known efficacy, such as the Strong African American Families Program (Brody et al. 2006), could be a place to start. Future research also might evaluate the impact of working directly with caregivers to facilitate more effective coping, more adaptive parenting, and fewer adjustment difficulties. In our previous work (Kliewer et al. 2004) we demonstrated that a high-quality relationship with one’s primary caregiver had a strong potential to affect children’s adjustment positively in the face of significant stressors. We argued in that paper that a first step in getting there is to attend to the mental health needs of parents, many of whom are suffering from depression or post-traumatic stress disorder. In terms of methodology, incorporating different types of data collection such as ecological momentary assessment that can capture regulation of affect as it is occurring would be a valuable addition in future studies. Additionally, incorporating biological assessments into the protocols would expand the domains of functioning covered. Lastly, future research might leverage findings on the importance of emotion regulation to implement and evaluate programs designed to promote positive youth development. South Africa is at a critical point in its history, now less than 25 years post-apartheid. After a period of improved economic growth and reductions in crime, these gains recently have begun to recede. Unemployment is on the rise (http://www.businessinsider.com/south-africaunemployment-rate-rises-2016-5) and crime is up (https://
1777
www.osac.gov/pages/ContentReportDetails.aspx?cid= 19814), making it even more essential to enact programming to promote positive youth development in the face of these societal trends. Acknowledgements We would like to thank the Global Education Office at Virginia Commonwealth University (VCU), Richmond, Virginia, USA and the University of KwaZulu-Natal (UKZN), Durban, South Africa for helping to fund this research, the undergraduate students at VCU and research staff at (UKZN) who worked on the project, the school staff who facilitated participation in the study, and the families who participated. Author contributions W.K.: designed the study, conducted the data analyses, and wrote the paper. B.J.P.: collaborated with the design and writing of the study. K.S.: collaborated with the design of the study, and executed part of the study. N.R.: collaborated with the design and writing of the study, and executed part of the study. A.B.: executed part of the study and collaborated in editing of the final manuscript. T. N.: collaborated with the design of the study, and executed part of the study. L.P.: executed part of the study and collaborated in editing of the final manuscript. T.G.: collaborated with the design and writing of the study. C.G.: collaborated with the design and writing of the study, and executed part of the study. L.J.: collaborated with writing of the study and in editing of the final manuscript. T.K.D.: collaborated with writing of the study and in editing of the final manuscript. A.W.W.: collaborated with writing of the study and in editing of the final manuscript. N.W.: collaborated with the design of the study and in editing of the final manuscript. Conflict of Interest interests.
All authors declare that they have no competing
Ethical Approval All procedures performed in these studies were in accordance with the ethical standards of the institutional research committees of the two educational institutions involved and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
References Achenbach, T. M. (1991). Integrative manual for the 1991 CBCL/418, YSR, and TRF profiles. Burlington, VT: University of Vermont Department of Psychiatry. Adamsons, K., & Johnson, S. K. (2013). An updated and expanded meta-analysis of nonresident fathering and child well-being. Journal of Family Psychology, 27, 589–599. doi:10.1037/ a0033786. Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Aldao, A., Nolen-Hoeksema, S., & Schweizer, S. (2010). Emotionregulation strategies across psychopathology: a meta-analytic review. Clinical Psychology Review, 30, 217–237. doi:10.1016/j. cpr.2009.11.004. Basen-Engquist, K., Edmundson, E. W., & Parcel, G. S. (1996). Structure of health risk behavior among high school students. Journal of Consulting and Clinical Psychology, 64, 764–775. doi:10.1037/0022-006X.64.4.764. Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246. doi:10.1037/0033-2909. 107.2.238. Berking, M., & Wupperman, P. (2012). Emotion regulation and mental health: recent findings, current challenges, and future directions.
