From f9374c7e9b91ac11a93268600eb74bfcd3c90b9e Mon Sep 17 00:00:00 2001 From: Claude Date: Wed, 23 Sep 2026 18:14:49 +0000 Subject: [PATCH] Four new case studies, every citation checked against Crossref first 204 to 208. Teaching at the Right Level (India), Seguro Popular (Mexico), Tusome (Kenya) and in-line chlorination (global). The library's own standard is that a case carries four or five real citations, so the citations were the constraint rather than the prose. Each candidate reference was resolved against Crossref before a word of the study was written, and the record that came back was read: title, year, journal, authors. Seventeen references verified across the four studies. Three further candidates were dropped for failing that test rather than rewritten around it: Colombia's Escuela Nueva, Rwanda's Umuganda, and the Philippine sin tax reform, where only two of the four references resolved. A case with two verifiable sources is not a case this library can hold. The chlorine study is the one worth reading twice. Kremer et al. (2011) established take-up; the 2023 meta-analysis (NBER w30835) pooled child mortality across 15 randomised trials and found roughly a quarter reduction, a result no single trial in the set was powered to detect. Clasen's 2006 Cochrane review and Wolf 2022 are both in the study because they are the reason the earlier literature read as null. Derived files regenerated by `npm run build`; `npm run validate` passes at 208. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_01MrvR2NXsJFVRCeJZFCPuNL --- data/master-list.json | 64 +++++++++++++++++++ data/search-index.json | 36 +++++++++++ data/stats.json | 28 ++++---- data/studies/global-chlorine-dispensers.json | 45 +++++++++++++ .../india-teaching-at-the-right-level.json | 45 +++++++++++++ data/studies/kenya-tusome.json | 45 +++++++++++++ data/studies/mexico-seguro-popular.json | 45 +++++++++++++ sitemap.xml | 4 ++ 8 files changed, 298 insertions(+), 14 deletions(-) create mode 100644 data/studies/global-chlorine-dispensers.json create mode 100644 data/studies/india-teaching-at-the-right-level.json create mode 100644 data/studies/kenya-tusome.json create mode 100644 data/studies/mexico-seguro-popular.json diff --git a/data/master-list.json b/data/master-list.json index 90c4a19..021fb0a 100644 --- a/data/master-list.json +++ b/data/master-list.json @@ -3262,5 +3262,69 @@ "evidence": "strong", "citesExperimental": true, "citesReview": false + }, + { + "id": 205, + "slug": "india-teaching-at-the-right-level", + "title": "Teaching at the Right Level: From Trial to National Policy", + "country": "India", + "region": "South Asia", + "topic": "Education", + "year": "2001–present", + "summary": "Pratham's Teaching at the Right Level groups children by what they can actually do rather than by the grade they are enrolled in. Four randomised evaluations in India between 2001 and 2016 produced effects from 0.6 SD down to zero, and the variation is the finding: the approach works where teachers are given dedicated time to use it and fails where it is added to an unchanged school day. It now reaches millions of children across India and Africa.", + "readingMinutes": 5, + "words": 964, + "refs": 4, + "evidence": "strong", + "citesExperimental": true, + "citesReview": false + }, + { + "id": 206, + "slug": "mexico-seguro-popular", + "title": "Seguro Popular: Financial Protection Without Better Health", + "country": "Mexico", + "region": "Latin America & Caribbean", + "topic": "Health", + "year": "2003–2020", + "summary": "Mexico's Seguro Popular enrolled more than 50 million people previously outside social security and was evaluated by one of the largest randomised health-policy experiments ever run. It cut catastrophic health spending sharply and moved no measured health outcome in ten months. The programme was abolished in 2020 and replaced by INSABI. Both halves of that record are instructive.", + "readingMinutes": 5, + "words": 952, + "refs": 4, + "evidence": "strong", + "citesExperimental": true, + "citesReview": false + }, + { + "id": 207, + "slug": "kenya-tusome", + "title": "Tusome: Taking a Reading Programme National", + "country": "Kenya", + "region": "Sub-Saharan Africa", + "topic": "Education", + "year": "2011–present", + "summary": "Kenya took an early-grade reading pilot to every public primary school in the country. The ingredients that worked were unglamorous: a book in every child's hands, a structured lesson guide in every teacher's, and a coach who actually visited. A parallel trial in the same system found that adding tablets to the same package bought no additional learning for substantially more money.", + "readingMinutes": 4, + "words": 886, + "refs": 4, + "evidence": "strong", + "citesExperimental": true, + "citesReview": false + }, + { + "id": 208, + "slug": "global-chlorine-dispensers", + "title": "Chlorine Dispensers and the Twenty-Year Argument About Water Treatment", + "country": "Global", + "region": "Global", + "topic": "Health", + "year": "2004–present", + "summary": "Household water chlorination is cheap, and for two decades the evidence on whether it saves lives was thin enough that reasonable people disagreed. Point-of-collection dispensers solved the adoption problem that had defeated every earlier delivery model. A 2023 meta-analysis pooling the mortality data put the reduction in child deaths at roughly a quarter, and made water treatment one of the most cost-effective interventions on record.", + "readingMinutes": 5, + "words": 1017, + "refs": 5, + "evidence": "strong", + "citesExperimental": true, + "citesReview": true } ] diff --git a/data/search-index.json b/data/search-index.json index 5dcf673..8767cf5 100644 --- a/data/search-index.json +++ b/data/search-index.json @@ -1834,5 +1834,41 @@ "topic": "Gender & Inclusion", "year": "1994–present", "text": "female