Rwanda’s pre-primary education infrastructure varies across and within districts

In what researchers describe as one of the most detailed portraits ever assembled of early childhood education in a low-income country, a new study of nearly 5,000 pre-primary settings across Rwanda has revealed stark, measurable gaps in the physical infrastructure and staffing quality of the classrooms where the country’s youngest children learn. The findings, published in the International Journal of Child Care and Education Policy, show that home-based nurseries—the backbone of early learning in rural Rwanda—lag centre-based schools by as much as 0.85 standard deviations on key structural measures, while the wealthiest sectors of the country concentrate nearly all of the formal, well-equipped provision.

The research, conducted by Remy Pages and Gilbert Munyemana, draws on an ambitious 2023 mapping exercise commissioned by Rwanda’s National Child Development Agency and carried out by Esri Rwanda. Combining direct observation, geographic information system (GIS) classification, and self-reports from centre managers, the dataset covers 4,875 pre-primary settings serving 207,521 children aged three to five, spread across 91 administrative sectors in seven districts spanning all of Rwanda’s provinces. The authors merged this operational census with the country’s 2022 national census, which counted 13,246,394 residents, of whom 1,071,139—just over 8 percent—were children aged three to five.

Rwanda’s pre-primary landscape is unusual in that three very different kinds of provision coexist under a single national policy framework. Centre-based settings are formal institutions, typically urban, offering structured curricula delivered by trained caregivers. Community-based settings are improvised arrangements in which communal buildings or temporary structures are converted into learning spaces by local residents. Home-based settings—the most informal modality—are private households designated by neighboring families to host children’s early learning during the day, usually charging minimal or no fees. Under Rwanda’s regulatory system, sector-level officers receive monthly reports from caregivers and parents’ committees and are expected to conduct supervisory visits, though the intensity of oversight varies sharply by setting type, with centre-based providers facing the most consistent monitoring from Sector Education Officers and Social and Economic Development Officers.

The scale of Rwanda’s expansion in early childhood education makes the quality question urgent. The country has made enrollment a national priority consistent with Sustainable Development Goal 4.2, which affirms universal, high-quality pre-primary education as the right of every child. But as the new analysis makes clear, expanding access and ensuring quality are not the same thing—and the mechanism of expansion shapes what children actually experience.

What the Numbers Show

The descriptive picture is unambiguous. Among centre-based settings, 93 percent have handwashing stations, 80 percent have piped water, and 92 percent have safe waste disposal. Home-based settings report 68 percent, 23 percent, and 68 percent, respectively—the piped water gap alone corresponds to an effect size of d = 1.39, an enormous difference by social science standards. Proximity to public infrastructure follows the same pattern: 87 percent of centre-based settings sit within a two-kilometer walk of a health facility and 96 percent near a primary school, compared with 69 percent and 82 percent for home-based settings.

Human resources show even wider gaps. In centre-based settings, 45 percent of educators hold formal qualifications; in home-based settings, the figure is 10 percent (d = 1.07). Centre-based settings operate an average of 5.06 hours per day against 3.84 for home-based provision, and roughly five days per week compared with fewer than four. Community-based settings fall consistently between the two extremes, with staffing constraints resembling those of home-based care but somewhat better physical infrastructure.

Enrollment patterns mirror the geography of the divide. Centre-based settings account for 65 percent of total enrollment—135,634 children across 1,370 centres—while home-based settings serve 48,366 children (23 percent) across 2,824 homes, and community-based settings serve 23,521 children (11 percent) in 681 locations. In the capital’s districts, centre-based provision dominates: in Gasabo, home to Kigali’s urbanized core, up to 78 percent of enrolled children attend centres, and in sectors like Kinyinya, Nduba, and Gisozi, centre-based settings enroll 50 to 66 percent of the eligible child population. Some central sectors, such as Kimihurura, post enrollment rates exceeding 100 percent of their resident eligible population—a signal that urban centres draw children from surrounding areas. In the rural district of Nyamasheke, by contrast, home-based settings are the primary access point, with sectors such as Bushenge reporting home-based enrollment rates of 90 percent of eligible children, and Kagano, Kanjongo, and Karengera ranging from 33 to 45 percent.

