How to Improve Education Outcomes Most Efficiently? A Comparison of 150 Interventions using the New Learning-Adjusted Years of Schooling Metric

Many low- and middle-income countries lag far behind high-income countries in educational access and student learning. Policymakers must make tough choices about which investments to make to improve education with limited resources. Although hundreds of education interventions have been rigorously evaluated, comparing their impacts is challenging. This paper provides the most recent and comprehensive review of the literature on effective education programs, with a novel emphasis on cost-effectiveness, covering the effectiveness and cost-effectiveness of interventions from over 200 impact evaluations across 52 countries. The analysis uses a unified measure – learning-adjusted years of schooling (LAYS) – that combines access and quality and compares gains to an absolute, cross-country standard. The results identify programs and policies that can be orders of magnitude more cost-effective than business-as-usual approaches, enabling policymakers to improve education outcomes substantially more efficiently.

Many low- and middle-income countries lag far behind high-income countries in educational access and student learning. Policymakers must make tough choices about which investments to make to improve education with limited resources. Although hundreds of education interventions have been rigorously evaluated, making comparisons between the results is challenging. This paper provides the most recent and comprehensive review of the literature on effective education programs, with a novel emphasis on cost-effectiveness. We analyze the effectiveness and cost-effectiveness of interventions from over 200 impact evaluations across 52 countries. We use a unified measure — learning-adjusted years of schooling (LAYS) — that combines access and quality and compares gains to an absolute, cross-country standard. The results identify programs and policies that can be up to an order of magnitude more cost-effective than business-as-usual approaches. Examples of some of the most cost-effective approaches include targeting instruction to students’ learning level rather than grade as well as structured pedagogy approaches. These results can enable policymakers to improve education outcomes substantially more efficiently.

OVER THE PAST 15 YEARS, cash transfer programs have become a core component of antipoverty policy strategies in the developing world. In Latin America in particular, cash transfer programs have adopted a multidimensional approach to poverty, whereby income support is provided together with simultaneous interventions in health, education, and nutrition. This "human development" approach to poverty reduction places a strong emphasis on tackling the intergenerational transmission of poverty through human capital investment (Levy and Schady 2013; Nio-Zaraza 2011; Levy 2006). Mexico's Progresa-Oportunidades-Prospera, Brazil's Bolsa Familia, Colombia's Familias en Accin, and Chile Solidario are prominent examples of this antipoverty policy framework.

Accreditation and professional organizations have recognized the importance of measuring medical students' perceptions of the learning environment, which influences well-being and professional competency development, to optimize professional development. This study was conducted to explore interactions between students' perceptions of the medical school learning environment, student demographic variables, and students' professional attributes of empathy, coping, tolerance of ambiguity, and patient-centeredness to provide ideas for improving the learning environment. Twenty-eight medical schools at 38 campuses recruited 4,664 entering medical students to participate in the two-cohort longitudinal study (2010-2014 or 2011-2015). The authors employed chi-square tests and analysis of variance to examine the relationship between Medical School Learning Environment Survey (MSLES) scores and student characteristics. The authors used mixed-effects models with random school and campus effects to test the overall variances accounted for in MSLES scores at the end of the first year of medical school. Student attributes and demographic characteristics differed significantly across schools but accounted for only 2.2% of the total variance in MSLES scores. Medical school campus explained 15.6% of the variance in MSLES scores. At year's end, students' perceptions toward the learning environment, as reported on the MSLES, differed significantly according to the medical school campus where they trained. Further studies are needed to identify specific factors, such as grading policies, administrative support, and existence of learning communities, which may influence perceptions of the learning environment at various schools. Identifying such variables would assist schools in developing a positive learning environment.

