This blog cross-posted with permission from the Center for Global Development's website where it was published first on January 15, 2025.
It was authored by Jonathan Stern, Christelle Saintis-Miller and Rachel Jordan. Photo: Kindergarten student in Senegal. (2023) Patricia Esteve for USAID/RELIT
Learning at Scale: One year later
In October 2023, the Learning at Scale study officially came to a close. The study was designed to investigate factors contributing to successful improvements in foundational learning outcomes in effective, large-scale early grade literacy and numeracy programs. These programs are typically in short supply in low- and middle-income countries (LMICs) and where they do exist, there has been limited research into how they can be used to inform other programs. The goal of Learning at Scale was to draw on the successful approaches and characteristics of these programs to provide policymakers and development practitioners with evidence-based strategies for improving instruction (and student outcomes) across contexts.
This research was implemented in two phases: the first phase examined commonalities of eight effective literacy programs working at scale with findings on effective instructional practice, teacher support and system support; the second phase examined commonalities of six effective numeracy programs working at scale (including two government-led programs) with a final international webinar hosted by CGD.
In this blog, we look back at the Learning at Scale study as a whole and reflect on the lessons learned, as well as areas that need further research.
Key takeaways from the Learning at Scale study
Large donors have primarily focused on literacy but smaller donors are making strides in numeracy.
The United States Agency for International Development (USAID) has been the largest investor in improving learning at scale for several years, and they served as the donor for six of the eight Literacy at Scale programs. Under Numeracy at Scale, however, selected programs tended to be supported by a different set of donors: such as the Japan International Cooperation Agency (JICA) or smaller, local foundations. This trend shows a prioritization of USAID on funding literacy programs, though we hope to see more balance in the coming years with a broader focus on foundational literacy and numeracy (FLN).
Research designs to estimate impact at scale require more investment.
Identifying programs for inclusion in the study was no easy task. Several large-scale donors simply had no candidate programs with both real scale and rigorous evaluations of learning outcomes. Identification of causal impacts in large-scale (particularly national-scale) programs can be complex but without evidence of what works, we are navigating without a map.
Literacy programs tend to follow similar approaches, while numeracy programs show diverse paths to success.
With greater investments in scaling, it is no surprise that successful literacy interventions tend to use common approaches for improving classroom instruction and supporting teachers. In numeracy programs, there were certain evidence-based practices found across all successful programs in the study but also considerably more variation in the approaches used. We would expect more commonalities as numeracy at scale becomes a higher priority.
Successful programs begin with the building blocks but don’t forget about higher order skills.
In both phases of the study, we found that having sufficient time to spend on teaching the building blocks of reading/math is essential. Successful literacy programs began with basic reading skills (typically including a phonics-based approach) while students engaged regularly with improved reading materials and spent significant time reading (leading toward greater reading outcomes). Quality math instruction followed learning progressions and moved students from concrete representations to pictorial representations, and then to the abstract concepts themselves. Additionally, building students higher-level problem-solving skills is critical, as is asking students to provide mathematical explanations and justification of their answers.
If you don’t know where students are, you don’t know what they need.
We found limited use of classroom assessments to understand students’ individual needs throughout the Learning at Scale study. However, emergent evidence suggests that use of these formative assessments to address differentiated student needs may increase student learning and equity in the classroom. Successful remediation programs (such as teaching at the right level) and other targeted instructional programs should be used to inform adjustments in core instructional approaches for assessment-informed instruction.
There must be a focus on making teachers’ lives easier.
Teachers in LMICs are facing significant challenges and these interventions are all asking them to do more, not less. Teachers need to be supported to build both pedagogical and content knowledge, through structured guidance (including teachers’ guides) and on-going support in the classroom. Another key finding from both phases of the study was a shift from classroom inspection to friendlier and more collaborative mentoring and coaching through the system. Paired with current work on reducing teachers’ workloads by focusing on high-impact teaching strategies, these approaches are all designed to make it easy for teachers to implement new practices in their classrooms.
Interventions should be built for scale rather than trying to scale up everything that works as a pilot.
