Skip to content

The evidence

Why a classroom in a bag works. Every number sourced.

For NGOs, foundations, governments and development partners: what the research says about structured lessons, why hardware alone has failed, and how each part of an EAP lesson maps to a body of evidence.

The scale of the need

350 million children and youth, aged 6 to 18.

Across sub-Saharan Africa, about 350 million children and youth have no access to interactive lessons. That includes 98 million who are out of school entirely.

≈250M
learners enrolled in school across sub-Saharan Africa with no interactive lessons (EAP model)
98M
children and youth aged 6 to 18 in sub-Saharan Africa who are out of school entirely1
≈350M
children and youth aged roughly 6 to 18 who could be reached

The enrolled figure is EAP’s modelled estimate. The out-of-school figure is UNESCO’s. Together they give about 350 million.

The problem in Kenya

Half of Kenya’s Grade 6 learners cannot read the English every other subject is taught in.

79%
of Kenyan children are in learning poverty: unable to read and understand an age-appropriate text by late primary.2
51%
of public-school Grade 6 learners cannot read and understand a Grade 3 English passage. English is the language every other subject is taught in.3
72,000
junior-school teacher shortfall, within a national shortage of 96,345. Teachers cover learning areas they were never trained in.4
1 : 46
public primary pupil-to-teacher ratio. 54% of observed classrooms hold more than 45 learners, and 11% hold more than 90.5
98%
of observed classrooms had no internet. Government devices reached 99.6% of primary schools, but usage stayed below 40%.6
6.5 hrs
a week the average teacher spends planning lessons. A junior-school teacher handed unfamiliar subjects needs far more.7

The expert panel’s verdict

Hardware alone is a Bad Buy. Structured pedagogy is a Great Buy.

The Global Education Evidence Advisory Panel, convened by the World Bank, FCDO, UNICEF and USAID, ranks what works per dollar.8,9

Great Buy

Structured pedagogy

Ready-made lesson plans, matching teaching materials and ongoing teacher support, delivered as one package.

≈ 3 years

of quality-adjusted schooling gained per $100 spent, for the best programmes. Kenya’s Tusome is the panel’s flagship example.

Bad Buy

Hardware on its own

Laptops, tablets, projectors or smartboards shipped without lessons, materials and support.

≈ 0 years

gained per $100. One Laptop Per Child in Peru and Kenya’s Digital Literacy Programme are the panel’s cautionary cases.

Where the ClassKit sits

All three parts, in one bag

  • Lesson plans: 6,000+ CBE-aligned lessons, every strand and sub-strand.
  • Materials: about 40,000 images and videos on a shared screen, offline.
  • Teacher support: an offline teacher app with five-minute preparation for every lesson, training videos and termly coaching visits.

The projector is the delivery vehicle, not the product.

Proven in Kenya

Pre-planned, standardised lessons have produced some of the largest learning gains ever measured in Africa.

Tusome · national programme

0.6 – 1.0 SD

gains in English and Kiswahili reading across all public primary schools. Grade 2 English gains roughly equalled an extra full year of schooling.10

Every 10 structured lessons a teacher covered added one more correct word per minute to pupils’ reading. Coverage drove learning.

Bridge · randomised trial

+0.81 SD

for primary pupils after two years of centrally written lesson guides, equal to 0.89 extra years of schooling. The effect is in the 99th percentile of at-scale programmes.11

Scholarships were awarded by lottery. The weakest pupils gained most, and girls and boys gained equally.

SD means standard deviation, the unit education trials use. An effect of 0.2 to 0.3 SD is a solid programme result. An effect of 0.8 SD is among the largest ever recorded.

Proven in this format

Same videos, two deliveries. On a shared screen led by the teacher, pupils gained. On their own tablets, pupils lost.

Classroom screen · the ClassKit model

+0.3 SD

in maths and science after four months: about 60% more learning than the control group. Pupils were 5 points more likely to pass the government exam, and teacher and pupil attendance rose.12

One tablet per pupil

−0.42 SD

on the same content loaded onto personal tablets with no classroom screen. Pupils learned less than with no technology at all.12

The trial

  • Government middle schools in Punjab, Pakistan, 2016–17.
  • Classroom arm: 59 schools and 2,999 Grade 8 pupils. Tablet arm: 71 schools and 3,614 Grade 6 pupils.
  • Short videos mapped chapter by chapter to the official curriculum, with review questions after each.
  • Teachers previewed each video before class, then showed it on a large screen: the same prepare-then-project loop as EAP’s teacher app.
  • Two days of training. 95% of teachers used it at least twice a week.

A trial of teacher-led, projected instruction in rural Ghana found a similar result: +0.23 SD in maths.13

Inside every lesson

Each part of an EAP lesson maps to a body of research.