Poor sleep quality, often resulting from poor sleep hygiene, is common among medical students. Educational interventions aimed at improving sleep knowledge are beneficial for sleep quality in healthy populations. However, sleep education is often given minimal attention in medical school curriculums. The aim of the study was to explore whether a short educational intervention could improve sleep knowledge, and consequently sleep quality, among medical students. We recruited preclinical- and clinical-stage medical students during the 2017-2018 academic year. Students completed a demographic survey, the Pittsburgh Sleep Quality Index (PSQI), the Epworth Sleepiness Scale (ESS), and the Assessment of Sleep Knowledge in Medical Education (ASKME) questionnaire. Students then attended a lecture on the physiology and importance of sleep. To assess the efficacy of the intervention, questionnaires were repeated 4 months thereafter. A total of 87 students (31 preclinical) with a mean age of 25.86 years (standard deviation [SD], 3.33), 51 of whom were women, participated in the study. At baseline, students had poor sleep quality with a PSQI mean score of 5.9 (SD, 2.37), without significant sleepiness, and a mean ESS score of 8.86 (SD, 4.32). The mean ASKME scores were consistent with poor sleep knowledge at 11.87 (SD, 4.32). After the intervention, the mean ASKME results improved to 14.15 (SD, 4.5; P < .001), whereas sleep quality did not. The effect was similar in preclinical and clinical medical students. Sleep knowledge was inadequate among medical students, who also experienced poor sleep quality. A short educational intervention improved sleep knowledge but was insufficient at improving sleep quality. Further studies are needed to determine which interventions may provide benefit in both sleep knowledge and sleep quality.

The performed cross-national tests with negative binomial regression models support the presence of a curvilinear relationship between the quantitative expansion of education (measured with mean years of schooling) and terrorist attack intensity. Growth of schooling in the least educationally developed countries is associated with a significant tendency towards the growth of terrorist attack intensity. This tendency remains significant when controlled for income level, type of political regime, unemployment, inequality, and urbanization; wherein the peak of the terrorist attack intensity is observed for a relatively low, but not zero level of the quantitative expansion of formal education (approximately three to six years of schooling). Further growth of schooling in more developed countries is associated with a significant trend toward the decrease of terrorist attack intensity. This tendency remains significant after being controlled for income level, political regime, unemployment, inequality, and urbanization. The most radical decrease is observed for the interval between seven and eight mean years of schooling. In addition, this quantitative analysis indicates the presence of a similar curvilinear relationship between GDP per capita and terrorist attack intensity with a wide peak from $4000 to $14,000. The explanation of a curvilinear relationship between GDP per capita and terrorist activity through mean years of schooling intermediary can only be partial. The regression analysis suggests that the growth of mean years of schooling with economic development of middle and high income countries may really be one of the factors accounting for the decrease of terrorist attacks in countries with GDP per capita growth. However, this regression analysis indicates that a partial role in the explanation of negative correlation between GDP per capita and terrorist attack intensity for middle and high income countries is also played by a lower level of unemployment rate in the high income countries, as well as by a very high share of consolidated democracies and an extremely low share of factional democracies among the high income states. It is especially worth noting that after the introduction of all controls, the coefficient sign for per capita GDP changes from negative to positive, i.e., GDP growth in middle and high income countries after the introduction of controls for inequality, education, unemployment, type of regime, etc. turns out to be a factor of increase rather than decline of the intensity of terrorist activity. On the one hand, this suggests that the negative correlation between per capita GDP and the level of terrorist activity in these countries is actually explained to an extremely high degree by the fact that per capita GDP growth here tends to be accompanied by an increase in the educational level of the population, a decrease in unemployment, a reduction in inequality, a decrease in the number of factional democracies, and an increase in the number of consolidated democracies. On the other hand, the positive sign (with a statistically significant correlation) indicates here that if in the middle and high countries economic growth is not accompanied by an increase in economic equality and education of the population, a decrease in unemployment, a decrease in the number of unstable factional democracies, and an increase in the number of consolidated democracies (that is, if in fact all the fruits of economic growth are captured by the elites, and almost nothing gets from this growth to the commoner population), then such economic growth would tend to lead to an increase in terrorist activity (and not to its reduction).

The phenomenon of globalization that concerns the modern era, pushed by technological evolution, has led to several changes in the field of education. Not only are education policies of the single States adapting by directing towards European models: the possibility for a student to choose a university is increasing all over the world, too. In order to facilitate students who want to enroll at a European university and who come from countries with less than 12 years of compulsory schooling, our University designed the Foundation Programme. This is an additional year that allows to earn 60 ECTS to reach the 12 years of schooling, a basic requirement for university access in Europe. The main feature of this project is that it is delivered online; this allows to reduce the costs of enrollment and allows students to attend it directly from home in their country. The design, structure and methodologies of the project are described and discussed in this article.

