Lee Crawfurd

Clark and Nielsen (2026) assemble a large dataset of quasi-experimental studies and argue that the causal return to an extra year of schooling is close to zero once one corrects for the selective publication of statistically significant findings. If correct, this would overturn a central pillar of human capital theory and weaken the empirical case for investing in education. This comment re-examines their own data with the standard battery of publication-bias corrections, including methods that guard against spuriously precise estimates and against p-hacking. Every method points to a return of around 5–6% per additional year of schooling—positive, statistically significant and far from zero—and the estimate stays well above zero even under the most conservative variants. Clark and Nielsen’s near-zero figure comes instead from a bespoke procedure that assumes large numbers of negative-return studies were suppressed; the distribution of published estimates does not support that assumption, and the same procedure understates even a genuine, positive return in simulation. Publication bias in this literature is real but modest: corrected for properly, the evidence is inconsistent with a near-zero causal return.

Addressing foundational literacy and numeracy in low-income countries is a major global education policy priority. Yet we have little direct evidence on the effect of these skills on adult outcomes. We use panel survey data from Indonesia, where skills are tested in early childhood, and the same individuals are then re-interviewed as adults. After adjusting for rich covariates, one standard deviation higher foundational skills are associated with a 9% increase in adult earnings. This association is partially but not primarily attenuated by controlling for completed schooling. If these correlations can be interpreted as causal, they would imply a large benefit–cost ratio for investments in foundational skills.

One in three children worldwide have elevated lead exposure. Lead-acid batteries are the main use of lead by weight, and many are recycled unsafely, but it is uncertain how much of human exposure can be traced to this recycling. In this paper we provide new modelled estimates suggesting that around 36% of lead exposure in low- and lower-middle-income countries may come from battery recycling, although there remains significant uncertainty. The vast majority of harm comes from mass population low-level exposure, rather than localised hotspots. Previous studies have typically focused on small populations with high exposure living within hundreds of metres of polluted sites, but recent reduced-form quasi-experimental evidence demonstrates smaller negative effects for people living within much wider areas, affecting many more people. Our simulation model reconciles these approaches, and shows that expanding the area of concern around each recycling site increases the estimated share of global lead exposure attributable to battery recycling by an order of magnitude, from just 0.55% when considering only the high exposures within a few hundred metres of a site.

Lead poisoning affects one in three children globally, with large but neglected impacts on health and education. One important source of exposure is the use of lead as a pigment in the form of lead chromates, in paints, plastics, and even food. Lead chromates are toxic and safe alternatives exist: Its continued use is effectively prohibited in the European Union and the United Kingdom and is highly restricted in other high‐income countries. In this paper, we present new data documenting the global trade in lead chromates. Exports are dominated by companies based in India, but we also show significant ongoing exports from high‐income countries including the United States, Canada, and Spain, despite these countries placing bans or strong restrictions on use in their own territories. Many exports go to countries with low capacity to regulate its end use and prevent highly hazardous applications. While the harm posed by any particular shipment or actor depends on the end use of the exported lead chromates, which is not possible to determine, the volume of exports from rich to poor countries could be sufficient to poison hundreds of thousands of children. We conclude with recommendations to policymakers to end the detrimental and unnecessary use of lead chromates.

Despite rapid increases in global access to primary school, average learning outcomes in many low and middle income countries remain low. International actors are increasingly focused on a policy agenda prioritizing foundational learning, measured by test scores in primary school. However, international actors have limited capacity to impose this agenda, which ultimately depends on national decision-makers. In this paper we present new evidence on the priorities and views of these national decision-makers. We report on a new survey of 931 senior government officials working on education in 35 low- and middle-income countries. We show with survey experiments that national policymakers place relatively low value on action to address foundational skills. We explain variation in preferences among policymakers as a function of three possible factors: different objectives for education (e.g., learning versus socialization), different beliefs about the state of the world (e.g., enrollment and learning levels), and different beliefs about the effectiveness of specific interventions. Misalignment with donor agendas is evident in all three dimensions. We also show experimentally that beliefs do respond to new evidence on the effectiveness of interventions.

Can effective school leadership enhance high-stakes test scores in low-income countries? To address this question, we examine the short-term impact of a school leadership professional development program that was implemented by VVOB across 525 primary schools in the six lowest-performing districts of Rwanda between 2018 and 2019. The program aimed to strengthen the leadership, management, and teacher support skills of school headteachers. We find that the program had small but statistically insignificant effects on Primary Leaving Examination scores within one to two years after the intervention. However, the program led to a five to six percentage point increase in teacher retention rates, with qualitative evidence suggesting that headteachers provided greater support to teachers, in particular during the onboarding process. Future research should focus on refining such programs and understanding which mechanisms are necessary to also improve learning outcomes of students.

