The learning crisis is an equity crisis on a global scale, in which low learning levels are both influenced by and contribute to severe global inequality. While the vast majority of children in high-income countries acquire basic literacy and numeracy skills, the vast majority of children in low-income countries do not (UNESCO, 2025). Much of the focus on low learning outcomes has been on this disparity between poorer and wealthier countries (Crouch et al., 2021).

While it is also important to consider inequality within countries, initial research on this topic finds that low learning levels are typically a broad, systemic challenge rather than one driven mainly by inequalities between different groups or individuals (Crawfurd, 2021; Crouch et al., 2021). Silberstein (2021), for example, uses household survey data from 18 countries to explore foundational literacy and numeracy skills and potential sources of inequality and exclusion. Differences between groups – such as between urban and rural children or between boys and girls – account for relatively little of the gap between children’s current performance and a scenario in which all children have mastered basic literacy and numeracy skills.

Modeling an estimate of full parity between rich, urban boys and poor, rural girls would close only 30% of the shortfall in universal literacy in the average country, leaving more than 45% of children still unable to read and understand a simple story (Silberstein, 2021). In other words, closing existing gender, socioeconomic status and other gaps would not be enough to achieve universal foundational skills.

These findings indicate that general reform is critical and that we need to dramatically improve average learning levels across low- and middle-income countries. If the learning crisis affects the majority of children, then we need solutions that can reach the majority of children. Yet how we approach this challenge might have significant consequences.

One approach would be to target the majority of children, or even the children for whom learning gains might be the easiest to achieve, in order to make the most of limited resources. This line of thinking can also be applied to schools: to reach the most children, we might start with the schools and regions where we can most efficiently improve learning, not the most remote or rural, which might take longer, cost more, and face more obstacles.

But what is at stake if we take this approach? If we ignore equity in our current work, do we run the risk of creating a future equity crisis?

An alternative approach would be to design interventions specifically for the children who are most behind and to address scaling by centring the most marginalised and disadvantaged learners. This could help avoid a situation where we make some gains in learning but deepen inequality, further disadvantaging the children most in need.

It is possible that either approach could yield unexpected results. General reforms might turn out to be most effective for the children who are furthest behind. Targeting the most disadvantaged children might improve outcomes for all children and prove less expensive than anticipated. But if we don’t investigate inequality in our current work, we won’t know. We may miss out on possible discoveries, and we may also risk celebrating progress that comes with costly setbacks.

The most important starting point is to learn from progress thus far and examine how existing interventions are affecting equity. Inspired by research in Malawi, we have been looking into the impacts of structured pedagogy and targeted instruction interventions. Our results suggest that we are not paying nearly enough attention to equity.

Malawi’s curriculum reform worked on average but not for the poorest or those who were initially most behind

Malawi faces a critical learning crisis. Only 18.9% of children aged 7–14 demonstrate foundational literacy skills, and only 12.6% demonstrate foundational numeracy skills (UNICEF, 2023). In response, the Ministry of Education, Science and Technology has invested significantly in improving learning outcomes, including through the National Mathematics Curriculum Reform (formerly the National Numeracy Programme), a structured pedagogy programme designed to improve the teaching of mathematics in primary schools.

A randomised controlled trial of the Curriculum Reform programme found that it improved average numeracy outcomes (Nchadze et al., 2025). In many ways, the programme can be considered a success and an important part of Malawi’s commitment to improving foundational skills. However, the trial also found that while the Curriculum Reform improved learning outcomes on average, it did not benefit the poorest children or those who were initially the furthest behind (Nchadze et al., 2025). Our additional analysis estimates that the Curriculum Reform nearly doubled the learning gap between poor and non-poor children and increased the learning gap for the lowest-performing learners by approximately 20%.

As part of the FCDO SCALE Programme, we are currently working with partners to determine whether a targeted instruction approach to the country’s remedial education programme could be used to complement the Curriculum Reform and improve outcomes for the children not currently benefitting from it. When we designed this project, one of our key questions was whether Malawi’s experience was specific to the country’s context or whether it reflected a broader pattern. Were the gaps observed in Malawi linked to the Curriculum Reform’s design or to certain contextual factors? Or could successful reforms in other countries also be improving average learning outcomes while exacerbating existing inequalities?