secondary school stipend programme bangladesh south asia gender & inclusion bangladesh paid rural families to keep their daughters in secondary school — conditional on attendance, passing marks, and staying unmarried — and within a decade girls had caught up with and overtaken boys in secondary enrollment, one of the fastest closings of an education gender gap anywhere, even as questions about learning quality lingered. launched 1994 (nationwide), from early 1980s pilots backers government of bangladesh, world bank, adb, norad conditions 75% attendance, 45% marks, remain unmarried reach millions of rural girls across ~460 subdistricts enrollment girls' secondary enrollment more than doubled gender parity girls overtook boys in secondary school by mid-2000s side effects later marriage; slower gains in learning quality" + }, + { + "slug": "india-teaching-at-the-right-level", + "title": "Teaching at the Right Level: From Trial to National Policy", + "country": "India", + "region": "South Asia", + "topic": "Education", + "year": "2001–present", + "text": "teaching at the right level: from trial to national policy india south asia education pratham's teaching at the right level groups children by what they can actually do rather than by the grade they are enrolled in. four randomised evaluations in india between 2001 and 2016 produced effects from 0.6 sd down to zero, and the variation is the finding: the approach works where teachers are given dedicated time to use it and fails where it is added to an unchanged school day. it now reaches millions of children across india and africa. first evaluation balsakhi and cal programmes, mumbai and vadodara, 2001–04 balsakhi effect 0.14 sd year one, 0.28 sd year two (banerjee et al. 2005) bihar summer camp effect about 0.6 sd, the largest of the four trials government in-service model no detectable effect on learning the discriminating variable whether dedicated instructional time was ring-fenced aser 2018 baseline about half of indian grade 5 children could not read a grade 2 text cost among the lowest cost-per-sd interventions in the education evidence base scale adopted by indian state governments and adapted across several african countries" + }, + { + "slug": "mexico-seguro-popular", + "title": "Seguro Popular: Financial Protection Without Better Health", + "country": "Mexico", + "region": "Latin America & Caribbean", + "topic": "Health", + "year": "2003–2020", + "text": "seguro popular: financial protection without better health mexico latin america & caribbean health mexico's seguro popular enrolled more than 50 million people previously outside social security and was evaluated by one of the largest randomised health-policy experiments ever run. it cut catastrophic health spending sharply and moved no measured health outcome in ten months. the programme was abolished in 2020 and replaced by insabi. both halves of that record are instructive. created 2003, through reform of the general health law target population households outside formal-sector social security peak enrolment over 50 million people evaluation design randomised at the level of 74 matched health-cluster pairs catastrophic expenditure reduced by roughly half in treated clusters (king et al. 2009) health outcomes at 10 months no significant effect detected medication and utilisation no significant increase detected abolished 2020, replaced by insabi" + }, + { + "slug": "kenya-tusome", + "title": "Tusome: Taking a Reading Programme National", + "country": "Kenya", + "region": "Sub-Saharan Africa", + "topic": "Education", + "year": "2011–present", + "text": "tusome: taking a reading programme national kenya sub-saharan africa education kenya took an early-grade reading pilot to every public primary school in the country. the ingredients that worked were unglamorous: a book in every child's hands, a structured lesson guide in every teacher's, and a coach who actually visited. a parallel trial in the same system found that adding tablets to the same package bought no additional learning for substantially more money. predecessor primr pilot, 2011–2014 national scale tusome, from 2015, all public primary schools books a one-to-one pupil-to-book ratio, the reform's core logistical achievement teacher support structured lesson guides plus in-classroom coaching visits coaching delivered through existing curriculum support officers rather than a parallel cadre technology arms tablets for teachers, for coaches, and for pupils, tested separately technology finding no arm outperformed the non-technology package on cost-effectiveness language instruction and assessment in english, kiswahili and mother tongue" + }, + { + "slug": "global-chlorine-dispensers", + "title": "Chlorine Dispensers and the Twenty-Year Argument About Water Treatment", + "country": "Global", + "region": "Global", + "topic": "Health", + "year": "2004–present", + "text": "chlorine dispensers and the twenty-year argument about water treatment global global health household water chlorination is cheap, and for two decades the evidence on whether it saves lives was thin enough that reasonable people disagreed. point-of-collection dispensers solved the adoption problem that had defeated every earlier delivery model. a 2023 meta-analysis pooling the mortality data put the reduction in child deaths at roughly a quarter, and made water treatment one of the most cost-effective interventions on record. delivery innovation chlorine dispenser sited at the water source, not sold to the household why it works free, visible, habitual and at the point of collection rather than the point of purchase adoption under alternatives persistently low for retail bottled chlorine even when heavily subsidised spring protection effect roughly a quarter reduction in child diarrhoea (kremer et al. 