Around half of children in low-income countries have elevated blood-lead levels. What role does lead play in explaining low educational outcomes in these settings? We conduct a new systematic review and meta-analysis of observational studies on the relationship between lead exposure and learning outcomes. Adjusting for observable confounds and publication bias yields a benchmark estimate of a -0.12 standard-deviation reduction in learning per natural log unit of blood lead. As all estimates are non-experimental, we present evidence on the likely magnitude of unobserved confounding, and summarize results from a smaller set of natural experiments. Our benchmark estimate accounts for over a fifth of the gap in learning outcomes between rich and poor countries, and implies moderate learning gains from targeted interventions for highly exposed groups (≈ 0.1 standard deviations) and modest learning gains (< 0.05 standard deviations) from broader public health campaigns.

Textbooks play a critical role in schooling around the world. Small sample studies show that many books continue to under-represent women and girls, and to portray men and women in stereotypical gendered roles. In this paper, we use quantitative text analysis to assess the degree of gender bias in a newly assembled corpus of 1,255 English language school textbooks from 34 countries that are publicly available online. We find consistent patterns of under-representation of female characters and portrayal of stereotypical gendered roles. Women and girls appear less frequently, are portrayed as more passive, are less likely to be associated with work or achievement, and are more likely to be associated with the home and traditionally female occupations. Comparing across countries, female representation in books is correlated with higher GDP and more legal rights for women. Under-representation and stereotypes are a particular problem in South Asia.

The first SDG4 target is to both expand access to universal secondary school and to ensure that all children achieve minimum learning outcomes in primary school. To the extent that action must be prioritized, this article argues that achieving universal secondary education is both more feasible at scale and has clearer benefits than improving learning outcomes in primary schools in developing countries. Removing barriers to access such as fees and exam requirements has been demonstrated to have consistent impacts at scale, even in contexts of weak state capacity. By contrast, efforts to improve school quality have been much less consistent. Wage gains from secondary schooling have been demonstrated repeatedly, even in contexts of low quality education. Wage gains from improving primary school quality have not. Governments should thus focus on reforms that reliably expand access, rather than complex interventions to improve learning that often fail at scale.

Improving school quality in low and middle income countries (LMICs) is a global priority. One way to improve quality may be to improve the management skills of school leaders. In this systematic review, we analyze the impact of interventions targeting school leaders’ management practices on student learning. We begin by describing the characteristics and responsibilities of school leaders using data from large, multi-country surveys. Second, we review the literature and conduct a meta-analysis of the causal effect of school management interventions on student learning, using 39 estimates from 20 evaluations. We estimate a statistically significant improvement in student learning of 0.033 standard deviations. We show that effect sizes are not related to program scale or intensity. We complement the meta-analysis by identifying common limitations to program effectiveness through a qualitative assessment of the studies included in our review. We find three main factors which mitigate program effectiveness: (1) low take-up; (2) lack of incentives or structure for implementation of recommendations; and (3) the lengthy causal chain linking management practices to student learning. Finally, to assess external validity of our review, we survey practitioners to compare characteristics between evaluated and commonly implemented programs. Our findings suggest that future work should focus on generating evidence on the marginal effect of common design elements in these interventions, including factors that promote school leader engagement and accountability.

Education systems regularly face unexpected school closures, whether due to disease outbreaks, natural disasters, or other adverse shocks. In low-income countries where internet access is scarce, distance learning – the most common educational solution – is often passive, via TV or radio, with little opportunity for teacher–student interaction. In this paper we evaluate the effectiveness of live tutoring calls from teachers, designed to supplement radio instruction during the 2020 school closures prompted by the COVID-19 pandemic. We do this with a randomised controlled trial with 4,399 primary school students in Sierra Leone. Tutoring calls led to some limited increase in educational activity, but had no effect on mathematics or language test scores, whether for girls or boys, and whether provided by public or private school teachers. Even having received tutoring calls, one in three children reported not listening to educational radio at all, so limited take-up may partly explain our results.

The private school sector has expanded with almost no public intervention to educate half of primary school children in many urban centers in Africa and Asia. Simple comparisons of test scores would suggest that these private schools may provide better quality than public schools, but how much of this difference is due to selection effects is unclear. Much donor and policymaker attention has proceeded on the basis that private schools do perform better, and focused on models of public subsidy to expand access, and investment in networks or chains to encourage expansion. We review the evidence of the effects of private schools on learning, and how that effect translates to public-private partnerships (PPPs). We also study the effects of private school chains. We conduct a systematic review for eligible studies, with transparent search criteria. The search resulted in over 100 studies on low-cost private schools and PPPs, with a large majority being on low-cost private schools. Our meta-analysis shows moderately strong effects from private schooling, although the limited number of experimental studies find much smaller effects than quasi-experimental studies. This advantage, though, is not nearly enough to help most children reach important learning goals. Turning to policy goals, we find that the private school advantage has not translated to public private partnerships, which have shown limited value in improving quality. They can however represent a lower-cost means of increasing access to school. We also find that private school chains perform little better than individual private schools and have little scope for achieving meaningful scale.

