Why parenting skills alone cannot carry early child development in low-income South African families

Introduction

Most early childhood programmes in low-income South African communities start from a simple idea: teach the caregiver, usually the mother, to talk, play and respond better, and the child will do better. There is some evidence behind that. In a township trial, eight weekly group sessions of book-sharing led caregivers to become less harsh and more sensitive, and their toddlers’ language and attention improved (Dowdall et al., 2021). The trouble is how much else is going on around the caregiver, as the birth cohorts show. Among 973 women followed from pregnancy to five years after birth in the Drakenstein Child Health Study, a persistent pattern of psychological distress was predicted by trauma exposure, smoking and gestational diabetes, and partner support protected against it (MacGinty et al., 2025). Training a woman in responsive talk while leaving that picture untouched is a thin kind of help.

This paper argues that parenting skills, important as they are, cannot carry early child development on their own in these families, and that the South African evidence points to three things programmes have to take on as well: caregiver distress and trauma, violence close to young children, and the near absence of fathers from both research and services. Parenting programmes do not fail on this evidence (Dowdall et al., 2021). They work inside a set of conditions that limit how far they can go. I take caregiver distress first, then violence, then fathers, and then ask what the trial evidence on programmes can and cannot support, drawing on cohort and trial studies from the Western Cape, the Eastern Cape and Soweto (MacGinty et al., 2025; Tsunga et al., 2024).

Caregiver distress and what children inherit from it

The trajectory work in Drakenstein is a good place to begin because it replaces the single snapshot of postnatal depression with a longer view. MacGinty et al. (2025) identified four patterns of maternal distress over the first five years of a child’s life: persistent symptoms, antenatal symptoms only, late postnatal onset, and low symptoms. Early and recent trauma exposure predicted the persistent group, whereas trauma during or before pregnancy predicted the antenatal group and postnatal trauma the late-onset group. Partner support was protective, which echoes work in Soweto linking partner involvement to mothers’ mood after birth (Drysdale et al., 2021). The authors call for social change around gender-based violence, which is a striking conclusion for a paper about mothers’ mental health, and I think it is the right one. Distress in these women seems tied to what has happened to them, and that matters for how services respond.

What it does to children is harder to pin down. Rayport et al. (2025) studied 39 mother-child pairs in the Western Cape and measured toddlers’ attention skills with eye-tracking at 15 months. Toddlers of mothers with low trauma exposure showed the expected differences in attention across conditions. Toddlers of mothers with high trauma exposure did not, which the authors tentatively read as a possible adaptation rather than a deficit. The sample is tiny and the main interaction was only partly borne out by the pairwise tests, so I would not build a theory on it. Still, it is a useful reminder that maternal trauma may reach the child through quite subtle routes, and that a parenting group built around observable warmth and talk may not touch them (MacGinty et al., 2025).

Put the three studies of distress side by side and they do not agree neatly. MacGinty et al. (2025) show that trauma predicts a persistent course of maternal distress, which is what one would expect. Rayport et al. (2025) find that the toddlers of highly traumatised mothers look different from the others, but not necessarily worse. Sherr et al. (2025) find that mothers with more than one child report more stress, yet their children, who have a sibling, score higher on cognitive development. If distress mapped straight onto child outcomes, none of that would happen. It suggests the path from a mother’s state to a child’s development runs through other things as well, such as who else is in the house, what childcare exists and what the child sees day to day. That is a more hopeful reading than a simple harm story, and also a more demanding one for programme designers, because it means there are several places to intervene and no single lever.

Young mothers carry an extra load. Sherr et al. (2025) drew on a cohort of 1,017 mothers aged 10 to 24 and their children in the Eastern Cape. Mothers with more than one child reported higher parental stress and lower social support than mothers with one. The surprise was on the child side: children who had a sibling scored higher on cognitive development, and formal childcare support was linked to better child outcomes. Because the study is cross-sectional, it cannot say whether siblings help, or whether families with more stable circumstances simply go on to have a second child. The authors lean towards delaying a second pregnancy and expanding early childcare. That reading is plausible, but the stress and support findings point somewhere else, namely towards the mother’s own circumstances as the thing to change (Cluver et al., 2025; Sherr et al., 2025).