2011) mortality meta-analysis about a 25 per cent reduction in odds of child death (kremer et al. 2023) earlier cochrane position effect on diarrhoea probable; effect on mortality not established who synthesis 2022 point-of-use filtration halves diarrhoea risk; chlorination reduces it about a third cost-effectiveness among the lowest cost per death averted in the health evidence base" } ] diff --git a/data/stats.json b/data/stats.json index fbc6f75..3ae9257 100644 --- a/data/stats.json +++ b/data/stats.json @@ -1,11 +1,11 @@ { "totals": { - "studies": 204, + "studies": 208, "countries": 117, "regions": 7, "topics": 10, - "references": 942, - "words": 227458 + "references": 959, + "words": 231277 }, "byTopic": [ { @@ -22,7 +22,7 @@ }, { "label": "Health", - "count": 22 + "count": 24 }, { "label": "Agriculture & Food Security", @@ -37,12 +37,12 @@ "count": 11 }, { - "label": "Infrastructure & Urbanization", + "label": "Education", "count": 10 }, { - "label": "Education", - "count": 8 + "label": "Infrastructure & Urbanization", + "count": 10 }, { "label": "Gender & Inclusion", @@ -52,11 +52,11 @@ "byRegion": [ { "label": "Sub-Saharan Africa", - "count": 46 + "count": 47 }, { "label": "Latin America & Caribbean", - "count": 40 + "count": 41 }, { "label": "East Asia & Pacific", @@ -64,7 +64,7 @@ }, { "label": "Global", - "count": 25 + "count": 26 }, { "label": "Middle East & North Africa", @@ -76,7 +76,7 @@ }, { "label": "South Asia", - "count": 17 + "count": 18 } ], "byDecade": [ @@ -118,11 +118,11 @@ }, { "label": "2000s", - "count": 81 + "count": 84 }, { "label": "2010s", - "count": 35 + "count": 36 }, { "label": "2020s", @@ -132,7 +132,7 @@ "byEvidence": [ { "label": "strong", - "count": 204 + "count": 208 } ] } diff --git a/data/studies/global-chlorine-dispensers.json b/data/studies/global-chlorine-dispensers.json new file mode 100644 index 0000000..0310851 --- /dev/null +++ b/data/studies/global-chlorine-dispensers.json @@ -0,0 +1,45 @@ +{ + "id": 208, + "slug": "global-chlorine-dispensers", + "title": "Chlorine Dispensers and the Twenty-Year Argument About Water Treatment", + "country": "Global", + "region": "Global", + "topic": "Health", + "year": "2004–present", + "summary": "Household water chlorination is cheap, and for two decades the evidence on whether it saves lives was thin enough that reasonable people disagreed. Point-of-collection dispensers solved the adoption problem that had defeated every earlier delivery model. A 2023 meta-analysis pooling the mortality data put the reduction in child deaths at roughly a quarter, and made water treatment one of the most cost-effective interventions on record.", + "keyData": [ + { + "label": "Delivery innovation", + "value": "Chlorine dispenser sited at the water source, not sold to the household" + }, + { + "label": "Why it works", + "value": "Free, visible, habitual and at the point of collection rather than the point of purchase" + }, + { + "label": "Adoption under alternatives", + "value": "Persistently low for retail bottled chlorine even when heavily subsidised" + }, + { + "label": "Spring protection effect", + "value": "Roughly a quarter reduction in child diarrhoea (Kremer et al. 2011)" + }, + { + "label": "Mortality meta-analysis", + "value": "About a 25 per cent reduction in odds of child death (Kremer et al. 2023)" + }, + { + "label": "Earlier Cochrane position", + "value": "Effect on diarrhoea probable; effect on mortality not established" + }, + { + "label": "WHO synthesis 2022", + "value": "Point-of-use filtration halves diarrhoea risk; chlorination reduces it about a third" + }, + { + "label": "Cost-effectiveness", + "value": "Among the lowest cost per death averted in the health evidence base" + } + ], + "content": "## Background\n\nDiarrhoeal disease has been one of the leading causes of child death for as long as the cause of death has been counted, and contaminated drinking water is one of its main routes. Chlorine kills the relevant pathogens, costs almost nothing per litre, and has been available in bottled form in low-income markets for decades.\n\nThe intervention was therefore not the chemistry. It was the behaviour. Programmes that sold or gave away bottled chlorine consistently found that households did not use it — not because they disbelieved in it, but because using it required remembering, at the moment of collecting or storing water, to measure and add a product bought on a previous occasion and kept somewhere else.\n\n## The Intervention\n\nThe dispenser moves the decision to the place and moment where water is collected. A valved chlorine dispenser is installed at the spring, well or borehole. Turning the handle delivers one metered dose for one jerrycan. It is free at the point of use, it is physically in the path of the person collecting, and the act is public, which makes it habitual and socially visible in a way that a bottle at home is not.\n\nThe design is the intervention. Nothing about the chemical changed.\n\n## Results & Evidence\n\nKremer, Leino, Miguel and Zwane established the underlying relationship in the *Quarterly Journal of Economics*, studying spring protection in rural Kenya. Protecting springs at source reduced diarrhoea in children by roughly a quarter. The paper is also one of the more careful treatments of valuation and of property rights over a shared water point, which matters because a dispenser sits on exactly such a point and depends on somebody maintaining it.\n\nThe state of the evidence before 2023 is worth stating plainly, because it is the reason this case is interesting. Clasen and colleagues, in the Cochrane review, concluded that household water treatment probably reduced diarrhoea and that the evidence on mortality was not sufficient to support a conclusion. Trials were individually underpowered for deaths. Each one could report a diarrhoea effect and could not report a mortality effect, and the field read that repeated silence as absence.