Sponsorship, HEART, and ELM Endline Evaluation Results

Malawi 2016


This study aimed to estimate the impact of ELM and HEART programming and understand how the programs benefit children differently. Save the Children conducted a quasiexperimental study with 553 children at 40 Community Based Childcare Centers.

Author: Jonathan Seiden, Francis Mabeti
Organization: Save the Children
Date: March 24, 2017
Country: Malawi


Save the Children implemented a year-long program of Healing and Education through the Arts (HEART) and Emergent Literacy and Math (ELM) at hundreds of Community Based Childcare centers in Save the Children’s Zomba Sponsorship area in Malawi. HEART is an arts-based approach to providing psychosocial support for children affected by serious or chronic stress. ELM is a program designed specifically to improve children’s early literacy and math skills.

Zomba district is in the rural south of Malawi, the area of the country with the highest rates of poverty. Zomba also has some of the highest rates of HIV/AIDs in the country. One in four adults is infected in Zomba, compared with fewer than one in ten nationally. The result is that there are many orphans and vulnerable children (OVCs) who have lost one or both parents to HIV/AIDS in Zomba. Up to 27% of children in Save the Children’s programming area have been identified as OVCs.

Sample and Design

In order to estimate the impact of the ELM and HEART and understand how the programs benefit children differently, Save the Children conducted a quasi-experimental study with 553 children at 40 Community Based Childcare Centers (CBCCs) in the Zomba Sponsorship area. All 40 CBCCs received the basic Save the Children quality ECD package. In addition to the basic package, 10 CBCCs received the HEART program,10 CBCCs received the ELM program, and 10 CBCCs received both ELM and HEART programs.

Children were assessed using the International Development and Early Learning Assessment (IDELA). The IDELA provides an overall early learning and development score and measures of domain-level skills in Motor, Social Emotional, Early Numeracy, and Early Literacy. During the baseline study in September 2015, 553 primarily five-year-old children were assessed. The sample represented the gender distribution at CBCCs and was quite even: 47% boys 53% girls. This reflects total enrollment at all CBCCs – slightly more girls than boys. Children in the four study groups appeared similar in terms of their early learning and development. There were no significant differences on any observed variables.

At the endline in June 2016, enumerators attempted to re-survey the same children they interviewed at baseline. Unfortunately, only 79% of the original children were able to be found and re-interviewed at endline. While there was not an overall different rate of attrition between the study groups, the reasons for attrition were recorded, and there were significant differences between the study groups.


The study found that children in the three treatment groups (that received ELM, HEART, or ELM + HEART) all had large gains in early learning and development above and beyond children in the comparison group. As Figure 1 shows, children in the HEART, ELM, and HEART + ELM groups gained an average of 18 percentage points on Total IDELA score whereas children in the comparison groups gained only 5 percentage points. This was a significant difference and is suggestive of a large programmatic impact.

While there were large differences between each of the three intervention groups and the comparison group, there were no differences in gains between the three intervention groups. The gains among children in ELM only, HEART only, and HEART + ELM were statistically identical for all domains. Figure 2 displays the domain-level results for each of the intervention groups and the comparison group.

In terms of equity findings, little demographic data was collected. However, there were no gender gaps at baseline or endline or any differences in the amount that girls or boys gained.


While findings are suggestive of a large programmatic impact, as a quasi-experimental study, they must be considered as correlational in nature. There may have been important unobserved baseline differences that bias results. In addition, differential attrition may have biased the endline sample in difficult to understand ways.


Conclusions and Recommendations

The positive findings of this study support both the theories that HEART and ELM can support children’s early learning and development. However, the lack of any significant differences of the impact of these programs is a puzzling result. Especially because HEART primarily targets psychosocial skills and ELM primarily targets literacy and math skills, the results are not consistent with expected programmatic outcomes.

Further investigation into potential contamination between study groups and the implementation of different programming models will be helpful to explain these findings. In the future, additional studies into the comparative effectiveness of HEART and ELM programming may give a clearer picture of how they can support children’s early learning and development in fragile situations.