Social protection policies and cash transfers in particular have expanded in important ways in developing countries to break the cycle of intergenerational poverty and inequality (Barrientos, 2013; Bagstagli et al., 2016). An evaluation synthesis study of social protection programs for children and families in developing countries shows positive well-being outcomes in different domains (UNICEF, 2015). However, cash transfers on their own are not able to address the complex and multi-dimensional needs of children and their families and complementary interventions are advocated in South Africa (Patel et al., 2017). There is growing support for combination cash transfers (social protection) and family care interventions to improve people’s lives (Banerjee et al., 2015) and particularly in promoting child and adolescent well-being in Sub-Saharan Africa (Roelen et al., 2017; Cluver et al., 2016). Different models of combination have been identified, also known as ‘cash plus’ programs that may include psychosocial support, in-kind transfers, improved access to social supports and linkages to services, education, communication and information strategies, behavior change and improving care giver practices to mention a few (Roelen et al., 2018). Family strengthening interventions are one kind of complementary service that could be provided to boost outcomes for disadvantaged children and their families along with cash transfers.

Some promising results are emerging of combined interventions focusing on the early years of life and on caregiver practices in developing countries (Arriagada et al., 2018; Britto et al., 2015). The evidence of ‘cash plus’ family strengthening interventions is scarce, particularly in assessing both child and family well-being in the foundation years of schooling (6–8 years). Holistic social investments in this particular phase are crucial in promoting a love of learning, fostering parental engagement in children’s education, promoting their psychosocial well-being through sound child and caregiver relations; all of which are associated with long-term improvements in education, employment, income and overall personal well-being (Heckman, 2008). However, a narrow focus on one or more child well-being outcomes tends to negate the significant role of caregivers and the overall family and community environment in ensuring their welfare. South Africa, like Brazil has one of the largest cash transfer programs in the developing world. It also has a fairly well-developed welfare system with social work services delivered by both professionals and paraprofessionals; although, this system remains under-resourced and biased towards urban areas. These services are largely remedial and are of a statutory nature. Limited examples exist of the effectiveness of carefully and deliberately designed combination programs to boost well-being outcomes of poor families with young children.

Over 12 million children and close to two-thirds of all children in South Africa receive a monthly flat-rate cash rate benefit, the Child Support Grant (CSG). It is a publicly funded, means-tested cash transfer that is paid to the primary caregiver of the child who may be a parent, relative or non-relative of the child. Despite the small monetary value of the grant of approximately US$ 30, the poverty reduction effects of the transfer are well-established (Department of Planning, Monitoring and Evaluation, Statistics South Africa and The World Bank, 2018). Other positive benefits include improved child nutrition (Agüero, Carter, & Woolard, 2007; Patel et al., 2017), school attendance and performance (DSD, SASSA, & UNICEF, 2012), and increased caregiver engagement in children’s well-being (Patel, Knijn, & van Wel, 2015) among others. However, child and family well-being is compromised by persistently high rates of household poverty and unemployment, unusually high rates of HIV and AIDS prevalence in poor families, high rates of physical and sexual violence perpetrated against children and women and higher than expected rates of stunting among young children (Hall et al., 2018). Poverty is recognized as a significant risk factor for the well-being of children and their families (Meinck et al., 2015) and leads to severe family stress. This situation is compounded by the cumulative disadvantage experienced by Black African and Coloured families (people of mixed race) due to the legacy of apartheid. How to boost cash transfers with other family interventions is a major challenge. The aim of the study was to assess whether combined cash transfers and a preventative family social work intervention improves social outcomes for children and their families.

Child Support Grant (CSG) Beneficiaries and Their Families

The CSG was rolled out in 1998 starting with children under 6 years but due to its positive impact it was expanded to all children under 18 years of age. About a third of all beneficiaries are under 8 years of age. Despite the expansive coverage of the program, about 17% of children are not receiving the grant due to a lack of identity documents, caregivers earning an income above the means test threshold and other familial factors (DSD, SASSA, & UNICEF, 2016). The grant is well-targeted at poor children and their families although the amount received is low. Based on national household survey data analyzed specifically for CSG children and their caregivers in the age cohort of interest to this study, the following descriptive statistics provide background information on the profile of grant receiving children and their families (see Patel et al., 2017).