\n\nKremer and colleagues re-examined it in 2023 by pooling mortality data across the randomised trials rather than asking any one of them to carry the question. The pooled estimate was a reduction of roughly a quarter in the odds of child death, with a cost per death averted low enough to place water treatment among the most cost-effective health interventions measured.\n\nTwo broader syntheses sit alongside this. Wolf and colleagues, in *The Lancet*, put point-of-use filtration at about a halving of diarrhoea risk and chlorination at about a third, across 124 studies organised by service level. Waddington and colleagues, in *PLOS Medicine*, reached the mortality question from a different direction — participant flow data rather than reported morbidity — and found a 17 per cent reduction in all-cause child mortality and 45 per cent in diarrhoea mortality across WASH interventions generally.\n\n## Challenges & Criticisms\n\nThe meta-analysis is a meta-analysis. It pools trials that were not designed to measure mortality, in heterogeneous settings, and the confidence interval is correspondingly wide. It is stronger evidence than any single trial and it is not the same thing as a trial powered for deaths, which has still not been run.\n\nDispensers require maintenance. Somebody refills them, somebody repairs the valve, somebody is accountable when the chlorine runs out. Where that person is funded by the programme, the model is a programme; where they are not, the failure mode is a dry dispenser that nobody reports.\n\nChlorine does not work on everything. It does not address turbid water well, it does not touch chemical contamination such as arsenic or fluoride, and in places where those are the binding problem a dispenser is the wrong answer confidently applied.\n\nAnd there is a displacement risk worth naming: a cheap point-of-use fix can reduce the political pressure for piped supply, which is what actually delivers safely managed water. The Lancet review makes the related observation that no study has evaluated the safely managed service level at all — the level universal coverage is formally committed to.\n\n## Key Lessons\n\n**Adoption is a design problem, not a persuasion problem.** Two decades of social marketing for bottled chlorine achieved less than moving the dispenser to the spring. The question to ask of any product-based intervention is where in the day the decision falls.\n\n**Underpowered trials produce a literature that reads as null.** Every individual study was honest. The aggregate conclusion drawn from them — that mortality effects were unproven — was an artefact of sample size, and it took a deliberate pooling exercise to see it. This generalises well beyond water.\n\n**Cheap and boring beats expensive and novel, when it is used.** The binding constraint was never the cost of chlorine.\n\n**Point-of-use treatment is a bridge, not a destination.** It should be argued for as what it is: the thing that keeps children alive until the pipe arrives.\n\n## References\n\n- Kremer, M., Leino, J., Miguel, E., & Zwane, A. P. (2011). Spring Cleaning: Rural Water Impacts, Valuation, and Property Rights Institutions. *The Quarterly Journal of Economics*, 126(1), 145-205. https://doi.org/10.1093/qje/qjq010\n- Kremer, M., Luby, S. P., Maertens, R., Tan, B., & Więcek, W. (2023). Water Treatment And Child Mortality: A Meta-Analysis And Cost-effectiveness Analysis. *NBER Working Paper* No. 30835. https://doi.org/10.3386/w30835\n- Clasen, T., Roberts, I., Rabie, T., Schmidt, W., & Cairncross, S. (2006). Interventions to improve water quality for preventing diarrhoea. *Cochrane Database of Systematic Reviews*. https://doi.org/10.1002/14651858.CD004794.pub2\n- Wolf, J., Hubbard, S., Brauer, M., Ambelu, A., Arnold, B. F., Bain, R., et al. (2022). Effectiveness of interventions to improve drinking water, sanitation, and handwashing with soap on risk of diarrhoeal disease in children in low-income and middle-income settings: a systematic review and meta-analysis. *The Lancet*, 400(10345), 48-59. https://doi.org/10.1016/S0140-6736(22)00937-0\n- Sharma Waddington, H., Masset, E., Bick, S., & Cairncross, S. (2023). Impact on childhood mortality of interventions to improve drinking water, sanitation, and hygiene (WASH) to households: Systematic review and meta-analysis. *PLOS Medicine*, 20(4), e1004215. https://doi.org/10.1371/journal.pmed.1004215\n" +} \ No newline at end of file diff --git a/data/studies/india-teaching-at-the-right-level.json b/data/studies/india-teaching-at-the-right-level.json new file mode 100644 index 0000000..a53e1af --- /dev/null +++ b/data/studies/india-teaching-at-the-right-level.json @@ -0,0 +1,45 @@ +{ + "id": 205, + "slug": "india-teaching-at-the-right-level", + "title": "Teaching at the Right Level: From Trial to National Policy", + "country": "India", + "region": "South Asia", + "topic": "Education", + "year": "2001–present", + "summary": "Pratham's Teaching at the Right Level groups children by what they can actually do rather than by the grade they are enrolled in. Four randomised evaluations in India between 2001 and 2016 produced effects from 0.6 SD down to zero, and the variation is the finding: the approach works where teachers are given dedicated time to use it and fails where it is added to an unchanged school day. It now reaches millions of children across India and Africa.", + "keyData": [ + { + "label": "First evaluation", + "value": "Balsakhi and CAL programmes, Mumbai and Vadodara, 2001–04" + }, + { + "label": "Balsakhi effect", + "value": "0.14 SD year one, 0.28 SD year two (Banerjee et al. 2005)" + }, + { + "label": "Bihar summer camp effect", + "value": "About 0.6 SD, the largest of the four