education · family: the context you didn’t design for
everything they learned, unlearned
BRAC's play-based pre-primary gains vanish within two years of entering government schools
Problem statement
BRAC's pre-primary program enrolls over 1 million children annually in play-based, child-centered learning environments. Rigorous evaluations show significant gains in school readiness, early literacy, and numeracy at pre-primary exit. But when these children enter government primary schools — which use rote memorization, teacher-centered instruction, and corporal punishment — the gains fade within 18–24 months. By Grade 3, BRAC pre-primary graduates are statistically indistinguishable from children who had no pre-primary education. The problem is not that play-based learning doesn't work; it's that the receiving system actively undoes what was built.
Why this matters
Bangladesh's government has adopted a national pre-primary policy, and BRAC's model heavily influenced its design. But if pre-primary gains cannot survive the transition to a pedagogically incompatible primary system, the entire investment — which now reaches millions of children across multiple countries — represents a temporary effect rather than a developmental foundation. This is not unique to Bangladesh: similar fadeout effects have been documented in Head Start (US), Perry Preschool, and multiple LMIC pre-primary interventions. The pattern suggests a structural incompatibility between early childhood pedagogy and primary school pedagogy that no amount of pre-primary quality improvement can resolve.
What’s been tried and why it hasn’t worked
BRAC has tried extending its model into primary school (BRAC Primary Schools use similar pedagogy through Grade 5), and BRAC school graduates do sustain gains. But BRAC cannot replace the government primary system — 85% of Bangladesh's 18 million primary students attend government schools. Efforts to train government teachers in child-centered methods have produced minimal change because teacher behavior is shaped by the assessment system, not by training content: national exams reward memorized recall, so teachers teach for memorized recall regardless of their training. The assessment system, not the teacher, is the binding constraint. BRAC has also experimented with "bridge" classes during Grade 1 transition, but these delay fadeout without preventing it.
What would unlock progress
The problem is fundamentally one of system alignment: pre-primary pedagogy optimizes for developmental competencies (curiosity, problem-solving, self-regulation), while primary pedagogy optimizes for content recall as measured by standardized exams. Changing either system independently doesn't work — the pre-primary system can't prepare children for rote learning without abandoning its developmental model, and the primary system can't adopt child-centered methods while being evaluated on memorization. Progress requires either reforming the assessment system (which drives teacher behavior) or designing a transition pedagogy that translates developmental competencies into forms that survive contact with a rote-learning environment.
Entry points for student teams
An education design team could ask which competencies survive the transition and which fade — not by fielding a new cohort study, which would take restricted BRAC data or multi-year fieldwork in Bangladesh, but by decomposing the published evaluations (Aboud & Hossain 2011 and Nath et al. 2013, cited above) and the wider fade-out literature (Head Start, Perry Preschool) by competency domain, building a persistence map from effect tables already in print. A policy team could analyze assessment reform efforts in comparable systems (Rwanda, Kenya) to identify what levers actually changed teacher behavior at scale. A measurement team could design assessments that capture developmental competencies (not just content knowledge) — and since piloting them in a school is an intervention with minors that outruns a semester, the deliverable is a pilot-ready package handed to a partner such as BRAC IED: the instrument, an administration protocol, and a pre-registered evaluation plan for whether its use changes instructional practice.
Genome — every gene is a door
Structural cousins — same reason stuck, other fields
Sources
BRAC Institute of Educational Development (BIED) research; Aboud & Hossain, "The impact of preprimary school on primary school achievement in Bangladesh," Early Childhood Research Quarterly, 2011; Nath et al., "Achievement of primary competencies: A comparison between government and BRAC schools," BRAC Research and Evaluation Division, 2013 (accessed 2026-02-25)
verification notes (working record)
The collection team’s own sourcing notes for this brief, kept verbatim:
BRAC Institute of Educational Development (BIED) has published extensively on this fadeout pattern. The framing here follows BRAC's own analysis, which identifies the assessment system — not teacher quality or training — as the binding constraint. This is a classic wrong-stakeholder pattern: pre-primary interventions target children's readiness, but the binding constraint is the receiving institution's assessment system. BRAC's own research points this direction but the organization's programmatic response has been to build parallel schools rather than to confront the assessment system directly.
Structural parallel: `education-curriculum-assessment-misalignment` documents the same claim with OECD 40-country data — curriculum reform targets content designers while assessment bodies' instruments determine classroom reality. Together, these briefs ground the "assessment as binding constraint" pattern across GS and OECD contexts.
Source type: Self-articulated Institutional source: BRAC (Bangladesh) Galaxy A tag: failure:wrong-stakeholder, constraint:behavioral, breakthrough:behavior-change
Reconciliation 2026-08-21: Entry-point repair per the C37 realism triage. Two suggestions failed the student-scale bar: the competency-persistence study of BRAC graduates in government schools (needs restricted BRAC partner data or multi-year fieldwork abroad) and the in-school assessment pilot (intervention with minors, plus school access). The triage's replacement hypothesis — public BRAC RCT replication microdata — did not verify: searches of the World Bank Microdata Library (https://microdata.worldbank.org/index.php/catalog) surfaced no public BRAC pre-primary evaluation microdata, and a UNICEF MICS check could not be completed (mics.unicef.org blocks automated access), so no dataset is claimed as public. The persistence question was recast as a domain-level decomposition of the already-cited published evaluations (Aboud & Hossain 2011; Nath et al. 2013) plus the fade-out literature — facility-free. The assessment pilot was recast per the design-the-trial default: instrument + protocol + pre-registered evaluation plan handed to BRAC IED. The Rwanda/Kenya desk-analysis door was kept untouched.