1778 Current Opinion in Psychiatry, 25, 128–134. doi:10.1097/YCO. 0b013e3283503669. Borre, A., & Kliewer, W. (2014). Parental strain, mental health problems, and parenting practices: a longitudinal study. Personality and Individual Differences, 68, 93–97. doi:10.1016/j.paid.2014. 04.014. Borsa, J. C., Damasio, B. F., & Bandeira, D. R. (2012). Cross-cultural adaptation and validation of psychological instruments: some considerations. Paidéia, 22, 423–432. doi:10.1590/198243272253201314. Briere, J., & Lanktree, C. (1995). The trauma symptom checklist for children (TSCC): Preliminary psychometric characteristics. Unpublished Manuscript, Los Angeles, CA: Department of Psychiatry, University of Southern California School of Medicine, 31. Brody, G. H., Murry, V. M., Kogan, S. M., Gerrard, M., Gibbons, F. X., Molgaard, V., Brown, A. C., Anderson, T., Chen, Y., Luo, Z., & Wills, T. A. (2006). The strong African American families program: a cluster-randomized prevention trial of long-term effects and a mediational model. Journal of Consulting and Clinical Psychology, 74, 356–366. doi:10.1037/0022-006X.74.2.356. Brown, J. L., Roderick, T., Lantieri, L., & Aber, J. L. (2004). The resolving conflict creatively program: a school-based social and emotion learning program. In J. E. Zins, R. P. Weissberg, M. C. Wang, & H. J. Walberg (Eds.), Building academic success of social and emotional learning. What does the research say? (pp. 151–169). New York, NY: Teachers College Press. Browne, M. W., Cudeck, R., Bollen, K. A., & Long, J. S. (1993). Alternative ways of assessing model fit. In K. A. Bollen, J. S. Long (Eds). Testing structural equation models (pp. 136–162). Beverly Hills, CA: Sage. Buckner, J. C., Mezzacappa, E., & Beardslee, W. R. (2003). Characteristics of resilient youths living in poverty: the role of selfregulatory processes. Development and Psychopathology, 15, 139–162. doi:10.1017.S0954579403000087. Derogatis, L. R., & Melisaratos, N. (1983). The brief symptom inventory: an introductory report. Psychological Medicine, 13, 595–605. doi:10.1017/S0033291700048017. Esbjørn, B. H., Bender, P. K., Reinholdt-Dunne, M. L., Munck, L. A., & Ollendick, T. (2012). The development of anxiety disorders: considering the contributions of attachment and emotion regulation. Clinical Child and Family Psychology Review, 15, 129–143. doi:10.1007/s10567-011-0105-4. Evans, G. W., & Kim, P. (2013). Childhood poverty, chronic stress, self-regulation, and coping. Child Development Perspectives, 7, 43–48. doi:10.1111/cdep.12013. Evans, G. W., Li, D., & Whipple, S. S. (2013). Cumulative risk and child development. Psychological Bulletin, 139, 1342. doi:10. 1037/a0031808. Farrell, A. D., Kung, E. M., White, K. S., & Valois, R. F. (2000). The structure of self-reported aggression, drug use, and delinquent behaviors during early adolescence. Journal of Clinical Child Psychology, 29, 282–292. doi:10.1207/S15374424jccp2902_13. Firth, N., Barkham, M., & Kellett, S. (2015). The clinical effectiveness of stepped care systems for depression in working age adults: a systematic review. Journal of Affective Disorders, 170, 119–130. doi:10.1016/j.jad.2014.08.030. Fowler, P. J., Tompsett, C. J., Braciszewski, J. M., Jacques-Tiura, J. J., & Baltes, B. B. (2009). Community violence: a meta-analysis on the effect of exposure and mental health outcomes of children and adolescents. Development and Psychopathology, 21, 227–259. doi:10.1017/S0954579409000145. Frey, K. S., Hirschstein, M. K., & Guzzo, B. A. (2000). Second step: preventing aggression by promoting social competence. Journal of Emotional and Behavioral Disorders, 8, 102–112. doi:10. 1177/106342660000800206.