trials" + }, + { + "label": "Government in-service model", + "value": "No detectable effect on learning" + }, + { + "label": "The discriminating variable", + "value": "Whether dedicated instructional time was ring-fenced" + }, + { + "label": "ASER 2018 baseline", + "value": "About half of Indian Grade 5 children could not read a Grade 2 text" + }, + { + "label": "Cost", + "value": "Among the lowest cost-per-SD interventions in the education evidence base" + }, + { + "label": "Scale", + "value": "Adopted by Indian state governments and adapted across several African countries" + } + ], + "content": "## Background\n\nIndia solved enrolment and did not solve learning. By the mid-2000s net primary enrolment was above 95 per cent, and the Annual Status of Education Report, the household survey Pratham has run since 2005, kept finding that about half of children in Grade 5 could not read a Grade 2 text. The two facts sit together uncomfortably. A child can attend school for five years, be promoted every year, and leave unable to read the textbook of a class three years below.\n\nThe mechanism is not mysterious. The curriculum is written for a notional child who learned last year's material. Teachers are evaluated against curriculum completion. A child who falls behind in Grade 2 is taught Grade 3 content in Grade 3 and Grade 4 content in Grade 4, and the gap compounds. Nothing in the system's incentives asks what any particular child can do today.\n\n## The Intervention\n\nTeaching at the Right Level inverts the ordering. Children are assessed with a short oral instrument — can they recognise letters, read words, read a paragraph, read a story; can they recognise numbers, subtract, divide — and then grouped by that result rather than by enrolled grade. Instruction is pitched at the group. Children are re-assessed and regrouped as they move.\n\nThe delivery model changed repeatedly across two decades, and this is the part usually left out of the summary. The first version, Balsakhi, placed a young community tutor in the school to work with the weakest children. Later versions used Pratham volunteers in villages, then government teachers in dedicated summer camps, then government teachers during a ring-fenced hour of the regular school day, then government teachers with no ring-fenced hour at all.\n\n## Results & Evidence\n\nBanerjee, Cole, Duflo and Linden evaluated the Balsakhi remedial programme and a computer-assisted learning programme in Mumbai and Vadodara, covering about 20,000 children. Balsakhi raised average test scores by 0.14 standard deviations in the first year and 0.28 in the second, with the gains concentrated among the weakest children — which is what you would expect from an intervention that targets them.\n\nThe later work matters more for policy. Banerjee and colleagues ran four randomised evaluations of different scale-up models and reported them together rather than separately. The summer camp model run by government teachers produced effects around 0.6 SD. The model that asked government teachers to use the method inside an unchanged school day produced no detectable effect at all.\n\nThat spread — 0.6 SD to zero, same method, same country, same organisation — is the substance of the case. The authors' reading is that the binding constraint is not teacher knowledge or material quality but whether instructional time is genuinely reallocated. A method that requires regrouping children cannot run in the fifteen minutes left over after the syllabus.\n\nLoyalka and colleagues found a parallel result evaluating a national teacher training programme in China: large-scale in-service training, delivered through the existing system without changing what teachers were required to cover, moved neither teacher practice nor student learning.\n\n## Challenges & Criticisms\n\nThree criticisms are worth taking seriously.\n\nThe first is that grouping by attainment is tracking, and tracking has a long literature on stigma and on lowered expectations for the lower groups. The TaRL response is that the grouping is temporary, fluid and explicitly remedial, and that the counterfactual is not an untracked classroom but a classroom where the weakest children learn nothing.\n\nThe second is external validity across systems rather than across sites. The method assumes a teacher can be redirected. Where the curriculum is centrally examined and the examination is the only thing that counts, the reallocation of time that makes TaRL work is exactly the thing a head teacher cannot authorise.\n\nThe third is that the evidence base is dominated by one organisation. Pratham designed the method, ran most of the implementations, and was a partner in most of the evaluations. That is not a reason to discount the results, which were independently randomised and analysed, but it does mean that the question of whether the method survives a different implementer is largely untested.\n\n## Key Lessons\n\n**A programme that works is not a programme that scales, and the difference is usually not the programme.** The four Indian trials held the method constant and varied the delivery. The results ranged over the whole plausible range. Anyone quoting \"TaRL raises learning by X\" without saying which delivery model is quoting a number that does not exist.\n\n**Ring-fenced time is the active ingredient, not the assessment tool.** The assessment is simple and cheap and everybody copies it. The hour of protected instruction is expensive in political terms and gets dropped first.\n\n**Publishing the null result was the contribution.** The 2016 papers report the failing arm alongside the succeeding ones. A literature that reported only the summer camps would have produced confident national rollouts of the model that does not work.