First, in relation to household income, made up of social assistance grants and other income, all families had an income below the upper bounds of the poverty line. The implication is that these families had insufficient access to food and other basic consumption goods. Most children lived with their biological parents, mainly the mother of the child (78%). Second, single parent families with relatives were the norm (34%) and care by multiple relatives in the household was not uncommon. Significant numbers of children (29%) lived with relatives only. Father absence was high with almost three-quarters of fathers not being present. On average, a caregiver received 2.2 CSGs. Children in this age group lived in larger households (6.5) compared to the national average (3.6). Third, the majority of CSG children were attending school (92%). While most were in the normal range for body mass index (82%), CSG children were three times more likely to be over-weight than underweight. This finding is likely due to diets that are high in carbohydrates. Four out of ten CSG households indicated that they experienced food scarcity. Fourth, an analysis of the characteristics of caregivers indicated that they were predominantly female (97%) although men may apply. Caregivers had mainly a secondary education and were largely unemployed (87%) although some were engaged in casual work and other micro-enterprise activities. Almost a third of caregivers were at risk of depression with the risk being higher for those with lower levels of education (see also Plagerson et al., 2014). Finally, although CSG beneficiary families lived in households with a medium level of access to basic services such as water, electricity and sanitation, the quality and efficiency of services were considered to be poor. Qualitative evidence of the family relations of CSG families is sparse. Data from six focus group discussions with caregivers in urban and rural areas affirmed the remarkable levels of care that families managed to achieve under very difficult circumstances. Challenges identified included parent child relations such as behavior management of children, discipline of children, inadequate social support, tensions with relatives and kin, depressive symptomatology and service delivery challenges. Participants also expressed interest in growing caregiving knowledge and skills and improving knowledge of financial literacy and nutrition. Although there was evidence of caregiver engagement with teachers, this engagement appeared to be inconsistent and some found interaction with school difficult (see Patel, Hochfeld, & Chiba, 2019; Hochfeld, Chiba, & Patel, 2020).

Ecological & Social Development Theory: Cash Transfers Plus Family Strengthening Interventions

The ecological theory that guided the study was based first on the notion that child well-being is multi-dimensional, and that these different dimensions of well-being are interrelated (Pollard & Lee, 2003). The dimensions include material (or economic), physical, cognitive, social and emotional well-being (Meinck et al., 2015; Minkkinen, 2013; September & Savahl, 2009). A second premise is that children’s well-being is integrally connected to the welfare of their caregivers, the nature and quality of family relations, and their family’s connectedness to their communities including access to services within an enabling and supportive world (Bronfenbrenner, 1979). Given the changing structure of families locally, particularly in regard to single parenthood (Wright et al., 2019) and high rates of father absence (Makusha et al., 2019), the family is broadly defined, acknowledging a diversity of family forms that is made up of a network of interdependent relationships that need to work synergistically to achieve optimal outcomes. The family intervention was designed to build on family strengths, and addresses particular risk factors associated with low levels of child and caregiver well-being, with the purpose of improving positive parenting practices through equipping caregivers with appropriate knowledge, skills and information (Tolan et al., 2004). Third, the following protective factors are hypothesized to be associated with better well-being outcomes for children and families: having access to social support (Cobb, 1976); parental involvement in children’s education (Bogenschneider et al., 2012); knowledge of nutrition, caregiver knowledge of children’s development, positive caregiver mental health (Black, 2012) and enhanced financial capabilities leading to greater personal efficacy (Sherraden, 2013). These assumptions are derived both from theory and empirical evidence emerging in different contexts and informed the design of the family intervention. Of particular relevance are Developmental-Ecological Risk theory (Bronfenbrenner, 1979; Tolan, Guerra & Kendall, 1995), systems thinking, the psycho-educational approach to family intervention (Tolan, Gorman-Smith & Henry, 2004). Ecological theory is consistent with the social development approach with its focus on integrating material (economic) and the psychosocial well-being of individuals and families; the construction of beneficiaries as active agents and partners in promoting family well-being; and its concern with poverty reduction, poverty prevention and finding appropriate solutions in development contexts (Patel, 2015). Taken together, it is assumed that combination interventions of this kind could disrupt the deep seated structural and systemic disadvantage that poor families experience in urban communities in South Africa. These ideas are central to the country’s developmental welfare policy and the developmental social work approach (Patel, 2015). Finally, it was anticipated that the combined intervention would lead to primary improvements in child and caregiver well-being in the following domains:

  • Family and caregiver relations: child, caregiver and family communication; bonding activities, use of positive parenting practices and perception of own parenting ability.

  • Social networks and social support: knowledge and use of resources; and strengthened social networks of support.

  • Caregiver and family involvement in child’s education: active support of child’s learning, attendance of school meetings, homework, school performance, and child’s enjoyment of school and functioning at school.

  • Financial capabilities: budgeting, borrowing and savings behavior; and communication in family about wants and needs.

  • Nutritional knowledge: knowledge of healthy eating principles.

  • Caregiver mental health: symptoms of depression.

Based on the above assumptions, it was hypothesized that the CSG, when combined with a family strengthening intervention (Sihleng’imizi program) would lead to improvements in child and family well-being in the above-mentioned six domains. The results of the study are reported on in relation to the question as to whether changes occurred in a positive direction in each of the identified domains in the intervention group compared to the non-intervention group.

