Evidence
We publish only results that trace back to an evaluation, a government assessment, or a provider report, and we label how strong each one is.
The trial that set the bar
0.8 to 1.3 SD
In a multi-year randomized trial in Kenya, children gained 0.8 to 1.3 standard deviations in learning, above the 99th percentile of education interventions. Most programs move learning by about 0.1.
Read the paperSource: Gray-Lobe, Keats, Kremer, Mbiti and Ozier (2022), Can Education Be Standardized? Evidence from Kenya
How strong is each result?
- Randomized controlled trialStrongest
- Quasi-experimentalStrong
- Pre/postModerate
- Provider-reportedIndicative
Results across the portfolio
Results use the unit each study reported, so compare them within a study rather than across studies.
| Provider | Where | Result | Study type |
|---|---|---|---|
| Rising Academies | Sierra Leone | +0.38 SD maths, +0.29 SD English | Randomized controlled trial (2022 to 2025) |
| Taleemabad | Balochistan, Pakistan | +0.23 SD Urdu | Randomized controlled trial |
| Educational Initiatives (Mindspark) | Delhi, India | +0.37 SD maths | Randomized controlled trial (Muralidharan et al. 2019) |
| PLANE | Nigeria | +0.41 SD literacy and numeracy | Quasi-experimental |
| NewGlobe (BayelsaPRIME) | Bayelsa, Nigeria | +0.22 to +0.57 SD reading fluency | Difference-in-differences (2023 to 2025) |
| EIDU | Kenya | +0.534 SD early literacy and numeracy (IDELA) | Comparison with non-participating peers |
| HPPI (Kadam+) | Uttar Pradesh, India | 84% at Grade 3 proficiency vs. 53% in similar schools | National Achievement Survey comparison |
| TaRL Africa | Nigeria | +26.6 percentage points more learners at target maths level | Provider-reported |
How we check every figure
- 1.
Sourced
Every figure comes from a published evaluation, government assessment, or provider report.
- 2.
Reverified
Our team checks each figure against its original source.
- 3.
Validated
A provider is validated once its core cost and learning data are confirmed.