We estimate the direct and external effects of levels of schooling on personal income in Ecuador in 2011, using data for 69,653 individuals in 567 municipalities. Using a Mincerian model that includes municipal levels of schooling and the size of the municipality and controls for endogeneity, we find that each year of individual schooling raises individual income by 8.5 percent and each year of municipal schooling raises individual income by 2.2 percent. The external effect of an additional year of schooling is larger for workers with more schooling, for those with higher incomes, and for those in more educated municipalities.

The benefits of interventions which improve early nutrition are well recognised. These benefits, however, only accrue to the extent that later life circumstances allow. Consequently, in adverse contexts many of the benefits will never be realised, particularly for the most vulnerable, exacerbating inequality. Returns to investment in early nutrition could be improved if we identified contextual factors constraining their realisation and interventions to weaken these. We estimate cost and impact of scaling 10 nutrition interventions for a cohort of South African children born in 2021. We estimate associated declines in malnutrition and mortality, and improvements in years of schooling and future earnings. To examine the role of context over the life-course we estimate benefits with and without additional improvements in school quality and employment opportunities by socio-economic quintile. Scale up reduces national stunting (height for age < = -2SD) rates among children at 24 months by 3.18 percentage points, implying an increase in mean height for age z-score (HAZ) of 0.10, and 53,000 years of additional schooling. Quintile 1 (the poorest) displays the largest decline in stunting, and largest increase in mean HAZ. Estimated total cost of increasing coverage of the interventions for the cohort is US$90 million. The present value of the additional years of schooling is estimated at close to US$2 billion. Cost-benefit ratios suggest the highest return occurs in quintile 5 (1:23). Reducing inequality in school quality closes the gap between quintile 5 and the lower quintiles. If school quality and labour force participation were equal the highest returns are in quintile 1(1:31). An enabling environment is key to maximising human development returns from investing in early nutrition, and to avoid exacerbating existing inequality. Therefore, particularly for children in adverse conditions, it is essential to identify and implement complementary interventions over the life course.

The effect of partners' education on women's reproductive and maternal health in developing countries

Formal education is correlated with entrepreneurial activity and success, but correlation does not indicate causation. Education and entrepreneurship are both influenced by other related factors. The current study estimates the causal effects of formal education on entrepreneurship outcomes by instrumenting for an individual’s years of schooling using cohort mean years of maternal schooling observed decades prior. We differentiate self-employment by industry employment growth and firm incorporation status. Policymakers are especially interested in entrepreneurship with the potential to create substantial employment growth. We find that an additional year of schooling increases self-employment in high-growth industries by 1.12 percentage points for women and by 0.88 percentage points for men. Education reduces the probability of male self-employment in shrinking industries. Education also increases incorporated self-employment for women and men and reduces unincorporated self-employment among men but not women. The overall probability of self-employment increases with education for women but is unaffected by education for men. The results suggest that formal education enhances entrepreneurship.