The Sihleng’imizi Family Strengthening Program

Sihleng’imizi (meaning ‘we care for families’) was designed to improve caregiver knowledge and practices through a psycho-educational intervention. Since there is a scarcity of tested programs in developing countries, the Sihleng’imizi Family Strengthening Intervention (FSI) was adapted from a previously rigorously tested family intervention, SAFE Children in the United States (Gorman Smith et al., 2007) and a South African innovation for adolescent children and caregivers, Sinovuyo Caring Families (Cluver et al., 2016). Systematic review evidence internationally shows that the principles undergirding parenting programs are transferable to diverse contexts (Gardner, Montgomery, & Knerr, 2016). The following distinguishing program features were first, that the whole family was engaged in the FSI; therefore all members were invited to participate including children and adults. Fourteen weekly training sessions of 2 h duration were held with five families in a group with a total of 25 people. Second, recruitment of families was done via schools in 10 of the poorest wards in the City of Johannesburg which had high uptake of the CSG with multiple levels of social and economic deprivation. Third, the curriculum included five content areas that corresponded with the dimensions of change identified in the theory above. In the curriculum, more sessions were allocated to the module on child-caregiver relations. Learning materials were translated into four languages. All sessions were conducted in vernacular and consisted of work books and homework exercises, group discussions, practical exercises and role plays. Each session consisted of family exercises involving caregivers, family members and children. Where this approach was not appropriate, children spent some of the time in a separate group activity that was tailored to address a particular aspect of the program content. Fourth, group facilitators were qualified social workers who were specifically trained to deliver the program and they were supported by qualified child care workers. Social workers received weekly supervision from senior social workers for the duration of the FSI. Delivery of the intervention followed the sessions in the manual and work books with homework exercises were provided. Fifth, each family was paired with another family (Sihleng‘imizi buddy) to support each other between sessions and after the groups terminated. Lastly, the sessions focused on real-life family needs, and opportunities were created to practice new skills learnt between sessions and for the co-creation of knowledge and learning in the groups. The content of the sessions is contained in Table 1.

Table 1 Content of sessions
Full size table

Method

Research Design

A qualitative design was employed to elicit the participants’ views of what changes occurred as a result of the family intervention. These types of responses could be best captured through open-ended and exploratory questioning with the caregivers and teachers using specially designed interview guides. Feedback from the children involved drawings and discussions with them about what the drawing meant to them. In order to assess change over time, a pre-post design was employed consisting of interviews with all three groups at pre-treatment assessment (start of the intervention), at end-point (after the intervention at the end of the 14 sessions) and nine-month follow-up interviews were conducted. For purposes of comparison, the sample was divided into an intervention and non-intervention group to assess the similarities and differences between the two groups which could enhance understanding of the potential effects of the change. The use of both intervention and non-intervention groups to assess change was considered appropriate for purposes of comparison and is acceptable in qualitative research designs (Lloyd-Jones, 2003).

Sampling

Each school provided a class list of the children in grades R and Grade 1. The selection criteria were that there was an identified child who: (a) was attending Grade R or Grade 1 in 2017; (b) the child was a recipient of the CSG; (c) the family lived locally, so that access to the group venue would be uncomplicated; and (d) the caregiver and the child agreed to the process. The class-lists of all the Grade R and Grade 1 classes in each selected school were provided by the school. Each child was assigned a number, which was then selected randomly from the class list using the randomizing feature on MS Excel. Seven numbers and seven substitute numbers were selected for the intervention group and also for the non-intervention group. There were no significant differences in the social profiles of the intervention and the non-intervention group. Fieldworkers contacted the families by phone, checked that the family met the sampling criteria, and then explained the intervention and the study to the intervention group and invited their participation. Only the study was explained to the non-intervention group.

The recruitment target for the intervention group was 60 families, the number recruited was 64 families, and the final number that completed the program was 40 families. Data were collected from 71 non-intervention group families at pre-treatment assessment, and 65 families at endpoint. However, in the non-intervention group only data for 20 families was analyzed as data saturation had been achieved. Nine months after the intervention, 25 families from the intervention group were again randomly selected from the participating families in the study and 15 families from the non-intervention group. Seven substitute families were again selected across both groups. In the selection of the families, care was again taken to ensure that all geographic areas were represented. Data analysis and reporting in this article is based on data collected for 131 families over the study period of 12 months. The retention rate of families in the intervention group was 62% from start to completion. This is lower than the over 90% retention in Sinovuyo Teens, a comparable family program (Cluver et al., 2016), but is similar to the average retention in parenting programs (Smokowski et al., 2018).