What other studies tell us about effects on inequality

To investigate these questions, we analysed a sample of 22 structured pedagogy studies and 19 targeted instruction studies. These studies were identified by the What Works Hub for Global Education as part of its work on evidence synthesis (Kaffenberger et al., 2026). We focused on the experimental and quasi-experimental studies and reviewed whether they examined how impacts varied across different subgroups; where they did, we also reviewed what those analyses found.

In total, 15 of the 22 structured pedagogy studies included some type of analysis by subgroup. The most prevalent analysis focused on gender. Eight studies analysed differential effects between boys and girls. Of these, four found no differential effects between girls and boys. Two studies found higher impacts on girls; one found higher impacts on girls only in specific language or cluster contexts; and one had mixed findings (girls benefited more on some subtests, boys on others). These results are somewhat encouraging in that there doesn’t seem to be a strong trend towards differential effects by gender. However, because fewer than half of the studies analysed impact by gender, it is hard to draw any conclusions.

More broadly, a critical limitation is that the studies did not analyse the same sources of heterogeneity, making it difficult to compare results across studies. Each of the following factors was analysed and reported by only one study: school infrastructure, class size, school average test score, student absenteeism, age, coach-to-school ratio and socioeconomic status. It is hard to know whether these studies investigated other sources of heterogeneity but did not report it, and whether their choices on what sources of heterogeneity to report were driven by what was deemed most important for a particular setting, or whether each study simply took a different approach to subgroup analysis. What is clear is that we cannot identify many trends across such a wide variety of factors.

Only five of the 22 structured pedagogy studies analysed differential effects based on baseline ability. Two of these found no differential effects by baseline skills; one found that effects were concentrated in the middle of the distribution (10th–45th percentile baseline scores); and two found that children with higher baseline scores benefited more from the programme in question. If Malawi’s Curriculum Reform were added to this sample of studies, half of the structured pedagogy studies analysing impact in relation to baseline ability would find that programme impacts were concentrated among students with higher initial learning levels, thus exacerbating existing inequality. This might suggest that Malawi’s experience is not unique. However, because less than a quarter of the studies include this analysis, this is not a definitive conclusion.

Turning to the studies of targeted instruction interventions, we reviewed 19 studies of Teaching at the Right Level, Catch-Up Clubs and related remediation programmes. As with structured pedagogy, the most common subgroup analysis focused on gender, with 14 of the 19 studies analysing differential effects between girls and boys. One study found higher impacts on boys, five found no differential effects between girls and boys, and nine found higher impacts on girls (six of which were focused only on literacy or oral skills). It is notable that across the structured pedagogy and targeted instruction studies, differential effects seem to favour girls. This might be an important trend to investigate further, as girls are increasingly showing higher foundational learning outcomes than boys in a growing number of low- and middle-income countries (Welmond & Gregory, 2021; Buhl-Wiggers et al., 2021) and key stakeholders, such as teachers, struggle to explain this trend (Kabay & Isika, 2026).

Twelve of the 19 targeted instruction studies reported differential effects in relation to baseline ability, and only five reported differences based on poverty or socioeconomic status. Encouragingly, of the studies that analysed differential effects in relation to baseline ability, five found no differential effects across baseline skills, and seven found positive effects in at least one domain for children who were initially lagging behind. Similarly, one study that reported impact by socioeconomic status found that children with lower socioeconomic status benefited slightly more. As with structured pedagogy, small numbers of studies investigated additional factors, such as ethnicity, grade level, parents’ literacy, school infrastructure and school average score.