Alcohol is a leading risk factor for the global burden of disease and contributes to a range of social and economic harms. Globally, alcohol is estimated to be the seventh leading risk factor in 2016 in terms of disability adjusted years of life lost, and alcohol use is the leading risk factor in disability adjusted years of life lost between the ages of 15 and 49 years 1. The 2016 global burden of disease analysis has confirmed more limited preventive effects from alcohol than have been previously claimed and identified a much larger risk of cancer due to alcohol 1. A non-communicable disease target of 10% relative reduction in alcohol consumption has been established by the World Health Organization (WHO) 2; alcohol is also recognised by the United Nations as a threat to sustainable development 3 and contributes economic costs of approximately 1%–2% of gross domestic product in several countries where these have been assessed 4. Policy measures to restrict alcohol availability, curtail affordability and restrict alcohol marketing, when implemented, have reduced alcohol-related harm 5-7, however, such policies have not, as yet, been widely implemented and, while summarised in the WHO Global Strategy to Reduce Harmful Use of Alcohol, they have not been encapsulated into an international health treaty comparable with the Framework Convention on Tobacco Control. Substantially less groundwork is available in alcohol control, when compared with tobacco, on monitoring and encouraging legislation and implementation of effective alcohol policy. For example, WHO developed the policy package MPOWER to monitor and assist with country-level implementation to reduce demand for tobacco 8. The Global Information System on Alcohol and Health (also developed by WHO) makes country-level alcohol consumption and policy data available, but does not provide resources for intervention implementation (although there have been some efforts at regional level 9 and a tool on taxation and pricing was recently published by WHO 10). The lack of progress in alcohol control at the national and international levels is highlighted by the fact that alcohol attributable DALYs have increased by more than 25% over the years 1990–2016, driven primarily by increased consumption in South Asia, Southeast Asia and Central Asia, among both men and women 1. Africa is now experiencing similar impacts to those in Asia as a result of targeting by the supranational alcohol corporations 11, 12. The implementation of alcohol policies is not only often politically difficult 5, but also more complex than that of tobacco for a number of reasons, including the availability of a range of beverages of different potencies and a wide range of prices in on- and off-premise drinking contexts. In addition, unlike for tobacco, there are policies related to intoxication such as restrictions of sale to intoxicated patrons and drink-driving legislation. The International Alcohol Control (IAC) study had its origins in several discussions with a colleague who participated in the International Tobacco Control (ITC) study 13, Professor Gerard Hastings, about the value of a study similar to the ITC pertaining to alcohol. A proposal was made to the Health Promotion Agency of New Zealand, and in 2010 New Zealand researchers, along with invited researchers from four other countries—three high-income (England, Scotland and Korea) and one middle-income country (Thailand), met in Scotland to plan the IAC; the planning drew on the expertise of staff in the Institute for Social Marketing, University of Stirling, who were participants in the ITC. The International Alcohol Control Study was developed by Casswell et al. 14 to provide detailed information on alcohol use, policy relevant behaviours and how these change in response to changing conditions. Subsequently, each participating country needed to raise its own funding to participate and resources have differed, resulting in some differences in approach. The International Development Research Centre of Canada has been a particularly important supporter of the IAC, funding participation by four middle-income countries in the full IAC project, and by three African countries in the use of the Alcohol Environment Protocol, and funding training and much of the dissemination to date. The methodology designed was comparable to the ITC. Longitudinal surveys of drinkers would collect information on consumption and policy relevant behaviour, allowing for assessment of the impacts of policy change when this occurred and comparison with countries in which the same policy change had not occurred. It also allowed, through the measures relating to specific policies, disentangling the effects of different policies if these were introduced as a package. Like the ITC, there was no attempt to collect measures of harm; rather, the IAC relied on very detailed consumption data as a proxy for harm. While the methodology of the IAC study allows for the evaluation of policy change, the reality is that policy change does not always occur at all, or when it is hoped for or anticipated. The secondary aim of the IAC study, therefore, was to collect accurate and detailed information on alcohol consumption and information on the policy environment and policy-relevant behaviours to inform policy debate. This is the focus of the papers in this Special Issue. Participation over many years in WHO meetings and consultations with officials and researchers in middle-income countries, particularly in Asia, made it apparent there was a growing level of concern about alcohol use. This reflected the expansion of the supranational alcohol corporations into new markets in middle-income countries with low drinking prevalence, growing economies and young populations, increasingly connected to the global youth culture, often in a digitally mediated environment. In these countries, as the need for research data to examine the use of alcohol became a priority, researchers new to the alcohol field often collected very basic consumption data. While this can be a useful first step (and the STEPS surveys supported by WHO in many countries was a valuable tool 15) the research lacked detail on drinking and, importantly, any reference to the policy context. The goal, therefore, was to provide a research platform which could be made available to researchers in not only high-income but also middle-income countries to collect robust and comparable data which could inform policy discussion. The policy focus of the study was on the ‘best buys’ of alcohol policy 16; those which research had shown, at least in high-income countries, were likely to be cost effective in reducing alcohol-related harm if implemented properly. These were policies restricting availability, control of price and affordability, restricting marketing of alcohol and legislating to prevent drink-driving. The areas for which there is less evidence, such as labelling, health warnings and education, were not included. The IAC study makes a unique contribution to the epidemiology of alcohol consumption. This is the first international collaborative project to collect general population survey data on alcohol consumption in such detail. It provides measures of typical quantities consumed, frequency of drinking and volumes consumed; the data are available by location of drinking and by beverages chosen. The survey instrument allows for very high coverage of alcohol available for consumption (based on sales or tax data) 17, 18 and is designed to provide comparable consumption data in different alcohol markets, including those with a sizeable proportion of informal alcohol. The IAC study also makes a unique contribution to alcohol policy research through the measurement of policy related behaviours. The survey data provide detail on key policy issues such as the prices paid, the location of purchase, time taken to access alcohol, the times of purchase and response to alcohol marketing. A second IAC tool is the Alcohol Environment Protocol which draws together data from legal and policy documents, administrative and commercial data, published research, observational studies and primary data collection of key informant perceptions. This framework allows for the collection of comparable data on policy settings and implementation. As of 2017, when the first cross-country analyses, published in this Special Issue, were carried out, 16 countries had engaged in some component of the IAC study, and of these 10 had successfully carried out at least one wave of a general population survey providing an insight into a wide range of alcohol markets These were five high-income countries [Australia, England, Scotland, New Zealand and St Kitts and Nevis (St Kitts and Nevis transitioned to high-income during the course of the project)]; three high middle-income (Thailand, South Africa and Peru); and two low middle-income countries (Mongolia and Vietnam). The countries participating in this project vary substantially in size, demography and social structure. One caution is that the data presented and discussed here are referred to by the country name, although several of the surveys did not sample the whole nation. For example, the Vietnamese sample is drawn from a number of provinces, South Africa surveyed in one large municipality, Peru surveyed in one area of Lima and Mongolia surveyed only in Ulaanbaatar. The countries vary greatly in population size and affluence (Table 1). The more affluent countries in this