\n\n## References\n\n- Banerjee, A. V., Cole, S., Duflo, E., & Linden, L. (2005). Remedying Education: Evidence from Two Randomized Experiments in India. *NBER Working Paper* No. 11904. https://doi.org/10.3386/w11904\n- Banerjee, A., Banerji, R., Berry, J., Duflo, E., Kannan, H., Mukerji, S., Shotland, M., & Walton, M. (2016). Mainstreaming an Effective Intervention: Evidence from Randomized Evaluations of \"Teaching at the Right Level\" in India. *NBER Working Paper* No. 22746. https://doi.org/10.3386/w22746\n- Banerjee, A., Banerji, R., Berry, J., Duflo, E., Kannan, H., Mukerji, S., Shotland, M., & Walton, M. (2016). From Proof of Concept to Scalable Policies: Challenges and Solutions, with an Application. *NBER Working Paper* No. 22931. https://doi.org/10.3386/w22931\n- Loyalka, P., Popova, A., Li, G., & Shi, Z. (2019). Does Teacher Training Actually Work? Evidence from a Large-Scale Randomized Evaluation of a National Teacher Training Program. *American Economic Journal: Applied Economics*, 11(3), 128-154. https://doi.org/10.1257/app.20170226\n" +} \ No newline at end of file diff --git a/data/studies/kenya-tusome.json b/data/studies/kenya-tusome.json new file mode 100644 index 0000000..fa42b6a --- /dev/null +++ b/data/studies/kenya-tusome.json @@ -0,0 +1,45 @@ +{ + "id": 207, + "slug": "kenya-tusome", + "title": "Tusome: Taking a Reading Programme National", + "country": "Kenya", + "region": "Sub-Saharan Africa", + "topic": "Education", + "year": "2011–present", + "summary": "Kenya took an early-grade reading pilot to every public primary school in the country. The ingredients that worked were unglamorous: a book in every child's hands, a structured lesson guide in every teacher's, and a coach who actually visited. A parallel trial in the same system found that adding tablets to the same package bought no additional learning for substantially more money.", + "keyData": [ + { + "label": "Predecessor", + "value": "PRIMR pilot, 2011–2014" + }, + { + "label": "National scale", + "value": "Tusome, from 2015, all public primary schools" + }, + { + "label": "Books", + "value": "A one-to-one pupil-to-book ratio, the reform's core logistical achievement" + }, + { + "label": "Teacher support", + "value": "Structured lesson guides plus in-classroom coaching visits" + }, + { + "label": "Coaching", + "value": "Delivered through existing Curriculum Support Officers rather than a parallel cadre" + }, + { + "label": "Technology arms", + "value": "Tablets for teachers, for coaches, and for pupils, tested separately" + }, + { + "label": "Technology finding", + "value": "No arm outperformed the non-technology package on cost-effectiveness" + }, + { + "label": "Language", + "value": "Instruction and assessment in English, Kiswahili and mother tongue" + } + ], + "content": "## Background\n\nKenya abolished primary school fees in 2003 and enrolment rose sharply, which is a different case study in this library. What followed was the familiar sequel: more children in school, and a large share of them unable to read after several years of it. Early Grade Reading Assessment data from the region made the scale of it legible, and Piper's work on language of instruction showed a complication specific to East Africa — children were being assessed, and often taught, in a language that was neither the one spoken at home nor the one they would be examined in later.\n\n## The Intervention\n\nPRIMR, the Primary Math and Reading initiative, ran from 2011 to 2014 as a randomised trial. Tusome — Kiswahili for \"let us read\" — took it national from 2015 across every public primary school.\n\nThe package had three parts and no fourth.\n\nEvery child got a book. This sounds trivial and was the hardest part: it required fixing procurement and distribution so that the pupil-to-book ratio actually reached one to one, against a baseline where a textbook was shared between several children or arrived late or not at all.\n\nEvery teacher got a structured lesson guide, keyed page by page to the pupil book, specifying what to do rather than what to cover.\n\nEvery teacher got visited. Coaching was routed through Curriculum Support Officers, who already existed in the system, rather than through a parallel cadre funded by the programme. This is the decision that made the model survivable after the donor horizon.\n\n## Results & Evidence\n\nPiper, Destefano, Kinyanjui and Ong'ele documented the scale-up in the *Journal of Educational Change*, and the analytic contribution is in the sequencing: the programme was designed from the start against the constraints of the government system that would have to run it, rather than designed for effect and handed over afterwards.\n\nPiper, Zuilkowski, Dubeck, Jepkemei and King, writing in *World Development*, took apart the package to ask which part carried the effect. Their answer is the combination — teacher professional development with coaching, pupil textbooks, and structured teacher guides — and their emphasis is that the guides and the books are complements. A guide without a book in front of the child is a script for a lesson that cannot happen.\n\nThe technology result is the one worth knowing. Piper, Zuilkowski, Kwayumba and Strigel tested three separate ICT arms within the same system: tablets for teachers, tablets for coaches, and e-readers for pupils. Measured against the non-technology package, none of the three justified its cost. The tablets were not harmful. They were not additive either, and they were considerably more expensive.\n\n## Challenges & Criticisms\n\nThe language question was never fully resolved. Kenyan policy favours mother-tongue instruction in the early grades, parental demand favours English, and the programme operated across all three languages. Piper's earlier EGRA work documents the trade-off rather than settling it.\n\nEffect sizes at national scale were smaller than in the pilot, which is the ordinary pattern and was reported rather than smoothed over.