In addition to the interviews with the caregivers of families, the drawings of 36 children who participated in the intervention were analyzed prior to the intervention; 25 at the end of the intervention and 15 children were assessed again at follow-up 9 months later. Children’s drawings could not be analyzed for the non-intervention group. Comparisons could therefore only be made between data collected at the start and end of the delivery of the family intervention. Additionally, 56 teachers were interviewed in the pre-treatment assessment and 19 at post-treatment assessment. The nine-month follow-up interviews could not be conducted with the teachers for various logistical reasons. No interviews were conducted with teachers in the non-intervention group. Here again comparisons could only be made between commencement and conclusion of the program. The children were all beneficiaries of the CSG and were randomly selected as described in the sampling procedure above. Teachers of children in Grades R and Grade one were engaged at the selected schools. All the schools were in poor communities with high uptake of the CSG. The communities were selected in consultation with the City of Johannesburg’s Department of Social Development which delivered social work services in these communities. Table 2 provides the sample sizes of the intervention and non-intervention groups over three data collection points.

Table 2 Sample sizes of caregivers, children and teachers in the intervention and non-intervention groups over three waves of data collected (N = 131 families)
Full size table

Research Tools and Data Collection

Triangulation of data occurred through collecting data from multiple sources namely, children, their caregivers and their educators. Data collection procedures were exactly the same for the intervention and non-intervention groups and over the three data collection points. At pre-treatment assessment, the identified child, the primary caregiver, and the child’s educator were interviewed, and these interviews were repeated at endpoint. At follow-up only the caregivers and the child were interviewed as educators were not available due to the demands of the school year. Short questionnaires were administered to collect specific information to assess caregiver depression, including a separate section to assess knowledge and practices in relation to family nutrition, provision of healthy meals and healthy food choices. Interviews were conducted in English, and the local vernacular languages that is, isiZulu, SeSotho and Tsonga/Shangaan, depending on the preferences of the interviewees. Audio-recordings were translated into English by professional translators who were fluent in all these languages. The research tools are summarized in Table 3.

Table 3 Research instruments
Full size table

Analysis of Data

Interviews with caregivers, educators and children were analyzed using a combination of closed and open coding and thematic analysis techniques. The children were asked to draw a picture of themselves and their families at pre-treatment assessment, endpoint and follow-up. The drawing analysis took the following factors into consideration: maturation of the child (e.g., Koppitz, 1968; Lowenfeld & Britain, 1970; Crawford, Gross, & Patterson, 2012); indicators depicting the emotional well-being of the child representing their perceived state of happiness; (e.g., Koppitz, 1968; Dunn, O’Connor & Levy, 2002; Haghighi et al., 2014); exclusion of family members (e.g., Gerhardt, Keller, & Ruberling, 2016); and cultural factors that may be pertinent in the local context (e.g., Betts, 2013; Baluch et al., 2017). Children’s drawings provide the opportunity to enable them to express their inner world of thoughts, feelings and relationships. However, these drawings cannot be relied upon as sole indicators of their emotional and cognitive state and their age, and level of maturation and cultural factors should be taken into account. Hence, the findings from the drawings were triangulated with caregiver interviews and the educator reports where this information was available. A short structured questionnaire including a depression index (CESDR-10) was administered at the commencement of each interview with the caregiver. The CESDR-10 is not a diagnostic tool but could provide an indication of the severity of symptoms of depression for an individual. The index gives a score; above a certain score (10 and above) the person is considered likely to be depressed and below that score the person is likely not to be depressed. The battery of questions was asked at pre-treatment assessment and post-treatment assessment and at 9 months’ follow-up of all the caregivers. The sample sizes of the caregivers are contained in Table 2 above although not all caregivers completed the depression questions. Since these sample sizes were small, no statistical tests of significance could be conducted. The findings provide a direction of the change and further research is needed with larger samples to confirm these results.

Trustworthiness

The research tools were pre-tested and then used in the pilot intervention program in 2016. Some minor changes were made before commencement of the study. Quality control took place in-field during data collection, at the transcription stage, and at the analysis stage where concerns arose, data-checking and re-collecting of data occurred. Transcription and translation checks led to corrections as needed. Hard copies of transcripts of interviews and group sessions were kept in files for the different participant groups and the various phases of the research. These were backed up with digital versions that were saved in a password-protected computer. Similarly, the data that were analyzed via Atlas ti were saved onto computer and hard copies retained in a file for easy access. Minutes of meetings of the research team were also kept. These documents provide a complete history and audit trail of the project from initial conceptualization to completion of the various research reports on the different phases.

Trustworthiness was also enhanced through moderation of the codes assigned using correspondence checking between the research team members (Townsend & De la Rey, 2008). Each researcher individually coded the data and categorized the themes that emerged. One person moderated the coding for consistency, and then all three researchers cross-checked the themes with one another to achieve correspondence. This process was managed using Atlas-ti software. The different sets of analyses were compared to check for similarities and differences, thereby enhancing the confirmability of the data.