study score higher on the United Nations Human Development Reports Education Index, which is calculated using mean years of schooling and expected years of schooling. The per capita consumption based on those aged 15+ years in Table 1, taken from the Global Information System on Alcohol and Health, show that the high-income countries had the highest per capita consumption, the high middle-income countries next and low middle-income least, with the exception of South Africa, which is drinking aggregate volumes similar to New Zealand with much lower prevalence of drinking 22. The proportion of abstainers among males, among females and among the total population is very different between the high-income and middle-income countries, with most of the high-income countries showing a prevalence of drinking at 80% or higher; the exception is St Kitts and Nevis. The gender ratio in prevalence is also very different, with high-income countries showing least difference between men and women and Thailand the greatest difference. The countries also differ in terms of the estimates of unrecorded alcohol, with Vietnam, the least affluent country, showing the largest proportion of unrecorded alcohol. Note that in all of the international comparison tables in this Special Issue, country data are presented in order of decreasing affluence. This Special Issue presents the first cross country analyses from 10 countries of the IAC. For a number of reasons, including omission of sections of the core questionnaire, programming issues (the survey is computer assisted) and lack of data to complete the Alcohol Environment Protocol, not all countries had data available for all analyses, and so the participating countries in each analysis vary. After describing the methodology, the first section provides an insight into the alcohol policy environment and policy relevant behaviours. The second section reports on consumption patterns and the relationships with policy-related behaviours and support for policy. The resources available and the context of the research varied across the 10 countries whose data are analysed in these papers. This affected implementation, but the goal of the methodology, described in the first section, was to provide a framework to allow data to be as comparable as possible in very different alcohol markets. In relation to aspects which necessarily differed—for example, the sampling designs employed—analytical techniques have been employed to minimise the effect in the analyses 23. Results from the Alcohol Environment Protocol, as reported from seven countries, described differences in the legislative and regulatory frameworks, and in key informants’ perceptions of the way alcohol policy was implemented and enforced 24. The level of implementation and enforcement was lower in lower-income countries and, in high-income countries where enforcement was stronger, policy was more liberal; marketing regulation was largely absent in all countries. Taxation systems and prices paid for alcohol by survey respondents from six countries were analysed; tax systems were seen to vary markedly, reflecting different objectives and histories 25. Data on prices paid and tax collected enabled calculation of the contribution taxes made to the prices paid in both off- and on-premise drinking and allowed comparisons with tobacco taxation. The final paper in this section gives an overview of survey data pertaining to access to alcohol, including by adults and those underage 26. These data supported the findings from the Alcohol Environment Protocol showing ease of access was high and those under the minimum purchase age could purchase alcohol more easily in middle-income countries (except Mongolia). In most of the countries take-away alcohol was a larger proportion of the alcohol market than on-premises drinking, and alcohol was available for access by the majority within 15 min. In the second section, an overview of drinking patterns by age and gender is provided for the 10 IAC countries 27. The patterns varied across countries and the proportion of high-frequency drinkers was higher in high-income countries whereas there were higher odds of drinkers in middle-income countries consuming 8+ drinks for men and 6+ for women (one drink = 15 mL absolute alcohol) on a typical occasion. The ratio of men to women's consumption varied somewhat, but men were the heavier consumers overall. A pattern of increasing frequency with age and declining quantity consumed in a drinking occasion was common but not universal. The relationship between heavy drinking and disadvantage (defined in terms of educational status and living in poverty) is examined in four high-income and three middle-income countries 28. Disadvantage is related to heavier drinking in high-income countries, but the reverse is the case in middle-income countries. A different approach was taken by looking at the alcohol market in each of the 10 countries and calculating what proportion of the market is consumed in harmful drinking occasions 29. These comprised an important component of the market in all countries and were higher in middle-income than in higher-income countries. Informal alcohol was less likely to be consumed in harmful drinking occasions than commercial alcohol. Policy-relevant behaviours (prices paid, time of purchase and liking for marketing) predicted larger typical quantities consumed in on-premise venues in a number of countries, and these behaviours were found to mediate the relationship between demographic characteristics and consumption, particularly in higher income countries 30. Support for alcohol policies among drinkers in seven countries is the subject of the final paper in the Special Issue 31. Across countries differences were found, with a cascade of support for alcohol-control policies, highest in low middle-income and lowest in high-income countries, suggesting the level of support was inversely related to the level of policy implementation. In this series of papers important differences were identified, often related to the level of affluence of the country and, in the case of Vietnam, the presence of a high proportion of informal alcohol. However, the other theme which emerged was the similarities between countries—for example, the easy access to alcohol, the widespread lack of regulation on marketing and the dominance and relative cheapness of take-away alcohol. The data allow comparison with tobacco, for example, in showing the proportion of alcohol's retail price which is made up of tax is much smaller than in the case of tobacco. A similarity with tobacco was the reliance of the industry on harmful use: in all of the countries a significant proportion of the alcohol market was consumed in harmful drinking occasions and this was a larger proportion, over half, in the middle-income countries. This reliance creates a conflict of interest for the producers of alcohol as sales would drop if effective policy reduced harmful drinking occasions and therefore these data support the exclusion of the alcohol industry from the policy environment. The current global context has meant important anticipated policy changes such as the introduction of minimum unit price in Scotland and the legislation banning alcohol marketing in South Africa have been delayed for many years. This has reduced the opportunity for evaluation of policy changes, as envisaged as part of the IAC study. However, the collection of policy-relevant data and the detail provided in the alcohol-consumption data have proven to be useful for individual participating countries and also in international comparisons. The IAC study has provided a research platform for diverse countries to collect alcohol consumption and policy-relevant data in comparable ways. This was achieved by adaptation of the IAC's two research tools, a survey framework and Alcohol Environment Protocol, to allow for country differences. We believe the cross-country analyses presented in the Special Issue of Drug and Alcohol Review provide valid and policy-relevant data to inform national and international policy debate and further research using the IAC platform would be valuable. The International Alcohol Control Study is led by Professor Sally Casswell. The IAC core survey questionnaire was largely developed by researchers at the SHORE & Whāriki Research Centre, College of Health, Massey University, New Zealand, with funding from the Health Promotion Agency, New Zealand. Further development involved collaboration between UK, Thai, Korean and New Zealand researchers. The funding sources for each country are: Australia—Australian National Preventive Health Agency and the Foundation for Alcohol Research and Education; England and Scotland—Medical Research Council National Prevention Research Initiative (Grant ref.: MR/J000523/1); New Zealand—The Health Promotion Agency and Health Research Council of New Zealand; St Kitts/Nevis—International Development Research Centre, Canada; Mongolia—World Health Organization; Peru—International Development Research Centre, Canada; South Africa—International Development Research Centre, Canada and South African Medical Research Council; Thailand—International Health Policy Program, Thai Health; Vietnam—International Development Research Centre, Canada. We would also like to acknowledge support from the UK Centre for Tobacco & Alcohol Studies and the excellent work of the interviewers and their supervisors and the time given by the survey respondents.

