30 New Findings in Global Education: RISE Conference 2019 (Repost from CGD)

Last week, Research on Improving Education Systems (RISE) held its annual conference, which economist Karthik Muralidharan has dubbed “the top conference on education in developing countries in the world.” Over the course of two days, researchers presented over 30 new findings on education systems. If you weren’t able to attend, you can find the full line-up and livestreams here. If you want a quick taste of this year’s research highlights, read on!

A word cloud of paper titles from the 2019 RISE conference

Learning inequalities and social mobility

• You want kids to go further in school because more school improves their life outcomes. Across five countries, if you close the learning gap between poor kids and rich kids at age 12, it narrows the gap in completed years of schooling by age 22 by between a quarter and half (Singh, with Das & Chang; video).
• In Benin, exposure to education in the colonial era increases “social mobility across three generations.” The largest effects are for grandchildren of people with only indirect exposure: They didn’t get an education, but they were in villages with a school (Wantchekon; video).
• Indonesians exposed to a major school construction program in their childhood are doing better now, in their 50s. They got more education, make more health investments, and did better on the marriage market. Their children are also getting more education, especially daughters whose mothers were exposed to the program (Kleemans, with Akresh & Halim; video).
• After examining 275 educational interventions across 50+ countries, two-thirds don’t report gender differentiated impacts, and general interventions deliver similar gains for girls as interventions that are targeted to girls (Evans, with Yuan; video; blog post).

Motivation, management, and incentives for teachers

• Leave your problems at home? Not so easy! In rural Ghana, personal and professional challenges in the lives of new teachers are associated with lower learning and reduced socioemotional development among their students (Fatima, with Wolf; video).
• When teacher incentives are removed, does performance drop? In Tanzania, non-monetary incentives (i.e., cool cell phones) increased test scores. After the incentives program ended, scores didn’t fall (Sabarwal, with Filmer & Habyarimana; video).
• In Rwanda, pay-for-performance contracts (using the pay-for-percentile design) attracted more money-oriented candidates, but those candidates were at least as effective as traditionally recruited candidates. Pay-for-performance boosted student learning (Zeitlin, with Leaver, Ozier, & Serneels; video).
• Poorer countries don’t just have higher pupil-teacher ratios: They also have higher variation in pupil-teacher ratios within the country. From a new data set of almost two million schools across 86 countries, we learn that changing how teachers are allocated across schools could significantly boost learning (Walter; video).

Inside the classroom

• Technology aided education improved learning for kids in a small scale after-school program in Delhi. Integrated into schools in Rajasthan, the program still significantly increases learning, and at much lower cost. But it doesn’t boost the test scores that teachers are evaluated on, creating a conundrum for the future (Muralidharan, with Singh; video).
• In middle schools in Pakistan, in-class technology together with teacher training (to help them integrate the technology effectively) increased both effort and test scores among students (Beg, with Lucas, Halim, & Saif; video).
• “Does gamification in education work?” Math games in low-performing primary schools in Chile lead to sizeable gains in math learning. It also increased “the idea among students that study effort can raise intelligence” (good) and math anxiety (bad) (Cristia, with Araya, Ortiz, & Bottan; video).
• Adolescent girls in Zambia received e-readers plus an empowerment curriculum and study groups: Literacy and non-verbal reasoning skills rose (Mensch, with 11 co-authors; video).
• What explains why Vietnam’s primary schools are so effective? A panel study shows that just as elsewhere, easily measured teacher attributes don’t explain it (Glewwe, with 5 co-authors; video).
• In Gambia, combining three effective interventions—para-teachers delivering after-school classes, scripted lessons, plus frequent monitoring and coaching—delivered big gains in learning, but it wasn’t cheap (Eble, with 6 co-authors; video).
• In South Africa, teachers who received coaching maintained their improved teaching practices in subsequent years. Student learning for new cohorts was still elevated, but not as high as in the year the teachers received coaching (Cilliers, with Fleisch, Kotze, Mohohlwane, & Taylor; video).
• Want to know how teachers are doing? Ask the students! A survey of 4,000+ primary school students and their parents showed that student perceptions of teacher performance—particularly teacher engagement and praise of students—were associated with students advancing in school (Weldesilassie, with Woldehanna, Oketch, & Sabates; video).