Violence close to young children

Violence is not a background condition for these children. It is something most of them meet early. In the Drakenstein cohort, 83% of 978 children had been exposed to some form of violence by age four and a half, according to parent report (Tsunga et al., 2024). Higher overall exposure, domestic victimisation and witnessing community violence were each linked to more internalising and externalising problems at age five. Poly-victimisation was linked to externalising behaviour but not clearly to internalising behaviour. Evidence for witnessing domestic violence was weaker, and community victimisation showed no association. Parents were the reporters, which may mean some exposure goes uncounted, especially if the parent is the source of it or is simply unaware. The Soweto cohort relied on the children’s own reports instead (Muchai et al., 2025).

The longer picture comes from Soweto’s Birth to Twenty Plus cohort. Muchai et al. (2025) followed children’s reports of physical and sexual violence between ages five and 18. Roughly two thirds of children (65.1%) fell into a group whose physical victimisation was limited to adolescence, but 34.9% were in a chronic, increasing group. Boys were more likely to be in that chronic group (adjusted odds ratio 1.67), and living in a middle rather than low socioeconomic household appeared protective. Having a mother with at least secondary education, compared with higher education, was associated with higher odds of chronic victimisation, an odd result that I would treat carefully until it is replicated. Even the few early-life factors that predicted these trajectories, household socioeconomic position among them, sit outside what a short group programme can alter (Tsunga et al., 2024).

Cluver et al. (2025) add something unusual because they could compare which services seemed to protect against several risks at once. Using data from 1,044 adolescent mothers in South Africa, they found that non-violent parenting was associated with lower odds of suicidality (odds ratio 0.20), mental health distress (0.44) and intimate partner violence (0.27). Food security was associated with being in school or work, and respectful clinics with lower distress. When all three were present, suicidality fell from 13% to 2% and intimate partner violence from 22% to 5%. These are associations in an observational cohort, so the numbers describe patterns, not guaranteed effects. Even so, putting parenting beside food and health services in a single list is telling. The authors themselves recommend economic support including cash transfers, parenting programmes and adolescent-responsive healthcare together, not one of them alone (Cluver et al., 2025).

Fathers who are missing from the data

If caregiver distress is partly about partner support and violence, fathers matter more than service design has allowed. Steventon Roberts et al. (2025) described fathers of children born to adolescent mothers in the Eastern Cape, using mothers’ reports. About 40% of the fathers had been adolescents themselves at the birth, and only 19.5% were counted as involved in childcare. Involved fathers were far more likely to be in a relationship with the mother (74.8% against 47.2%) and to live with her and the child (14.7% against 3.7%). Mothers with poor mental health were more likely to report domestic violence by the father (8.2% against 3.3%) and arguments about money (30.0% against 17.0%). The picture is one of fathers who are young, mostly not living with the mother and child, and entangled in the mother’s distress in ways that could go either direction (Drysdale et al., 2021).

It would be easy to conclude that more father involvement will improve child outcomes, and I do not think the South African data support that yet. In the same Eastern Cape study, composite scores for child cognitive development did not differ by father age or involvement (Steventon Roberts et al., 2025). In Soweto, Drysdale et al. (2021) studied 212 mother-baby pairs inside a larger trial. Only 43% of partners attended an early ultrasound, 33% attended antenatal care and 1% were at the birth. Father involvement was directly associated with postnatal depression, but not with birth weight or exclusive breastfeeding at six weeks. So the clearest link runs through the mother’s mood, not through the infant’s body. That fits the logic of the first section, where partner support protected mothers from persistent distress (MacGinty et al., 2025).