Can governments contract out school management at scale? In 2016 the Government of Punjab transferred management of over 4,000 failing primary schools to private operators. Schools remained free to students. Private operators received a government subsidy per enrolled student of less than half per-student spending in government schools. This paper evaluates the effects on performance of converted schools. Comparing early converters to later converters, we estimate that enrolment in treated schools increased by over 60 percent, and test scores declined sharply.

Are Labour party politicians anti-Semitic, and are Conservative party politicians Islamophobic? In this correspondence study we measure the responsiveness of elected local representatives in the United Kingdom to requests from putative constituents from minority religious groups. We send short email requests to 10,268 local government representatives from each of the main political parties, from stereotypically Islamic, Jewish, and Christian names. Response rates are six to seven percentage points lower to stereotypically Muslim or Jewish names. The two major political parties both show equal bias towards the two minority group names. Results suggest that the bias in response may be implicit. Bias is lower in more dense and diverse locations.

Highlights • Learning profiles allow us to trace learning progress through schooling.• In this paper I estimate a learning profile for Rwanda using a household-based survey of children.• I use a detailed schooling history to show that observational learning profiles that don’t explicitly account for repetition and dropout may substantially over-state learning gains per year.

SUMMARY Public support for global development in rich countries is critical for sustaining effective government and individual action. But the causes of public support are not well understood. Does spending time living in a developing country play a role in generating individual commitment to development? Addressing this question is fraught with selection bias, as individuals are rarely exogenously assigned to spend time in different countries. In this paper I address this question using a natural experiment – the quasi‐random assignment of missionaries from the Church of Jesus Christ of Latter‐day Saints to two‐year missions in different world regions. I provide the first causal estimates of the effect of travel to a developing country on attitudes to global development. Data comes from a new survey gathered through mission alumni Facebook groups. Missionaries assigned to low‐income and middle‐income world regions (Africa, Asia, Latin America, and Caribbean) have similar prior characteristics to those assigned to high‐income Europe. Those assigned to Africa self‐report greater interest in global development and greater charitable attitudes and behaviours. However, they also express stronger opposition to immigration from poor countries, and are less likely to be involved in political campaigns to address global development. Spending time in lower income countries may lead to greater support for charity but less support for political change.

How persistent are teacher effects on student outcomes? In this paper we present estimates of teacher effects on long‐run student outcomes from two low‐ and middle‐ income countries. We first estimate teacher value‐added using the Young Lives School Survey data from Ethiopia and Vietnam. We then track students taught by these teachers 2 and 5 years later and use data from the Young Lives Household Surveys to estimate the effects of teacher quality. We find no persistent effect after 2 years, but better mathematics (0.08 σ ) and reading (0.06 σ ) test scores after 5 years, from being taught by a 1 σ better Grade 5 teacher. We find no persistent effects of good teachers on measures of more “generalized” cognitive ability, aspirations, well‐being, or “grit.”

Value‐added models ( VAMS ) are commonly used in high‐income countries for measuring the quality of teachers and schools, on the grounds that they are a fairer reflection of true quality than simple average test scores, as they account for differences in student intake. Not accounting for student's prior test scores can give a misleading impression of school quality. In this article, we adapt the current VAM of secondary school quality to the Ugandan context, and test its robustness. Using official test score data from Uganda, we test the robustness of the model to a range of different empirical specifications, including sensitivity to the inclusion of controls for student socioeconomic status. We find that the model is robust to a variety of different specifications and control variables. The VAM is low cost and has the potential to provide a clearer signal to parents, teachers, schools, and policy‐makers about how much learning is actually happening in different schools. This approach could be carried out at low cost in a wide range of low‐income countries that have similar testing regimes.

Can the quality of school management explain differences in student test scores? In this paper, I present the first internationally benchmarked estimates of school management quality in Africa (based on the ‘World Management Survey’). The level and distribution of management quality is similar to that found in other low and middle-income countries (India and Brazil). I combine this data with individual student panel data, and demonstrate that differences in school management quality matter for student value-added—a standard deviation difference in management is associated with a 0.06 standard deviation difference in test scores. Finally, I contribute to understanding the role of the private sector in education in a low-income setting. Contrary to common perception, I find no difference between the quality of school management in government, private or public–private partnership (PPP) schools (despite the higher level of autonomy available to them). An exception is an internationally owned chain of PPP schools, which are as well managed as schools in the UK.

How many children in low- and middle-income countries attend school near sources of pollution? We match the locations of 2.6 million schools across 17 countries to 11,301 documented toxic sites and find that 9.7% of schools in our sample lie within 5 km of a site. Weighting by enrollment, 12.7% of students in the 7 countries with enrollment data attend a school within this distance. These remain lower bounds: the available data capture only a fraction of actual contaminated sites. Proximity is overwhelmingly an urban phenomenon—urban schools are 4 to 28 times more likely than rural schools to be near a site, depending on the country. Within countries, schools in the wealthiest neighborhood quintile are about 14 times more likely to be near a site than schools in the poorest (34.5% versus 2.4%), reflecting the spatial concentration of industry in wealthier urban areas of LMICs. Where data on school management are available, private schools are also more likely than public schools to be near sites in all eight such countries.