What every lesson hasStrongest studyHeadline finding
The same structure every lesson: objective, vocabulary, content, check, activity, closeTusome, Kenya100.6 to 1.0 SD. One more correct word per minute for every 10 structured lessons covered.
Images and video on a shared screenMayer’s multimedia principle; Pakistan eLearn trial14,12Learners did better from words and pictures than words alone in 13 of 13 experiments. +0.3 SD when projected and teacher-led.
Two vocabulary words per slide, with a definition and a Kenyan exampleStahl and Fairbanks meta-analysis15Up to 0.97 SD on comprehension of taught words. Definition plus context works best.
A check question in the middle of the lessonAdesope et al.: 272 effect sizes, 15,427 participants16Practice testing g = 0.61. Feedback is worth about six months’ progress.
A learning activityFreeman et al.: 225 studies17+0.47 SD. Pupils taught by lecture alone were 1.5 times as likely to fail.
A closing questionRoediger and Karpicke1861% recalled after a week with retrieval practice, against 40% with re-reading.
Preloaded and offline, one kit per classKenya Digital Literacy Programme evaluations6Devices reached 99.6% of schools but were used in under 40%. 98% of classrooms are offline.

Lab and university effect sizes run larger than classroom effects in low-income settings. We quote them as design evidence, not forecasts.

Our commitment

No independent study has evaluated EAP yet. This is how we will produce the first number, within one school year.

  1. 01

    Baseline and endline tests

    A 10-minute oral reading fluency check (words per minute, the Tusome measure) and a numeracy check for every class, at launch and at the end of the school year.

  2. 02

    Usage logs on every kit

    Lessons opened and time per slide, giving the same dose-response data Tusome had.

  3. 03

    Teacher time survey

    Planning hours per week, subjects taught outside training and confidence, at baseline and endline.

  4. 04

    Teacher app usage

    Preparation opened before each lesson and training videos watched, plus one coaching visit per term.

  5. 05

    Independent evaluator

    A matched comparison across 20 to 40 schools over one academic year, run by a Kenyan research partner, with outcomes pre-registered in standard deviations.

  6. 06

    Foundations reading strand

    Grade 3-level English reading with audio inside junior-school lessons: teaching at the right level, the panel’s other Great Buy.

Our target is about 0.3 SD, which would place the ClassKit alongside the most cost-effective programmes the expert panel has reviewed. Funders and research partners can help design the evaluation with us.

Sources

Where every number comes from.

  1. UNESCO Institute for Statistics, “244M children won’t start the new school year”, September 2022. Sub-Saharan Africa: 98 million out of school, ages 6 to 18. Read it
  2. World Bank and UNESCO Institute for Statistics, Learning Poverty brief for Kenya, 2024 (74% of Grade 6 below minimum English reading proficiency in AMPL-b 2023, adjusted for out-of-school children).
  3. Usawa Agenda, Foundational Literacy and Numeracy Assessment (FLANA), 2025.
  4. Teachers Service Commission figures, as reported by the Daily Nation, 2025–26.
  5. Parliament of Kenya, Senate report on pupil-to-teacher ratios; Edu-Design classroom observation study, 2023.
  6. Edu-Design, 2023; University of Notre Dame Pulte Institute, evaluation of Kenya’s Digital Literacy Programme.
  7. OECD, Teaching and Learning International Survey (TALIS), 2018. International benchmark.
  8. Global Education Evidence Advisory Panel (World Bank, FCDO, UNICEF, USAID), “Cost-effective approaches to improve global learning”, 2023. Read it
  9. Angrist, Evans, Filmer, Glennerster, Rogers and Sabarwal, “How to Improve Education Outcomes Most Efficiently?”, World Bank Policy Research Working Paper 9450, 2020. Read it
  10. Piper, Destefano, Kinyanjui and Ong’ele, “Scaling up successfully: Lessons from Kenya’s Tusome national literacy program”, Journal of Educational Change, 2018. NORC endline evaluation for USAID. Read it
  11. Gray-Lobe, Keats, Kremer, Mbiti and Ozier, “Can Education Be Standardized? Evidence from Kenya”, Becker Friedman Institute Working Paper 2022-68. Read it
  12. Beg, Halim, Lucas and Saif, “Engaging Teachers with Technology Increased Achievement, Bypassing Teachers Did Not”, American Economic Journal: Economic Policy 14(2), 2022. Read it
  13. Johnston and Ksoll, evaluation of interactive, teacher-led video instruction in rural Ghana, 2022.
  14. Mayer, Multimedia Learning, Cambridge University Press. The multimedia principle: 13 of 13 experiments.
  15. Stahl and Fairbanks, “The Effects of Vocabulary Instruction: A Model-Based Meta-Analysis”, Review of Educational Research, 1986. Read it
  16. Adesope, Trevisan and Sundararajan, “Rethinking the Use of Tests: A Meta-Analysis of Practice Testing”, Review of Educational Research, 2017. Education Endowment Foundation Toolkit, feedback. Read it
  17. Freeman et al., “Active learning increases student performance in science, engineering, and mathematics”, PNAS, 2014. Read it
  18. Roediger and Karpicke, “Test-Enhanced Learning”, Psychological Science, 2006. Read it

Bring structured lessons to the schools you support.

Get our partner pack, or see a ClassKit in a classroom.

Get the partner pack