This paper investigates whether mothers’ participation in post‐compulsory education impacts on children’s relative inequalities across four developmental outcomes. The empirical analysis uses information from children born in 1958 in Britain. Mothers of the 1958 British cohort were affected by the 1947 school leaving age reform, which increased the age of compulsory schooling from 14 to 15 years. We selected the first‐born cohort members whose mothers were born in 1933 and 1934 and whose mothers completed compulsory schooling only. We found that the additional year of maternal schooling was significantly associated with relative improvements in mathematics attainment for their children, but no significant differences for reading or behavioural outcomes. The impact on mathematics was mainly for boys. These results suggest wider dispersion in mathematics attainment between sons whose mothers benefited from the additional year of schooling in 1947 and those whose mothers did not.

BackgroundAn estimated 2·3 Million individuals are newly infected with HIV each year. Existing cross-sectional and longitudinal studies have found conflicting evidence on the association between education and HIV risk, and no randomized experiment to date has identified a causal effect of education on HIV incidence.MethodsA 1996 policy reform changed the grade structure of secondary school in Botswana and increased educational attainment. We use this reform as a ‘natural experiment’ to identify the causal effect of schooling on HIV infection. Data on HIV biomarkers and demographics were obtained from the 2004 and 2008 Botswana AIDS Impact Surveys, nationally-representative household surveys (N = 7018). The association between years of schooling and HIV status was described using multivariate OLS regression models. Using exposure to the policy reform as an instrumental variable, we estimated the causal effect of years of schooling on the cumulative probability that an individual contracted HIV up to his or her age at the time of the survey. The cost-effectiveness of secondary schooling as an HIV prevention intervention was assessed in comparison to other established interventions.FindingsEach additional year of secondary schooling induced by the policy change led to an absolute reduction in the cumulative risk of HIV infection of 8·1% points (p = 0·008), relative to a baseline prevalence of 25·6%. Effects were particularly large among women (11·6% points, p = 0·046). Results were robust to a wide array of sensitivity analyses. Secondary school was cost-effective as an HIV prevention intervention by standard metrics.InterpretationAdditional years of secondary schooling had a large protective effect against HIV risk, particularly for women, in Botswana. Increasing progression through secondary school may be a cost-effective HIV prevention measure in HIV-endemic settings, in addition to yielding other societal benefits.FundingTakemi Program in International Health at the Harvard School of Public Health, Belgian American Educational Foundation, and Fernand Lazard Foundation.