There is also a measurement question that tends to be skipped. If father involvement is judged by child outcomes alone, South African studies will keep finding little, because the child measures used so far were cognitive scores or birth outcomes (Steventon Roberts et al., 2025; Drysdale et al., 2021). The outcomes that did move were on the mother’s side, and the same Eastern Cape data tie maternal mental health to domestic violence by the father. A study that asked whether a father’s presence lowers violence at home, or whether it lifts mothers out of persistent distress, would be asking a question the existing data hint at without testing. I would expect the answer to depend on who the father is, since a young man who is himself violent or absent is hardly a resource (Steventon Roberts et al., 2025).

What happens when programmes try to include fathers? Jeong et al. (2023) reviewed 33 evaluations of father-inclusive interventions in low- and middle-income countries, covering young children under five. Most targeted fathers together with their partners, mainly to improve nutrition and health. Effects were mostly positive for maternal, paternal and couple outcomes, with more variation for child outcomes, and the study designs were often weak. The review did not focus on South Africa, so I treat it as a direction rather than a verdict. Its message to South African programme designers is nonetheless fairly clear. The evidence on fathers is not strong enough to promise child gains, but it is strong enough to stop treating them as an afterthought, particularly given the low involvement reported by Drysdale et al. (2021).

What parenting programmes can and cannot do

Return to the book-sharing trial. Dowdall et al. (2021) randomised 140 caregivers and their 21 to 28 month old children in a low-income township to eight weekly group sessions or a wait-list. They assessed families before, immediately after and six months later. The programme improved child language and attention, but not behaviour problems, prosocial behaviour or theory of mind. Caregivers became less verbally and psychologically harsh and showed more sensitivity and more complex cognitive talk. That is a real gain at low cost, and nothing here should be read as an argument against such programmes. It is, though, a narrow set of effects. Behaviour problems of the kind Tsunga et al. (2024) linked to violence exposure were not among the outcomes the programme improved.

There is a practical side to the choice of outcome as well. The book-sharing trial measured what a book-sharing programme should change, namely talk, attention and caregiver sensitivity (Dowdall et al., 2021). The adolescent-mother studies, by contrast, point to outcomes that no parenting group would be expected to shift directly, such as food security, school or work engagement and access to formal childcare (Cluver et al., 2025; Sherr et al., 2025). A programme can be perfectly well designed for its own outcomes and still sit too far from the things that decide whether a young family copes.

A second limit concerns who the programme reaches and what it assumes. A caregiver with persistent distress, a history of trauma or an absent partner may find it harder to attend eight sessions, to practise at home or to hold on to new habits, though none of the studies I read tested this directly (MacGinty et al., 2025; Dowdall et al., 2021). The same holds for fathers, who were not the target in the book-sharing trial, and for the young mothers described by Sherr et al. (2025), whose stress and low support suggest they carry more than any weekly group can lift. I am inferring here from the pieces, and the inference should be tested in trials that measure caregiver distress as a moderator.

A third point is about the evidence base itself. Two of the studies discussed here use the same Eastern Cape cohort of young mothers (Sherr et al., 2025; Steventon Roberts et al., 2025), and a third studies adolescent mothers with overlapping authors (Cluver et al., 2025). Two come from Drakenstein, two from Soweto, and the others from one township trial and one small Western Cape sample. Most are cross-sectional or observational, and most rely on maternal report. The agreement across them is encouraging, but it comes from a narrow set of places and designs. We know less about rural Limpopo, informal settlements in Gauteng, or families raised by grandmothers than the confident language of policy documents would suggest.

Conclusion

Parenting programmes belong in South African early childhood services. The book-sharing trial shows that responsiveness can be taught to caregivers in a low-income township and that toddlers benefit in language and attention (Dowdall et al., 2021). The wider evidence, however, indicates that this is a partial remedy. Caregiver distress follows trauma and weak partner support (MacGinty et al., 2025), most children in the Drakenstein cohort had met violence by the age of five (Tsunga et al., 2024), and fathers are largely absent from both the data and the services (Steventon Roberts et al., 2025).