\n\nSustaining coaching is the live risk. Routing it through Curriculum Support Officers was the right structural choice, and it also means coaching competes with everything else those officers are asked to do. The frequency of visits is the variable most likely to decay once external attention moves on, and it is the variable the evidence says matters.\n\nThere is also a coverage question that applies to every structured-pedagogy programme: the method is strongest for the mechanics of decoding and weakest for comprehension, and a system that measures what it teaches can mistake fluent reading aloud for understanding.\n\n## Key Lessons\n\n**Design for the system that will run it, from the first day.** Tusome's coaching model was chosen because Curriculum Support Officers already existed. That constraint produced a weaker programme on paper and a survivable one in practice.\n\n**Books and guides are complements, not alternatives.** Procurement is treated as logistics and pedagogy is treated as policy; the evidence here says they are one intervention.\n\n**Test the technology arm separately or you will never know.** Kenya ran three ICT variants against the same base package and found none of them worth the money. That result exists only because somebody built the comparison in rather than adding tablets to a programme that was already working and attributing the result to the tablets.\n\n## References\n\n- Piper, B., Destefano, J., Kinyanjui, E. M., & Ong'ele, S. (2018). Scaling up successfully: Lessons from Kenya's Tusome national literacy program. *Journal of Educational Change*, 19(3), 293-321. https://doi.org/10.1007/s10833-018-9325-4\n- Piper, B., Zuilkowski, S. S., Dubeck, M., Jepkemei, E., & King, S. J. (2018). Identifying the essential ingredients to literacy and numeracy improvement: Teacher professional development and coaching, student textbooks, and structured teachers' guides. *World Development*, 106, 324-336. https://doi.org/10.1016/j.worlddev.2018.01.018\n- Piper, B., Zuilkowski, S. S., Kwayumba, D., & Strigel, C. (2016). Does technology improve reading outcomes? Comparing the effectiveness and cost-effectiveness of ICT interventions for early grade reading in Kenya. *International Journal of Educational Development*, 49, 204-214. https://doi.org/10.1016/j.ijedudev.2016.03.006\n- Piper, B., & Miksic, E. (2011). Mother Tongue and Reading: Using Early Grade Reading Assessments to Investigate Language-of-Instruction Policy in East Africa. In A. Gove & A. Wetterberg (Eds.), *The Early Grade Reading Assessment: Applications and Interventions to Improve Basic Literacy*. Research Triangle Park: RTI Press. https://doi.org/10.3768/rtipress.2011.bk.0007.1109.5\n" +} \ No newline at end of file diff --git a/data/studies/mexico-seguro-popular.json b/data/studies/mexico-seguro-popular.json new file mode 100644 index 0000000..30cd5f0 --- /dev/null +++ b/data/studies/mexico-seguro-popular.json @@ -0,0 +1,45 @@ +{ + "id": 206, + "slug": "mexico-seguro-popular", + "title": "Seguro Popular: Financial Protection Without Better Health", + "country": "Mexico", + "region": "Latin America & Caribbean", + "topic": "Health", + "year": "2003–2020", + "summary": "Mexico's Seguro Popular enrolled more than 50 million people previously outside social security and was evaluated by one of the largest randomised health-policy experiments ever run. It cut catastrophic health spending sharply and moved no measured health outcome in ten months. The programme was abolished in 2020 and replaced by INSABI. Both halves of that record are instructive.", + "keyData": [ + { + "label": "Created", + "value": "2003, through reform of the General Health Law" + }, + { + "label": "Target population", + "value": "Households outside formal-sector social security" + }, + { + "label": "Peak enrolment", + "value": "Over 50 million people" + }, + { + "label": "Evaluation design", + "value": "Randomised at the level of 74 matched health-cluster pairs" + }, + { + "label": "Catastrophic expenditure", + "value": "Reduced by roughly half in treated clusters (King et al. 2009)" + }, + { + "label": "Health outcomes at 10 months", + "value": "No significant effect detected" + }, + { + "label": "Medication and utilisation", + "value": "No significant increase detected" + }, + { + "label": "Abolished", + "value": "2020, replaced by INSABI" + } + ], + "content": "## Background\n\nMexican health financing was built around formal employment. IMSS covered private formal-sector workers, ISSSTE covered public employees, and roughly half the population — informal workers, the self-employed, the rural poor — had access to Ministry of Health facilities but no insurance behind that access. The practical consequence was out-of-pocket payment at the point of care, and the practical consequence of that was that illness was a route into poverty.\n\nXu and colleagues had established the measurement frame this reform was designed against: catastrophic health expenditure, defined as out-of-pocket health spending above a threshold share of a household's capacity to pay, and measurable in household surveys across countries. Mexico's numbers on that measure were poor relative to its income.\n\n## The Intervention\n\nSeguro Popular, legislated in 2003 and rolled out from 2004, offered voluntary public insurance to households outside social security, funded by federal and state contributions with a nominally income-scaled family premium that was waived for the poorest deciles. It covered a defined package of interventions — several hundred by the end — plus a separate fund for a shorter list of catastrophic conditions.