Ethics

The research was approved by the primary authors’ university research ethics committee, by the Gauteng Education Department’s District Directors and by each school principal. Participants were informed that participation was voluntary and that the information would be confidential. Caregivers and educators provided written consent to participate in the study while children provided verbal consent. The intervention participants were offered the program as an incentive, but nothing was offered to the non-intervention group who received a cash transfer, free schooling and school lunch daily. The right to the CSG is guaranteed by the Constitution of the Republic of South Africa (1996) and the Social Assistance Act 13 of 2004 (as amended) which is applicable to both the intervention and non-intervention groups. Free schooling is available by law for all children living in poor communities and these children also receive a school lunch (Department of Education, 2006). All participants were provided with relevant referral information.

Results

The findings are presented with respect to each of the dimensions where changes were anticipated as set out in the overall hypothesis above and the hypotheses for each of the dimensions reported on in the results. Only the substantive changes that occurred over time are reported on in this article. Comparisons are made between the intervention and non-intervention groups; similarities and differences are identified; and areas where no changes occurred or where the evidence was inconclusive. Caregiver perceptions of change are presented and children and teacher perspectives are integrated where appropriate.

Family and Caregiver Relations

In line with the overall hypothesis, changes were assessed with respect to child and caregiver relations in terms of communication between the treatment and comparison groups. Of the 30 caregivers who responded to questions regarding child-caregiver and family communication, all (40) participants emphasized positive changes that occurred compared to pre-treatment assessment in communication with children and adults in their family following attendance of the FSI. The most frequent changes mentioned were in the use of positive communication skills (35 out of 40) followed by increased problem-solving behavior (9 out of 40) and active listening (7 out of 40). These changes were sustained at follow-up where 35 out of the 40 participants indicated the use of positive communication practices. Although positive communication was mentioned by some participants in the non-intervention group at end point and follow-up, the number of times this change was mentioned compared to the intervention group was small. Caregivers described the changes that occurred in the intervention group since they started the program and at post-treatment assessment, this is what some of them said:

Since I started there, now I can see changes in them [siblings], we are closer … She listens now, and she is more comfortable speaking to me now, they were always fighting, hitting each other, but now they have become close, the way they speak to each other—Caregiver LS

There has been improvement in communication. Myself and T, I think we have learned to say how we feel, instead of reacting before we speak to each other. That has been the biggest highlight of the programme…. it has been very helpful… I have noticed that he’s more confident, and there is more willingness to tell me how he feels—Caregiver Y-LG

I listen to them. Before I wasn’t listening—Caregiver SM

At 9-month follow up, caregivers continued to report changes in communication. One of the caregivers explained how she communicated with her children about the tensions they experienced with their extended family (Caregiver R) while three other caregivers mentioned how they talked about correcting behavioural difficulties and how to talk to people (Caregiver L; Caregiver P and Caregiver S).

….I have learnt…to avoid the children getting into those problems [with extended family], I would take them aside and sit them down and make them understand that it’s like this and that and you would go and visit your aunt—Caregiver R

When they do something… when they make a mistake I am able to talk to them. When I also make a mistake, they can also talk to me and say ‘mom, I don’t like what you did’—Caregiver L.

Talking to people, on how to talk to people—Caregiver P

Before I never spoke. I knew that when the child does something wrong, you have to beat them. But at least now we do communicate, and there is a change because we also don’t just spend time talking, also playing—Caregiver S.

Good communication is assumed to lay the foundation for positive parenting practices which are hypothesized to improve child caregiver relations. Increased use of skills obtained in the FSI were assessed in terms of the use of praise, consequence management, encouragement of good behavior and the use of alternative forms of discipline. At the end of the intervention caregivers used skills such as praising the child, teaching him/her the consequences of negative behavior, and rewarding and encouraging good behavior. Of the 30 respondents (out of 40) who commented specifically, 20 caregivers illustrated the use of these techniques to varying degrees and attributed these skills to what they learnt in the intervention group.

I used to like to shout at her and when she had done something wrong, I used to beat her up, but now when she has done wrong, I call her and make her take note that she has done something wrong and I ask her what she is supposed to do when she has done something wrong. And then she knows that she should apologize.—Caregiver NM

When a child made a mistake, I won’t say I will punish him next time. Now I have found an alternative that ‘you won’t get what you wanted, yogurt, I will give it to you later, please fix your mistake here’.—Caregiver OM

I praise her if for instance she helped me wash dishes, I would tell her, ‘thank you for washing the dishes… you are getting wise and growing up now’.—Caregiver NS

After the intervention, caregivers were asked how they felt when the program started and how they felt by the end of the program. This open-ended question had been coded and the following emerged from the analysis: In regard to the use of alternative forms of discipline, at end point over two-thirds of caregivers in the intervention group were more able to discipline their children without resorting to physical punishment. They felt more confident about their parenting abilities; they were better able to balance the different responsibilities in their life; and were more hopeful of their child and family’s future. Asked about changes that had occurred in the way they disciplined their children following the program, 20 respondents (out of 24) reported that they no longer beat their children. While it cannot be ruled out that some respondents may have given socially desirable responses, the answers suggest a heightened awareness of the negative impact of these practices. This finding is important given that Richter, Mathews, Kagura and Nonterah (2018) reported that 50% of pre-school and primary school children in the Johannesburg-Soweto area had experienced physical punishment by their parents.