Pilots are essential for understanding what works and for testing out new innovations. However, not everything that works at a pilot stage is meant to be scaled. Scaling is more than just applying a particular approach in more schools or more regions. One strength of the government-led programs we studied is that from inception, they were designed to be implemented through the system, following an initial investment in curriculum and materials development. These programs drew on existing government resources and limited ongoing maintenance of the intervention to those activities the government could afford, making sometimes difficult decisions about what to prioritize. When making decisions with limited resources, investments should explicitly target activities that improve classroom instruction.
What we still don’t know
While the Learning at Scale study revealed commonalities among the included programs, there were limitations to what questions the study could ultimately answer.
More evidence is needed on teacher adoption of new instructional practices.
Learning at Scale provided evidence on what teachers are doing but there remains limited data on why teachers adopt particular practices (and why they are resistant to others). This requires more research in LMICs on teacher expectations, priorities and definitions of success in the classroom (e.g., King & Gove, 2024); how to develop teachers’ content knowledge for complex subjects, such as early math (e.g., Sitabkhan et al, 2024); and identifying what types of materials and high-impact teaching strategies are most essential for engaging students and improving instruction.
It is essential to understand intervention effects across subjects, as well as their impact on long-term outcomes.
The interaction of reading and math interventions is especially pertinent when there is one teacher in the classroom handling both subjects. Investigations of these programs could help systems to leverage best practices, understand complementarities, and reduce the burden on teachers. Additionally, more research is needed to understand impacts across subjects (e.g., math on reading outcomes; reading on math outcomes), as well as the impact of early grade reading and math interventions in LMICs on later learning outcomes.
In one of the first longitudinal studies to examine the long-term impacts of early grade reading in LMICs, this recently published paper (Stern et al, 2024) explores the lasting effects of the Early Grade Reading Study (EGRS I) in South Africa. It tracks EGRS I students over a seven-year period, showing significant, sustained improvements in oral reading fluency and written comprehension in both Setswana and English, four years after they directly benefited from the intervention.
Moving from what to why: understanding why core components work in some contexts and not others.
Unfortunately, what made these Learning at Scale programs scalable and successful doesn’t necessarily make their success replicable in other contexts. For example, while structured teachers’ guides or coaching models were deemed as integral to the success of many of these programs, many other programs using these same approaches have had far less success. Therefore, simply using these components alone does not guarantee success. A deeper understanding of the core components and key causal pathways of approaches is essential for ensuring successful implementation moving forward.
Integration is great…but institutionalization is better.
Nearly all of the programs selected for Learning at Scale were successful in integrating certain practices into the formal education system; far fewer were able to institutionalize these practices for longer-term sustainability. To be clear, not everything from a successful program could or should be institutionalized. However, additional research is needed on how systems have successfully navigated institutionalization (such as ESMATE in El Salvador and R Maths in South Africa), and what continued support may be required in lower-capacity contexts.
Dig in deeper on equity.
The Learning at Scale study identified 14 foundational learning programs that achieved improvements in learning at scale. More analysis is needed, however, to understand whether those gains in performance were achieved equitably. Building on this, we also need to investigate the extent to which the middle tiers of education systems may interact with schools in ways that reinforce existing patterns of inequity, and what a commitment from governments (and societies) to equity in education would look like.
How can donor-funded programs become government-led programs?
One limitation of the Learning at Scale study was the lack of government-run programs in the literacy phase of the activity. For the numeracy programs, only two government-implemented initiatives were included. This raises questions around what happens after donor funding runs out and if/how successful programs can be handed over to governments. For example, what are the capacity needs (and continued support necessary) for governments to take on the core components of successful interventions,such as those under Literacy at Scale?
Success alone doesn’t tell the whole story.
In evaluating programs for Learning at Scale, we only included those deemed “effective” with evidence of causal impact at scale or causal impact at pilot with evidence of effective scale-up. However, there are certainly limitations of examining only successful programs. A comprehensive analysis approach that includes both successful and unsuccessful programs would provide a more balanced and nuanced perspective and help to inform strategies for literacy and numeracy programs going forward.
Special thanks to Yasmin Sitabkhan, Wendi Ralaingita, Peggy Dubeck, Matthew Jukes, and Justin Sandefur for their valuable inputs to this blog.