J Child Fam Stud (2017) 26:1768–1779 Goncy, E. A., & van Dulmen, M. H. M. (2010). Fathers do make a difference: parental influences on adolescent alcohol use. Fathering, 8, 93–108. doi:10.3149/fth.0801.93. Gratz, K. L., & Roemer, L. (2004). Multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. Journal of Psychopathology and Behavioral Assessment, 26, 41–54. doi:10.1007/s10862-008-9102-4. Greenberg, M. T., Kusche, C. A., Cook, E. T., & Quamma, J. P. (1995). Promoting emotional competence in school--aged children: the effects of the PATHS curriculum. Development and Psychopathology, 7, 117–136. doi:10.1017/ S0954579400006374. Hadley, W., Houck, C. D., Barker, D., & Senocak, N. (2015). Relationships of parental monitoring and emotion regulation with early adolescents’ sexual behaviors. Journal of Developmental and Behavioral Pediatrics: JDBP, 36, 381–388. doi:10.1097/ DBP.0000000000000160. Kelly, B. M., Schwartz, D., Gorman, A. H., & Nakamoto, J. (2008). Violent victimization in the community and children’s subsequent peer rejection: the mediating role of emotion dysregulation. Journal of Abnormal Child Psychology, 36, 175–185. doi:10. 1007/s10802-007-9168-6. Kim, J., & Cicchetti, D. (2010). Longitudinal pathways linking child maltreatment, emotion regulation, peer relations, and psychopathology. Journal of Child Psychology and Psychiatry, 51, 706–716. doi:10.1111/j.1469-7610.2009.02202.x. Kliewer, W. (2016). Victimization and biological stress responses in urban adolescents: emotion regulation as a moderator. Journal of Youth and Adolescence, 45, 1812–1823. doi:10.1007/s10964015-0398-6. Kliewer, W., Cunningham, J., Diehl, R., Parrish, K., Walker, J. M., Atiyeh, C., Neace, B., Duncan, L., Taylor, K., & Mejia, R. (2004). Violence exposure and adjustment in inner-city youth: child and caregiver emotion regulation skill, caregiver-child relationship quality, and neighborhood cohesion as protective factors. Journal of Clinical Child and Adolescent Psychology, 33, 477–487. doi:10.1207/s15374424jccp3303_5. Kliewer, W., & Robins, J. (2017). Cumulative risk and physiological stress responses in African American adolescents. Biological Research for Nursing. (in press) Kovacs, M. (1981). Rating scales to assess depression in school-aged children. Acta Paedopsychiatrica: International Journal of Child & Adolescent Psychiatry, 46, 305–315. Kovacs, M. (1985). The children’s depression, inventory (CDI). Psychopharmacology Bulletin, 21, 995–998. Lengua, L. J. (2002). The contribution of emotionality and self‐regulation to the understanding of children’s response to multiple risk. Child Development, 73, 144–161. doi:10.1111/1467-8624. 00397. Lockhat, R., & Niekerk, V. (2000). South African children: a history of adversity, violence and trauma. Ethnicity & Health, 5, 291–302. doi:10.1080/713667462. Lopez-Tamayo, R., Robinson, W. L., Lambert, S. F., Jason, L. A., & Ialongo, N. S. (2016). Parental monitoring, association with externalized behavior, and academic outcomes in urban AfricanAmerican youth: a moderated mediation analysis. American Journal of Community Psychology, 57, 366–379. doi:10.1002/ ajcp.12056. Mason, W. A., Hawkins, J. D., Kosterman, R., & Catalano, R. F. (2010). Alcohol use disorders and depression: protective factors in the development of unique versus comorbid outcomes. Journal of Child & Adolescent Substance Abuse, 19, 309–323. doi:10. 1080/1067828X.2010.502491. Maughan, A., & Cicchetti, D. (2002). Impact of child maltreatment and interadult violence on children’s emotion regulation abilities