What this means for Malawi and the field

We are encouraged by the results of the research on targeted instruction and by the fact that these remedial education approaches may have potential to improve outcomes for children who are furthest behind. After working with partners to pilot the targeted instruction programme in Malawi, we are now designing an impact evaluation to test whether the programme can improve outcomes for all children. This represents just one approach to addressing inequity, and we hope other projects will explore different approaches. Across programme design, implementation and research, we need to consider key questions such as how to design for equity from the start and whether sequencing should be an important part of education reform. One area for exploration is whether any single intervention can address all challenges or whether complementary approaches are needed.

The most important finding of our evidence review is that we need to draw more attention to equity. With so few studies investigating key sources of inequity, we don’t know whether the current solutions we are championing might be exacerbating existing inequities. As a first step, our team is conducting additional analysis of these same studies to further investigate impact by key subgroups and explore other concerns, including whether the intersection of factors such as gender and socioeconomic status warrants further investigation and whether we should be exploring inequity at the school or regional level rather than individual level.

Moving forward, one of our primary recommendations is that researchers should investigate whether the impact of interventions varies in relation to children’s initial learning levels. In every context, the specific factors that influence children’s learning and define disadvantage or marginalisation will vary. It is also important to recognise how certain factors intersect, and across settings, actors need to better understand what affects children’s learning and how. That said, focusing on children with the lowest initial learning levels offers a clear, universally relevant way to foreground equity in efforts to improve learning, while still recognising the complexity of these issues.

To truly realise the transformative potential of education and improve foundational skills, we must ensure that efforts to improve learning outcomes do not further disadvantage the children who are already most behind.

References

Buhl-Wiggers, J., Jones, S., & Thornton, R. (2021). Boys lagging behind: Unpacking gender differences in academic achievement across East Africa. International Journal of Educational Development, 83, Article 102382. https://doi.org/10.1016/j.ijedudev.2021.102382

Crawfurd, L. (2021). Accounting for repetition and dropout in contemporaneous cross-section learning profiles: Evidence from Rwanda. International Journal of Educational Development, 85, Article 102443. https://doi.org/10.1016/j.ijedudev.2021.102443

Crouch, L., Rolleston, C., & Gustafsson, M. (2021). Eliminating global learning poverty: The importance of equalities and equity. International Journal of Educational Development, 82, Article 102250. https://doi.org/10.1016/j.ijedudev.2020.102250

Kabay, S., & Isika, V. K. (2026). Teachers’ perceptions of gender and education during the return to school in Kenya. Comparative Education Review, 70(2), 468–490. https://doi.org/10.1086/740457

Kaffenberger, M., Angrist, N., Hwa, Y.-Y., Kayton, H. L., Jukes, M., & Stern, J. (2026, January). Core components of Teaching at the Right Level (What Works for Global Education Core Components Synthesis 2026/001). https://doi.org/10.35489/BSG-WhatWorksHubforGlobalEducation-WP_2026/001

Nchadze, L., Açıkol, A., French, E., Kadzamira, E., Pugatch, T., Chiwaula, L., & Meke, E. (2025). Malawi National Numeracy Programme evaluation: Endline report. Oregon State University; Center for Educational Research and Training, University of Malawi; University at Buffalo. https://iati.fcdo.gov.uk/iati_documents/D0007393.pdf

Silberstein, J. (2021). Measuring, visualising, and simulating solutions to the learning crisis: New evidence from learning profiles in 18 countries. (RISE Insight Series No. 2021/029). Research on Improving Systems of Education. https://doi.org/10.35489/BSG-RISE-RI_2021/029

UNESCO. (2025). International Literacy Day 2025: Promoting literacy in the digital era. https://unesdoc.unesco.org/ark:/48223/pf0000395395

United Nations Children's Fund (UNICEF). Malawi - Multiple Indicator Cluster Survey 2019-2020, Round 6 (MICS 2019-2020). Ref: MWI_2019_MICS_v01_M. Downloaded from https://microdata.worldbank.org/index.php/catalog/4286 on April 2025

Welmond, M. J., & Gregory, L. (2021). Educational underachievement among boys and men. World Bank Group. https://documents1.worldbank.org/curated/en/111041644611110155/pdf/Educational-Underachievement-Among-Boys-and-Men.pdf