A Multilevel Lens

What distinguishes the study methodologically is its attempt to separate two distinct sources of inequality. Because settings are clustered within Rwanda’s 91 administrative sectors, the authors used multilevel structural equation modeling to isolate within-sector differences between setting types from between-sector variation tied to sector-level socioeconomic conditions. Sector-level socioeconomic status was operationalized as the average years of completed schooling among residents aged twelve and older—a proxy strongly negatively correlated with the Multidimensional Poverty Index (Pearson’s r = –0.89). The authors handled the over-representation of Kigali’s 35 sectors with a binary indicator and reported cluster-robust standard errors throughout.

Infrastructure quality itself was modeled as three correlated latent factors derived from confirmatory factor analysis of binary and continuous indicators—a approach the authors argue is more robust than simple sum-score indexes because it accounts for shared variance among indicators and improves the precision of the effects of interest. The three factors captured basic physical facilities (handwashing stations, waste disposal, piped water), public infrastructure proximity (distance to health facilities, government offices, and primary schools), and staff quality and operational metrics (educator qualifications, daily hours, and weekly operating days). The measurement model fit the data well, with a comparative fit index of 0.955, an RMSEA of 0.06, and an SRMR of 0.05.

The results of the multilevel models are the study’s core contribution. Within sectors—holding local socioeconomic conditions constant—home-based settings scored 0.73 standard deviations below centre-based settings on physical facilities (95 percent confidence interval: –0.77 to –0.70) and 0.85 standard deviations lower on operational quality. Community-based settings showed similar structural constraints. Intraclass correlation coefficients reinforced how much of the action lies at the sector level: 51 percent of the variance in waste disposal access, 46 percent for proximity to government offices, and 42 percent for qualified educators resided between sectors rather than within them.

Yet a surprising nuance emerged when sector socioeconomic status entered the models. Between-sector disparities in infrastructure quality, which appeared substantial in the raw data, were no longer meaningful once the analysis adjusted for sector-level SES. In other words, the geographical clustering of poor infrastructure is largely explained by the socioeconomic composition of the sectors themselves rather than by geography per se. The poorest sectors—where average adult educational attainment is 4.99 years, compared with 6.21 years in sectors hosting centre-based provision—are precisely the sectors where home-based settings predominate and where structural quality is weakest.

The ‘Double Inequality’

The authors frame this pattern as an instance of what the international literature calls a “double inequality.” Socioeconomic status shapes not only which settings families can access but also the quality of those settings: higher-SES families cluster in better-resourced urban centres, while lower-SES families, facing constrained choices, prioritize affordability and proximity, landing in settings with minimal infrastructure, unqualified caregivers, and schedules dictated by household routines. Prior research also suggests that higher-SES families are better positioned to distinguish among care options and select environments conducive to development, amplifying the gap in children’s actual experiences.

The stakes are high. Decades of evidence from economics and developmental psychology show that high-quality early education improves school readiness and later outcomes, with the strongest benefits accruing to children in lower-resourced contexts—the very children most likely to be enrolled in Rwanda’s informal settings. The authors note that in low- and middle-income countries, structural inputs remain central to both access and quality, whereas in high-income countries, where universal baselines were achieved decades ago, structural features now show weaker associations with outcomes. Rwanda thus represents a system at an earlier stage of the same trajectory: three modalities under one policy roof, differing dramatically in resources and oversight.

The study also documents the fragile governance of the informal sector. Home-based caregivers are expected to report monthly to cell-level authorities, and parents’ committees oversee daily operations, sometimes in coordination with community health workers. But this supervision rests heavily on voluntary parent leadership and in-kind community contributions, raising questions about sustainability. Community-based settings, though formally embedded in the National Child Development Agency’s supervision cascade, are similarly constrained by their reliance on improvised facilities and unpredictable community mobilization.

The policy implications the authors draw are targeted rather than sweeping. Given that home-based and community-based settings are often the only viable early learning option in remote areas—and may deliver greater marginal value there than analogous settings in wealthier countries—the authors recommend prioritizing a minimum structural package for home-based settings: basic water, sanitation, and hygiene infrastructure, paired with educator training, sustainable financing mechanisms, and supportive oversight. Without such an intervention, the informal settings that made Rwanda’s enrollment expansion possible risk entrenching the very inequities the expansion was meant to erase.

For a country that has staked much of its development strategy on human capital formation, the message of the data is clear: the children furthest behind live not just in the poorest districts, but in the informal classrooms those districts can afford.