J Child Fam Stud (2017) 26:1768–1779 and socioemotional adjustment. Child Development, 73, 1525–1542. doi:10.1111/1467-8624.00488. McLaughlin, K. A., Hatzenbuehler, M. L., & Hilt, L. M. (2009). Emotion dysregulation as a mechanism linking peer victimization to internalizing symptoms in adolescents. Journal of Consulting and Clinical Psychology, 77, 894–904. doi:10.1037/a0015760. McLaughlin, K. A., Hatzenbuehler, M. L., Mennin, D. S., & NolenHoeksema, S. (2011). Emotion dysregulation and adolescent psychopathology: a prospective study. Behaviour Research and Therapy, 49, 544–554. doi:10.1016/j.brat.2011.06.003. Miller-Johnson, S., Sullivan, T. N., & Simon, T. R., Multisite Violence Prevention Project. (2004). Evaluating the impact of interventions in the multisite violence prevention study: samples, procedures, and measures. American Journal of Preventive Medicine, 26, 48–61. Muthén, B., & Muthén, L. (2015). Mplus version 7: User’s guide. Los Angeles, CA: Muthén & Muthén. National Institutes of Health (2007). Office of disease prevention, prevention research coordinating committee. Working Definition of Prevention Research. http://prevention.nih.gov/research.aspx. Reynolds, C. R. (1980). Concurrent validity of what I think and feel: the revised children’s manifest anxiety scale. Journal of Consulting and Clinical Psychology, 48, 774 doi:10.1037/0022006X.48.6.774. Reynolds, C. R., & Richmond, B. O. (1978). What I think and feel: a revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6, 271–280. doi:10.1007/ BF00919131. Richters, J. E., & Saltzman, W. (1990). Survey of exposure to community violence: Self-report version. Rockville, MD: National Institute of Mental Health. Roberton, T., Daffern, M., & Bucks, R. S. (2012). Emotion regulation and aggression. Aggression and Violent Behavior, 17, 72–82. doi:10.1016/j.avb.2011.09.006. Roberts, B., Gordon, S., & Struwig, J. (2013). Family matters: an overview of family values and preferences. HSRC Review. Pretoria: Human Sciences Research Council. Retrieved from: http:// www.hsrc.ac.za/en/review/hsrc-review-july-2013/family-mattersan-overview-of-family-values-and-preferences. Schwartz, D., & Proctor, L. J. (2000). Community violence exposure and children’s social adjustment in the school peer group: the mediating roles of emotion regulation and social cognition.
1779 Journal of Consulting and Clinical Psychology, 68, 670–683. doi:10.1037/0022-006X.68.4.670. Seedat, M., Van Niekerk, A., Jewkes, R., Suffla, S., & Ratele, K. (2009). Violence and injuries in South Africa: prioritising an agenda for prevention. The Lancet, 374, 1011–1022. Shields, A., & Cicchetti, D. (2001). Parental maltreatment and emotion dysregulation as risk factors for bullying and victimization in middle childhood. Journal of Clinical Child Psychology, 30, 349–363. doi:10.1207/S15374424JCCP3003_7. Solomon, J., Card, J., & Malov, M. (2006). Adapting efficacious interventions. Advancing translational research in HIV prevention. Evaluation & the Health Professions, 29, 162–194. Thompson, R. A. (1994). Emotion regulation: a theme in search of definition. Monographs of the Society for Research in Child Development, 59, 25–52. doi:10.2307/1166137. United Nations Office on Drugs and Crime. (2013). Global study on homicide 2013 (United Nations publication, Sales No. 14.IV.1). Retrieved from http://www.unodc.org/documents/gsh/pdfs/2014_ GLOBAL_HOMICIDE_BOOK_web.pdf. Walker, L. S., Beck, J. E., Garber, J., & Lambert, W. (2009). Children’s somatization inventory: psychometric properties of the revised form (CSI-24). Journal of Pediatric Psychology, 34, 430–440. doi:10.1093/jpepsy/jsn093. Weinberg, A., & Klonsky, E. D. (2009). Measurement of emotion dysregulation in adolescents. Psychological Assessment, 21, 616–621. doi:10.1037/a0016669. Wills, T. A., Pokhrel, P., Morehouse, E., & Fenster, B. (2011). Behavioral and emotional regulation and adolescent substance use problems: A test of moderation effects in a dual-process model. Psychology of Addictive Behaviors, 25, 279–292. doi:10. 1037/a0022870. Winters, K. C., & Henly, G. (1989a). Assessing adolescents who abuse chemicals: the chemical dependency adolescent assessment project. NIDA research monograph, 77, 4–18. Winters, K. C., & Henly, G. A. (1989b). Personal experience inventory (PEI): Manual. Los Angeles, CA: Western Psychological Services. Zeman, J., Shipman, K., & Suveg, C. (2002). Anger and sadness regulation: predictions to internalizing and externalizing symptoms in children. Journal of Clinical Child and Adolescent Psychology, 31, 393–398. doi:10.1207/S15374424JCCP3103_11.