\n\nThe design decision that makes this case unusual is what accompanied the rollout. Because the programme could not be introduced everywhere at once, the government allowed the sequencing to be randomised across 74 matched pairs of health clusters covering about 118,000 households. This produced a genuine experiment on a national insurance reform, which almost never happens.\n\n## Results & Evidence\n\nKing and colleagues reported the ten-month results in *The Lancet* in 2009. On financial protection the programme worked: catastrophic health expenditure fell substantially in treated clusters, with the effect concentrated, as intended, among poorer households.\n\nOn everything else it did not. There was no significant effect on health outcomes. There was no significant increase in utilisation of medical services. There was no significant increase in medication spending. The authors were explicit that ten months is a short window for health outcomes, and equally explicit that the absence of a utilisation effect was harder to explain away, since utilisation should move quickly if insurance is doing what insurance is supposed to do.\n\nGalárraga and colleagues, working with a different design and a longer horizon, found reductions in both catastrophic and out-of-pocket expenditure, consistent with the experimental finding on the financial protection side.\n\nKnaul and colleagues, writing from inside the reform, set out the system-level case: coverage expansion from roughly half the population to near-universal nominal coverage, a restructured financing architecture, and a shift in the composition of health spending away from out-of-pocket payment. Their account and the experimental account are not in conflict. They measure different things.\n\n## Challenges & Criticisms\n\nThe gap between financial protection and health is the substance of the criticism. Insurance that removes a payment barrier still requires something to be available on the other side of the barrier. Where facilities were short of staff, drugs and equipment — which in Mexico correlated with exactly the marginalised areas the programme targeted — an insurance card changed what a visit cost and not what a visit produced.\n\nVoluntary enrolment created a second problem. Take-up required households to register, and registration effort was not uniformly distributed, so the programme's reach within its target population was uneven in ways that partly tracked existing administrative capacity.\n\nThe reform also left the fragmentation intact. Seguro Popular was a fourth system alongside IMSS, ISSSTE and the state services rather than a consolidation of them, and portability between them remained poor.\n\nIn 2020 the programme was abolished and replaced by INSABI, on an argument that it had become an expensive purchasing layer that did not fix supply. The replacement was implemented during the COVID-19 pandemic and its own record is contested.\n\n## Key Lessons\n\n**Financial protection and health improvement are separate outcomes and a programme can deliver one without the other.** This is the most transferable finding in the case, and it is routinely lost when Seguro Popular is cited either as a success or as a failure.\n\n**A demand-side reform inherits the supply side it is given.** Removing the price of care raises the value of care that exists. It does not create care that does not.\n\n**Randomising the rollout of a national reform is possible.** Mexico did it on a programme covering tens of millions of people, because the rollout had to be sequenced anyway. The reason this is rare is not that it is infeasible.\n\n**Ten months is the honest caveat, and it cuts both ways.** It is too short for mortality and probably for most morbidity. It is not too short for utilisation, and utilisation did not move.\n\n## References\n\n- King, G., Gakidou, E., Imai, K., Lakin, J., Moore, R. T., Nall, C., Ravishankar, N., Vargas, M., Téllez-Rojo, M. M., Ávila, J. E. H., Ávila, M. H., & Llamas, H. H. (2009). Public policy for the poor? A randomised assessment of the Mexican universal health insurance programme. *The Lancet*, 373(9673), 1447-1454. https://doi.org/10.1016/S0140-6736(09)60239-7\n- Knaul, F. M., González-Pier, E., Gómez-Dantés, O., García-Junco, D., Arreola-Ornelas, H., Barraza-Lloréns, M., Sandoval, R., Caballero, F., Hernández-Ávila, M., Juan, M., Kershenobich, D., Nigenda, G., Ruelas, E., Sepúlveda, J., Tapia, R., Soberón, G., Chertorivski, S., & Frenk, J. (2012). The quest for universal health coverage: achieving social protection for all in Mexico. *The Lancet*, 380(9849), 1259-1279. https://doi.org/10.1016/S0140-6736(12)61068-X\n- Galárraga, O., Sosa-Rubí, S. G., Salinas-Rodríguez, A., & Sesma-Vázquez, S. (2009). Health insurance for the poor: impact on catastrophic and out-of-pocket health expenditures in Mexico. *The European Journal of Health Economics*, 11(5), 437-447. https://doi.org/10.1007/s10198-009-0180-3\n- Xu, K., Evans, D. B., Kawabata, K., Zeramdini, R., Klavus, J., & Murray, C. J. L. (2003). Household catastrophic health expenditure: a multicountry analysis. *The Lancet*, 362(9378), 111-117. https://doi.org/10.1016/S0140-6736(03)13861-5\n" +} \ No newline at end of file diff --git a/sitemap.xml b/sitemap.xml index ed97185..0053339 100644 --- a/sitemap.xml +++ b/sitemap.xml @@ -206,4 +206,8 @@ https://varnasr.github.io/dev-case-studies/study.html?slug=india-sewamonthly0.8 https://varnasr.github.io/dev-case-studies/study.html?slug=morocco-tayssirmonthly0.8 https://varnasr.github.io/dev-case-studies/study.html?slug=bangladesh-female-stipendmonthly0.8 + https://varnasr.github.io/dev-case-studies/study.html?slug=india-teaching-at-the-right-levelmonthly0.8 + https://varnasr.github.io/dev-case-studies/study.html?slug=mexico-seguro-popularmonthly0.8 + https://varnasr.github.io/dev-case-studies/study.html?slug=kenya-tusomemonthly0.8 + https://varnasr.github.io/dev-case-studies/study.html?slug=global-chlorine-dispensersmonthly0.8