After 9 months, the above sentiments were sustained for seven out of ten participants in the intervention group who continued to express confidence in their parenting capability and remained hopeful for their children’s future. A reduction in corporal punishment was sustained in the intervention group for ten out of the 25 respondents. In the non-intervention group at end point there appeared to be a greater awareness of the inappropriateness of harsh forms of discipline compared to pre-treatment assessment. However, those who did not have access to the intervention were not able to provide concrete examples of the use of alternative forms of discipline. Despite these positive changes in parenting practices for the intervention group at follow-up, eight participants said they felt more overwhelmed by their situation after the program had ended.

Views of child participants (36) on family relations based on children’s drawings depicted their families engaged in various activities such as shopping, baking, watching television and birthday celebrations. There were no important differences in their depiction of family activities and family happiness between pre-treatment assessment and end point for the intervention and non-intervention groups. At 9-month follow up, 19 of the 25 children still drew pictures that depicted happy families and 14 of the child participants had happy memories of the FSI. What is important is that four of the children’s drawings depicted use of corporal punishment by caregivers and other family members, experiences of hostility between siblings and friends and bullying. When triangulating children’s views with that of the caregivers, it is evident that corporal punishment continued for a small number of children. At follow-up two children identified hardships experienced due to poverty and feelings of deprivation relative to other children such as not having money to go on school trips.

Caregivers’ Involvement in Children’s Education

The assumption undergirding the overall hypothesis was that participation in the FSI could improve parental involvement in children’s education assessed in terms of changes in school behavior such as doing homework, enjoyment of school and learning, and improved performance. Accordingly, it was hypothesized that the FSI was likely to improve parental engagement in children’s education compared to the non-intervention group. When asked about differences in school-related behavior that caregivers in the intervention group perceived to have occurred in their children between pre-treatment assessment and endpoint, participants spoke of improvements in behavior and performance outcomes. Enjoyment of school, engagement in schoolwork and improvements in subjects were cited most frequently by intervention group participants with far fewer such references in the non-intervention group before and after the intervention.

There is no doubt that he [child] enjoyed school … this is evident by what he does and how he speaks of his teacher and his class mates … I would say it is because of the program… Caregiver T-LG

He’s happy since the group, he struggled with counting and he didn’t have an interest to do maths, now he can even count using his hands and also take stickers to count with.—Caregiver KM

…. when she got home and we asked her if she has homework, she would say she doesn’t have homework… But now she is able to say ‘dad, here I have homework, please help me where I don’t understand’.—Caregiver AK

Positive caregiver and parental engagement in the child’s school activities appeared to be sustained at follow-up with nine participants out of 40 indicating improvements in the child’s functioning at school compared to their functioning at the start of the program. A few participants in the non-intervention group identified similar improvements in school work which may have been due to other unknown factors.

Perceptions of Educators

Nineteen educators for whom data were available at pre-treatment assessment and endpoint indicated that eight children showed improvements, while one child deteriorated substantially. For the remaining 10 children, there was no apparent change with four out of ten continuing to present with problems and six continuing to function well.

Because they were coached and motivated to attend school regularly, she is no longer absenting herself from school … In terms of her behavior, she has never had any problems. However, she is more talkative now, she reports everything. She discusses whatever they are discussing at [group] meetings—Teacher (reporting on NS)

Since the last time we had the interview with you, there is a drastic change from M, especially academically. I am happy about his performance so far… so far with M I have seen a lot of improvement in his work…, the mother is still young but she is very supportive. She [now] comes to the school. You know she is taking care of him and he is very clean, very neat.—Teacher (reporting on MM)

Since last time I have noticed slight changes, but what I forgot to say in the previous interview that A’s writing is a problem… it’s almost like his letters are back to front’—Teacher (reporting on AM)

I’m not sure if its hyperactivity. It’s not like he has a learning problem. He knows what to do and he’s a good reader, he can identify words… he is all over the place, he doesn’t sit still, he is always fidgeting.—Teacher (reporting on T-LG)

I cannot say he has changed… socially he fights. He beats other children. He swears at them… I even told his mother and she said ‘ma’am I don’t blame him, the situation at home is not well’. So what can I say? This one really we need to monitor him… He is very aggressive.—Teacher (reporting on NM)

In summary, teachers expressed genuine concern for the learners. While some children were doing better following the program, learning difficulties were identified in some instances and continued behavioral challenges were noted with some of the children. There were no referrals to school psychologists by teachers, but these recommendations were however made by the social work facilitators engaged in the FSI. Follow-up interviews were not conducted with teachers.