In practice, that points to programmes that screen for caregiver distress and link it to care, that treat violence in the home and neighbourhood as part of what the programme is up against, and that make a real attempt to include fathers while accepting that the child-level gains are unproven (Jeong et al., 2023; Drysdale et al., 2021). It also points to pairing parenting support with food security and respectful health services, as the adolescent-mother cohort suggests (Cluver et al., 2025). None of this is cheap, and nothing guarantees the combined package works better than its parts. What the evidence does support is the weaker claim that skills training alone leaves too much untouched, and that is reason enough to test the wider package properly.

Reference list

Cluver, L., Jochim, J., Sidloyi, L., Armstrong, A., Gulaid, L., Banougnin, B., Saliwe, B., Steventon Roberts, K., Thabeng, M., de Graaf, K., & Toska, E. (2025). Accelerators for improved health among adolescent mothers in South Africa: HIV and violence prevention, sexual reproductive health and education success. BMJ Global Health, 10(6), Article e017614. https://doi.org/10.1136/bmjgh-2024-017614

Dowdall, N., Murray, L., Skeen, S., Marlow, M., De Pascalis, L., Gardner, F., Tomlinson, M., & Cooper, P. J. (2021). Book-sharing for parenting and child development in South Africa: A randomized controlled trial. Child Development, 92(6), 2252-2267. https://doi.org/10.1111/cdev.13619

Drysdale, R. E., Slemming, W., Makusha, T., & Richter, L. M. (2021). Father involvement, maternal depression and child nutritional outcomes in Soweto, South Africa. Maternal & Child Nutrition, 17(S1), Article e13177. https://doi.org/10.1111/mcn.13177

Jeong, J., Sullivan, E. F., & McCann, J. K. (2023). Effectiveness of father-inclusive interventions on maternal, paternal, couples, and early child outcomes in low- and middle-income countries: A systematic review. Social Science & Medicine, 328, Article 115971. https://doi.org/10.1016/j.socscimed.2023.115971

MacGinty, R., Hoffman, N., Koen, N., Zar, H. J., Stein, D. J., & Pellowski, J. (2025). Longitudinally investigating patterns of maternal psychological distress in a South African birth cohort. BMC Public Health, 25(1), Article 3409. https://doi.org/10.1186/s12889-025-24445-x

Muchai, L., Naicker, S., & Kagura, J. (2025). Early life factors associated with childhood trajectories of violence among the Birth to Twenty-Plus Cohort in Soweto, South Africa. PLOS One, 20(11), Article e0294207. https://doi.org/10.1371/journal.pone.0294207

Rayport, Y. K., Hu, Y., Gimenez, L. A., Du Plessis, C., Odendaal, H. J., Fifer, W. P., & Shuffrey, L. C. (2025). Maternal experiences of trauma and toddler multisensory attention skills in a South African community cohort. Infancy, 30(2), Article e70017. https://doi.org/10.1111/infa.70017

Sherr, L., Haag, K., Steventon Roberts, K., Cluver, L., Jochim, J., Saal, W., Langwenya, N., Wittesaele, C., Tolmay, J., Marlow, M., & Toska, E. (2025). Not one but two: Cross-sectional associations among repeat pregnancy, maternal mental health and child cognitive outcomes among adolescent and young mothers in South Africa. BMJ Global Health, 10(10), Article e019909. https://doi.org/10.1136/bmjgh-2025-019909

Steventon Roberts, K., Smith, C., Cluver, L. D., Toska, E., Kelly, J., Thomas, A., Tolmay, J., Marlow, M., & Sherr, L. (2025). Men matter: A cross-sectional exploration of the forgotten fathers of children born to adolescent mothers in South Africa. BMJ Open, 15(7), Article e092723. https://doi.org/10.1136/bmjopen-2024-092723

Tsunga, L., Heron, J., Lake, M. T., Halligan, S. L., Malcolm-Smith, S., Hoffman, N., Zar, H. J., Fraser, A., Stein, D. J., & Donald, K. A. (2024). Exposure to violence and mental health outcomes among pre-schoolers in a South African birth cohort. Research on Child and Adolescent Psychopathology, 52(10), 1635-1646. https://doi.org/10.1007/s10802-024-01211-y