Girls’ poor ability to manage menstrual health (MH) imposes barriers to education and general wellbeing, especially in low- and middle-income countries. This paper presents the results of the Ritu trial, a 2-year clustered randomized controlled trial, examining the effect of a multi-faceted menstrual health intervention in Bangladesh. We randomized 148 schools from one rural district, into one of three groups; i) receiving a school program (sanitation facilities, MH education and support); ii) a school program and a targeted household program (parental MH education); iii) or the control group. The primary beneficiaries are schoolgirls in grades 6 until 8, age 11-15. We measure short- to medium-term impacts on school attendance, a set of psychosocial outcomes, and menstrual health outcomes. We use three sources of school attendance data: information from administrative records; self-reported survey responses and data from spot-checks where someone from the research team would appear unannounced and on randomly selected dates at school to record attendance. We find significant treatment effects in both treatment arms. The Ritu trial significantly improved menstrual health outcomes. Moreover, it significantly reduced school absence rates and reduced the likelihood of dropping out of school before grade 8. The program also improved psychosocial wellbeing and empowerment during menses, but this did not translate into substantial improvement of general wellbeing and empowerment. We find little evidence that the additional household program generated larger effects than the school program alone – which is important for programming purposes given the substantial additional costs of the household intervention. Our findings show the school program was considerably more cost-effective than the combined program, both when measured in ‘additional years of schooling’ and ‘learning-adjusted years of schooling’. Even though MH programs are becoming more prevalent, supporting quantitative evidence on their impact is very limited. We show a positive impact of a menstrual health program on school attendance and other psychosocial outcomes of adolescent girls.