Social Networks and Social Support

Participation in the FSI was hypothesized to improve the social networks and social support systems of caregivers compared to the non-intervention group. Caregivers were asked if there had been any change in their social networks following their attendance at the Sihleng’imizi group. Of the 32 participants who responded to this question, 15 had plans to keep contact with their Sihleng’imizi group buddy. Ten caregivers out of 32 considered the changes in their networks to be positive and included learning from others, experiencing love and care, enhancing relationships with people, learning other languages, gaining understanding, improving communication, and knowing where to seek help with problems. These aspects would appear to be important advantages derived from the FSI.

With me being the youngest of them all with two kids, I learned a lot from them. We had this one in the group, she was the grandmother… I have learned such a lot from her—Caregiver LS

The love among each other, the caring about us, yes… I like to be with other people, and see how other people are working, and seeing their ideas—Caregiver QJ

Relations with extended family members also seemed to have improved for seven out of 32 caregivers.

It has changed because we can communicate now, they have also been asking me what I am doing with the social workers. You know when you go to the social workers, it means there’s a huge problem. But I explained to them … they ask how it’s going… my older sister can involve herself in the issue because [previously] she was also someone who didn’t care.—Caregiver NR

There was a time where we didn’t get along with them [my family] for many years. So after I started Sihleng’imizi, I thought its useless holding grudges… Let me go to them and apologize, even though I know I didn’t do anything wrong. I went there to talk to them; I firstly spoke with their father…and the father welcomed me…and yesterday I was talking to the mother. And I sat down with her and said as a family there shouldn’t be conflict between us. We talked and she was happy and she said thank you.—Caregiver SM

In the case of three respondents, it was not clear whether there had been any changes following exposure to the program or if there had been satisfactory pre-existing relations with family and neighbors before the program commenced. The positive connections continued at follow-up with 10 participants continuing to keep in touch with other group members.

Financial Capabilities

Participation in the FSI was hypothesized to increase the financial capabilities of the participants compared to the non-intervention group. Improved financial capabilities were assumed to lead to improvements in the material well-being of CSG beneficiary families. Important improvements emerged in the intervention group over time in improving their financial capabilities assessed in terms of increased savings and budgeting behavior, an ability to differentiate between wants, needs and obligations, and awareness of the consequences of loans and indebtedness. These changes were most evident among participants in the intervention group with 22 participants out of 40 indicating changes in savings behavior despite their low levels of income.

It [the group] taught me about saving. As a person, you have to save so that you can be able to do what you want. It helped me not to waste money….it taught me that I should go to a bank [and to keep] my money safe by putting it in a bank, not in the house. It’s not safe.—Caregiver TS

We have always had the belief that for someone to save money, they should have a lot of money.—Caregiver SK

It made me very conscious of what I spend my money on, and how I spend it. And this is an investment … so, instead of buying a dress for R150, I can buy a dress for R50 and take the R100 and put it in the savings for [something]. So my mindset has changed, my way of thinking has changed.—Caregiver T-LG

I didn’t understand what it is to really budget. And when we did the activities we found … we used sticks and stones. And we had to put them, we have R500 for example and we had to use the R500 for the month. We used the strategy of how we are going to pay everyone, how much we are going to save, and how much we are going to have for our needs and wants. So it was very interesting.—Caregiver T-LG

One participant specifically stated that she continues to take loans because she is unable to make ends meet. In two cases, it was not clear whether behavior had changed or remained the same. At follow-up, nine intervention group participants out of 20 continued to save while eight others still applied the knowledge gained in different ways. For instance, two participants indicated that they joined a community savings scheme. Although the non-intervention group participants expressed a desire to save at different points in time over the study period, they were not able to display the same level of knowledge and behavioral changes in financial management when compared to the intervention group.

Nutrition

Increased knowledge and skills of nutrition provided via the FSI was hypothesized to lead to healthier food choices, better meal planning, preparation of balanced meals and weighing of cost considerations with nutritional value when purchasing food compared to the non-intervention group. Nutritional knowledge and skills are associated with better physical well-being outcomes for children and the family as a whole. Caregivers had a fair knowledge of nutrition in the pre-treatment assessment based on interviews conducted with them. There were modest differences in the knowledge of participants between pre-treatment assessment (60 for the intervention group and 71 for the non-intervention group) and the post-treatment assessment (40 participants). There was a slight increase in the number of intervention group caregivers identifying breakfast as the most important meal of the day at end point from 27 to 29 participants. Nutritional value of food was considered most important in choosing or buying food (39 out of 40 participants) rather than cost which was the primary factor at pre-treatment assessment (29 participants). The shifts in the non-intervention group were small. At follow-up the majority of the participants in the intervention group still identified the importance of making healthy food choices (26), having balanced meals and acknowledged the importance of breakfast. However, in the non-intervention group at follow-up, healthy food choices were cited as important for almost all the participants although it was evident from the interview data that many could not identify the elements of a balanced diet. In view of the similarity of the responses between the intervention and the non-intervention groups, it cannot be confidently concluded that the identified changes could be attributed to the program.

