Start here
Draft wording for the prototype. 3ie to write or approve the final text.
3ie publishes several kinds of evidence. They answer different questions, so it helps to know which one you are reading.
Tests whether one programme caused a change, by comparing people who received it with similar people who did not.
Finds every rigorous study on one question and weighs them together.
Charts which programmes and outcomes have been studied in a field, and where there is nothing.
A short summary of one or more studies, written for people making decisions.
Research shared before it is final, so others can use and comment on it early.
Still not sure where to begin? Browse by area and start with the briefs.
Studies on what works in development. Impact evaluations, systematic reviews, evidence gap maps and briefs, all free to read, download and cite.
Sectors: Health, nutrition and population · Agriculture and rural development · Education · Social protection · Multisector · Finance · Public sector management · Environment and disaster management
Figures as at
Chosen by 3ie’s editors, alongside what readers opened most this quarter.
These are the filters the results page uses. Tick as many as you like here and the results open with them already on.
612 publications match. Nothing ticked yet, so this is the whole collection.Filtering by .
View 612 publications
Pulls together the 2025 gap map and four evidence summaries into one account of what drives people to move, and which interventions have been tested against it.
Training, schooling and health investments weighed against the reasons people leave. The schooling evidence is much thinner than the training evidence.
Ten evaluations of work visas and labour mobility agreements, and what they changed about who migrates and on what terms.
Bundling weather insurance with a loan got more farmers borrowing. It did not change what they chose to plant.
Baseline survey for a sister project, running the same design in a very different market.
Charts 1,200 studies across the migration cycle and shows where the evidence is strong, thin, or absent altogether.
Border and readmission programmes have been evaluated far less often than the money spent on them would suggest.
Thirty-one studies of campaigns meant to change migration decisions. Awareness shifts; behaviour mostly does not.
Reviews how 180 nutrition studies chose what to measure, and why so few of them can be compared with each other.
Teaching children first in their own language raises reading and keeps them in school. The gains are biggest where teachers were retrained too.
Showing 1 to 10 of 612 publications
The rest of 3ie’s evidence, and the work behind these studies.
Filters
Narrow these results
74 publications
in Education, newest first.
BriefRapid response brief2026
Teaching children first in their own language raises reading and keeps them in school. The gains are biggest where teachers were retrained too.
BriefSystematic reviewMar 2026
Training, schooling and health investments weighed against the reasons people leave. The schooling evidence is much thinner than the training evidence.
ReportSystematic reviewFeb 2026
Attendance rose almost everywhere meals were served. Test scores moved only where the school day was also lengthened.
ReportImpact evaluationJan 2026
Paying for measured gains lifted scores in the tested subjects and left the untested ones flat.
BriefImpact evaluation2025
Attendance held while payments continued. Two years after they stopped, the difference had closed in four of the six districts.
Interactive mapEvidence gap map2025
318 studies mapped. Almost all of them stop at the end of the programme, and very few follow a child into the next grade.
BriefImpact evaluation2024
Fee waivers kept more girls enrolled. Distance to the nearest secondary school predicted dropout better than cost did.
PaperWorking paper2024
Radio and SMS reached the most children. Nothing in the evidence base measures what was learned rather than what was delivered.
PaperWorking paper2024
Coaching raised reading in the schools that kept it up for two full years. Where it lapsed after one, the gains went with it.
ReviewSystematic review2023
Attendance rises almost everywhere meals are served. Learning follows only where the teaching was already sound.
Brief2026
Teaching children first in their own language raises reading and keeps them in school. The gains are biggest where teachers were retrained too.
BriefMar 2026
Training, schooling and health investments weighed against the reasons people leave. The schooling evidence is much thinner than the training evidence.
Interactive mapEvidence gap map2026
412 studies on vocational training, mapped. Almost nothing follows young people beyond two years after a course ends.
PaperWorking paperMar 2026
Pulls together the 2025 gap map and four evidence summaries into one account of what drives people to move.
BriefFeb 2026
Ten evaluations of work visas and labour mobility agreements, and what they changed about who migrates.
