Research, Measurement & Evaluation
Evidence on what works in education across 5 countries and 20+ programs.
Evidence on what works in education across 5 countries and 20+ programs.
A literacy programme across northern Nigeria tested whether a standardized tool would fix teachers misjudging their own students' reading skills. Comparing 1,844 pupils assessed by both their teacher and an independent supervisor, agreement was only slight (Cohen's κ = 0.107) — teachers systematically underestimate high performers and overestimate low performers.
Open Interactive Dashboard →Angrist et al. synthesize experimental evidence across five countries on what it takes for structured pedagogy programmes to durably raise literacy and numeracy outcomes at scale.
Three-year follow-up showing sustained and accelerating gains — Primary 3–4 pupils gained the equivalent of a full year's learning in one term.
The data behind Part 1's story: why teachers are shaped far more by daily school culture than by a periodic outside coach.
Read article →
A story about teacher attendance in rural Zambia — and why the "positive deviance" of a single exceptional teacher can't be replicated at scale.
Read article →
Learning outcomes are the Brazil of data — but is there a runner-up? Why pupil attendance may be the Argentina of education measurement.
Read article →