School accountability by parents and communities

• Eight years after a school report card intervention in Pakistan, improvements in test scores and price declines in private schools are maintained, together suggesting large improvements in productivity in the education market (Das, with Andrabi and Khwaja; video).
• In Uganda, a scorecard about the quality of schools increased student learning, but only when communities had input into what would be monitored on the scorecard. New work tries to understand whether this would work administered at the district level (Kabay; video).
• School council members—teachers, parents, & community—in Pakistan received regular phone calls encouraging them to spend the funds they’d received. It worked: Spending rose. But there was no impact on attendance or learning (Asim; video).

What parents do

• Lots of people pay for after-school tutoring, but in the slums of Delhi, offering prices to different families shows that paying a higher price leads to higher attendance and lower prices reduce dropout rates (from tutoring). There is “no evidence that tutoring impact average test scores” (Mukherjee, with Berry; video).
• Do parents invest in their children’s education in order to maximize household income, or do they seek to reduce inequality across their children? A lab-in-the-field experiment in Malawi shows that parents care about both, but that they sacrifice a significant proportion of income to avoid inequality (Jagnani, with Berry & Dizon-Ross; video).

Reducing the cost of schooling

• A scholarship program aimed at improving high school graduation rates in Mexico had no impact. The program was mistargeted, largely missing the poorest students. But many eligible students also lack the minimum skills to finish high school. As de Hoyos said in the presentation, “If kids don’t know how to add, and you want them to do calculus, you can give them a million US dollars & they still won’t be able to do calculus” (de Hoyos, with Attanasio & Costas-Meghir; video).
• Merit-based or needs-based scholarships? Nine years after the scholarships were distributed in Cambodia, beneficiaries of both had more schooling, but only beneficiaries of merit scholarships had higher learning or well-being (Barrera-Osorio, with de Barros & Filmer; video).

Innovating at the system level

• If we want education systems to align around learning, then the idea that providing quality education is the responsibility of the government needs to shift to all actors in the system. To do that, we need to create space for innovation throughout the system. Levy illustrates this with examples from South Africa (Levy; video).
• Trust and control reinforce each other in South Africa’s education system. Whether that is positive or negative depends on how the accountability is implemented: Is it followed up with support? Is it meaningful for schools? (Ehren, with Paterson & Baxter; video)
• In Kenya, an effective pilot to improve literacy was scaled up nationally, retaining the essential program inputs—classroom visits with feedback to teachers, updated textbooks, detailed learning guides—and delivering improved early grade literacy throughout the country. (Piper and DeStefano, with Kinyanjui and Ong’ele; video).

Measurement and methods

• Can we construct quantitative, meaningful measures of bureaucracy in education? A new instrument measures coherence in terms of whether bureaucrats share a common understanding of which jobs are whose. Across four countries in Latin America (the Dominican Republic, Guatemala, Peru, and certain states in Brazil) and levels, bureaucrats identify just 56 percent of tasks officially assigned to them as their own (Adelman, with Lemos, Nayar, & Vargas; video).
• A new, first-of-its-kind dataset measures learning across 164 countries and territories—98 percent of the world’s population. It suggests that girls do better than boys on learning but have less schooling. Learning is associated with growth, but it matters more for countries at certain levels of development (Angrist, with Djankov, Goldberg, & Patrinos; video).
• How big a problem is cheating on large-scale national exams? Indonesia introduced an integrity score in 2015, which showed a full third of middle schools with evidence of substantial cheating. But new, computer-based exams is making cheating more difficult (Berkhout, with Pradhan, Wati, Suryadarma, & Swarnata; video).
• There is so much variation within categories of education interventions that trying to figure out “which interventions work” is unlikely to yield success. Rather, try to figure out why specific interventions work (Masset; video).
• A large-scale randomized trial Andhra Pradesh, India shows that “paper-based tests overstate achievement a lot” relative to tablet-based tests (Singh; video).
• Teacher skills matter for learning, but so do the initial distribution of student skills, the pace of the curriculum, and other factors. In different cases, the factor that matters the most for improving learning outcomes will vary (Kaffenberger, with Pritchett; video; blog post).

Beyond research

In addition to the research presentations, the conference included insightful policy discussions.

• Kwabena Tandoh, Deputy Director General of Quality & Access for the Ghana Education Service discussed “Hard Choices: From Long Lists to Prioritised Action” with Karen Mundy (video).
• Laura Savage of the UK’s Department for International Development led a discussion with Pinelopi Goldberg (World Bank), Raphaelle Martinez (Global Partnerships for Education), and Bronwen Magrath (Oxford University) on Big Efforts in Education.