ReportImpact evaluation2026
Bundling weather insurance with a loan got more farmers borrowing. It did not change what they chose to plant.
BriefImpact evaluationFeb 2026
Training-led support added little on top of a market that was already growing fast.
PaperWorking paperJan 2026
Baseline survey for a sister project, running the same design in a very different market.
ReportImpact evaluationNov 2025
Door-to-door case finding cut the time to treatment by two days. Mortality did not shift over the trial period.
BriefSystematic reviewSep 2025
Facility births rose wherever payments depended on attendance. Newborn outcomes moved only where staffing rose with them.
PaperWorking paperJun 2025
Stockouts fell with a shared district buffer. The saving disappeared once transport was counted.
ReportImpact evaluationDec 2025
Approval rates rose for applicants with a phone history. Default among them tracked the existing portfolio.
BriefSystematic reviewAug 2025
Savings balances grow reliably. Evidence that the groups raise business investment is thin and mostly short-term.
PaperWorking paperMar 2025
A one percent rise in fees moved low-balance users off the service faster than any other group.
ReportImpact evaluationOct 2025
Reported service targets were met more often. Independent spot checks found no change at the facilities.
Interactive mapEvidence gap mapJul 2025
211 studies mapped. Almost all of them look at price, and very few at whether the goods arrived.
PaperWorking paperFeb 2025
Publication on its own changed little. Where a civil society group read and challenged the figures, allocations moved.
ReportImpact evaluationNov 2025
Warnings reached most households in time. Whether anyone moved depended on having somewhere to go.
BriefSystematic reviewSep 2025
Yields held through one bad season in most trials. Adoption fell away where seed had to be bought again each year.
Interactive mapEvidence gap mapApr 2025
154 studies mapped against nine adaptation sectors. Coastal protection has almost none.
PaperWorking paperOct 2025
Combined packages beat single interventions in eleven of fourteen trials. None of them isolates which part did the work.
BriefSystematic reviewMay 2025
Consumption gains persist three years out. The cost per household is the number that varies most.
ReportImpact evaluationSep 2025
Private wages rose in the villages that got the scheme. The effect faded once the works season ended.
Showing 1 to 8 of 74
Filters
Narrow these results
0 publications
in Education, newest first.
There are 34 publications on mother tongue, but none of them is in Education, tagged Gender equality and available in French at the same time. Try one of these.
Spelling: we also looked for mother-tongue and mothertongue.
Rapid Response Brief 14
Also published as French: L’apprentissage en langue maternelle · Spanish: La lengua materna y el aprendizaje
March 2026
Governments across sub-Saharan Africa and South Asia are revisiting the language children are taught in during their first years of school. 3ie was asked by two education ministries to summarise what the evidence says before those decisions are taken.
This brief synthesises 38 impact evaluations and 4 systematic reviews covering mother-tongue and bilingual instruction in 19 low- and middle-income countries. Teaching children first in a language they already speak raises early reading scores and keeps them in school, with the largest effects where teachers were retrained in the language of instruction and where readers and workbooks were available in that language. Where materials were translated but teachers were not retrained, the gains largely disappeared. The evidence is thinner on what happens at the point of transition to a national or international language, which is where most programmes place that transition between grades three and five. Three of the 38 studies follow children beyond the transition year, and none follows them to the end of primary school, so the long-run effect on literacy remains unmeasured. The brief closes with four questions a ministry should be able to answer before committing to a language policy, and names the two evaluations currently in the field that will report against them.
Read or download
Every file and language version held against this record
BriefMar 2026
Training, schooling and health investments weighed against the reasons people leave. The schooling evidence is much thinner than the training evidence.
Same programme, same topic
Interactive mapEvidence gap map2026
412 studies on vocational training, mapped. Almost nothing follows young people beyond two years after a course ends.
Different programme, same topic
BriefFeb 2026
Ten evaluations of work visas and labour mobility agreements, and what they changed about who migrates.
Same programme, different topic
Checked against the BRD
58 lines govern these pages. Click any outlined element.
Not this page
Your words on the left, the screen that answers them on the right.