Depression Symptomology of the Caregivers

It was hypothesized that the depression scores of the caregivers in the intervention group would be lower at posttest and follow-up compared to the non-intervention group. This section describes the findings from the depression index (CESDR-10) which was administered for all families in the intervention and non-intervention groups. At pre-treatment assessment the depression scores for the intervention (N = 59) and comparison groups (N = 71) were similar (53%). Depressive symptomatology dropped for both groups between pre-treatment assessment and endpoint for the intervention group by 16% (6 out of 38) and by 12% (8 out of 65) in the comparison group. The cause of this decline in both groups is unknown. It is possibly attributable to the Hawthorn Effect, which has been noted in other South African research (Cluver et al., 2018; Rosenburg et al., 2018). Contamination across the intervention and control groups is unlikely as participants were randomly selected and allocated and were recruited individually (not in a group setting).

At follow-up 9 months later, it appears that there was an increase of 3.9% (or 5 out of 20) in depressive symptoms in the intervention group which was possibly due to a lack of support after the program ended. Although the sample for the non-intervention group was smaller (N = 15) in the follow-up study, more people reported having depressive symptoms compared to a year previously when the pre-treatment assessment study was conducted. The lower rates of depression in the intervention group compared to the non-intervention group may have been attributable to the buddy system which they continued to utilize. It is possible that the deteriorating economic conditions in the country and rising levels of poverty and unemployment across urban communities may explain the results. Although changes in the intervention group were sustained over a period of 12 months with a small increase at follow up, the results are inconclusive and should be treated cautiously.

Discussion

In discussing the findings, the following limitations are worth noting: first, it was likely that participants provided socially desirable responses. Second, there were instances where the changes could not be attributed to the intervention. Other interacting environmental factors cannot be ruled out e.g. participant exposure to media messaging about healthy eating habits. Third, the real-life situation of families is precarious and changeable making it difficult to capture these changes and their impacts. A fourth source of bias was that raters were not blind as to whether the drawings were from the pretest or posttest. Lastly, the study showed some similarities between the intervention and non-intervention group in some areas which illustrates prior knowledge of caregivers about parenting. These are viewed as strengths and the level of care that they are providing despite important challenges (Patel, Hochfeld, & Chiba, 2019). The results are however indicative of the overall direction of the changes that occurred.

Combination interventions consisting of cash transfers and a family strengthening intervention improved child and family well-being in specific domains. The hypotheses were confirmed in the following domains namely, Improvements were noted in child-caregiver and family relations; strengthened networks of social support and caregiver engagement in schooling as well as enhanced parenting, financial capabilities and knowledge of healthy food choices and in the buying of food. Although symptoms of depression were reduced at end point, it began to rise after the intervention ended. An increase in depressive symptomatology was evident in both the intervention and non-intervention group, albeit to a smaller extent 9 months after the intervention. This finding may be explained by extraneous factors such as rising rates of unemployment and poverty nationally, which is most marked among women CSG beneficiaries with low levels of education and skills (Patel et al., 2017). Since the evidence was inconclusive, the hypothesis regarding improvements in the mental health of the caregivers was not confirmed. Similarly, the hypothesis regarding improvements in the nutritional knowledge and skills of the caregivers were not confirmed. Further research with larger samples will be needed to assess the direction of the latter changes.

Based on these findings, the offering of a family strengthening intervention to complement the CSG is proposed to accelerate well-being outcomes for beneficiaries, particularly in poor urban communities in South Africa, with high uptake of the grant. There is need to move beyond the provision of cash transfers only to incorporate the concept of care to achieve more comprehensive social outcomes for poor families with children. Social investments of this kind in the foundation years of schooling have the potential to improve outcomes in the longer term, but this hypothesis needs further testing and evaluation. Further, it cannot be assumed that similar results may be achieved in other contexts such as rural areas or child and family welfare agencies with different mandates. Also, it is not known for which beneficiary groups this type of combination intervention is likely to be more effective than the provision of a cash transfer only. The qualitative design of the study provided rich data on the positive direction of the changes that could be anticipated as a result of the program. However, it is limited in identifying the magnitude of the changes and for which groups of people the intervention will be most beneficial. In order to achieve this goal, an experimental design such as a randomized control trial may be needed. This approach will need to be augmented with qualitative research to delve deeper into the underlying meanings and understandings of the participants of the outcomes, to explore anomalies in the findings and contextual variations in how these might impact on program outcomes.