This study investigates the causal effects of repeat performance evaluations, under Chile’s national teacher evaluation system. The paper’s identification strategy exploits a recent change in the evaluation assignment mechanism, together with a discontinuity in the system’s scoring mechanism. Jointly, these two factors provide plausibly exogenous variation in which teachers get newly evaluated, after two years. The study’s main results suggest that student learning, teacher beliefs and teaching behaviors remain unaffected by a teacher’s re-evaluation, both in the year of the evaluation and in the year thereafter. The article confirms that these findings are not driven by a teacher’s level of work experience, by student sorting, or by systematic attrition.
Do Students Benefit from Blended Instruction? Experimental Evidence from India
This experimental study investigates the causal effect of a teacher capacity building program that promotes blended instruction, on student learning. It will be implemented in government schools in Haryana, India, in collaboration with a large, local NGO (“Avanti Fellows”). The program’s objective is to positively affect the instruction of mathematics and science, in grades nine and ten. The study hypothesizes that student learning improves if teachers are given resources and training, to enrich their instruction with video-based learning materials. Secondly, the study hypothesizes that the intervention’s cost-effectiveness outperforms that of an alternative model of teacher capacity building, which does not rely on infrastructure upgrades and uses printed workbooks only.
This research uses a randomized experiment to investigate the long-run effects of additional education on prosocial behavior. The article builds on an intervention in Cambodia that offered scholarships to students as they were beginning the fourth grade of primary school. The study follows these students as they are now approximately 21 years old. To measure prosocial behavior, I conduct large-scale, in-the-field dictator games (with tangible outcomes, and respondent deception).
With Alejandro Ganimian and Karthik Muralidharan
Despite improvements in enrollment and attendance, learning levels in the developing world remain low (the “Global Learning Crisis”). At the same time, little is known about what students can and cannot do. We use large-scale, representative test data from India to develop and estimate a Cognitive Diagnosis Model (CDM). In particular, we investigate the extent to which students in grades four, six, and eight have mastered a fourth-grade level understanding of five mathematical skills: “Fractions and Decimals”, “Measurement”, “Number Concepts and Number Theory”, “Operations on Whole Numbers”, and “Shapes and Geometry”. Our research documents low levels of student learning overall, and only slightly higher levels of mastery for students in higher grades. However, we also find high levels of heterogeneity in mastery; for individual skills, across geographic units (i.e., states), and in terms of learning gaps between male and female students. Overall, our study highlights the value of a more fine-grained understanding of student learning.
Action: Experimental Evidence on Activity-based Instruction in India
With Johanna Fajardo-Gonzalez, Paul Glewwe, and Ashwini Sankar
The number of rigorous studies on “what works” to foster education in less developed countries has strongly increased, but there is surprisingly little evidence on how to improve child learning through changes in instructional practice. We study the effect of an innovative program in Karnataka, India, that promotes activity-based learning through teacher training, community engagement, and additional inputs. In a Randomized Controlled Trial (RCT), we assign 98 administrative units (Gram Panchayats) and 294 of their schools to either receiving the program or not. Our primary outcome of interest is child learning, in mathematics, for students enrolled in grade four (at baseline). Sub-group analyses focus on differential effects by students’ initial ability level, by gender, and by geographic location (i.e., by district). The study’s secondary analyses investigate changes in observed instructional behaviors and the program’s implementation fidelity.
With Alejandro Ganimian and Karthik Muralidharan
There is mounting evidence indicating that schoolchildren in many developing countries lag far behind their expected grade-level performance. Remedial education can help these low-performing students, but it does not address the needs of their high-performing peers. Ability-based grouping can benefit both types of students, but it may be too coarse to address students’ individual learning needs. We will conduct an experiment in 15 public “model” schools in Rajasthan, India to evaluate the impact of personalized instruction for 3442 students in grades 6 to 8, as delivered by a computer-assisted learning software. We will compare a version of the software that provides students with only grade-appropriate activities (the typical approach used by most software products) with a fully and a partially customized version of the program, as well as with a remedial version of the program. We plan to use these comparisons to understand what type of personalization is most (cost-)effective in improving student learning.
With Alejandro Ganimian, Anuja Venkatachalam, and Karthik Muralidharan
In mathematics instruction, it is quite common for teachers to explain a topic and then assign their students a number of practice exercises – either as classwork or homework – to build “procedural knowledge” (i.e., knowledge about the algorithms to be followed to solve a specific problem) and “fluency” (i.e., capacity to solve these problems rapidly). Although there is general agreement among educators that this strategy is beneficial, prior studies have not determined the extent to which students benefit more from this practice. We will use a computer-assisted learning (CAL) software called “Mindspark” to randomly assign 4,350 students in grades 4-7 in 8 private schools in India to receiving or not receiving practice exercises after they learn a new mathematical concept, and to assess the impact on their procedural knowledge and fluency.
Levers for Learning: The Relation Between School-level Factors and Literacy Outcomes in Low-income Schools in Colombia
With Felipe Barrera-Osorio, Sarah Dryden-Peterson, Bethany Mulimbi, Nozomi Nakajima, and Paola Uccelli
We study the relation between literacy outcomes and four levers for learning – instructional practice; school-community engagement; student, teacher, and parent well-being; and community belonging – in ten public schools in a mid-size city in Colombia, serving similar, low-income communities. We find that (1) some schools show significantly higher literacy outcomes, despite serving similar students; and (2) the most salient school-level factors in explaining literacy performance are higher levels of student sense of belonging and low levels of bullying.
- Under review
Long-Term Impacts of Alternative Approaches to Increase Schooling: Experimental Evidence from a Scholarship Program in Cambodia
With Felipe Barrera-Osorio and Deon Filmer
This randomized trial investigates the long-term effects of a primary school scholarship program in rural Cambodia. We estimate impacts—nine years after program inception—on educational attainment, cognitive skills, socioemotional outcomes, labor market outcomes, and well-being. Our results point to systematic improvements in attainment, but no average impacts on long-term cognitive or socioemotional outcomes. A merit-based (as opposed to poverty-based) targeting strategy did, however, increase cognitive outcomes, especially for poorer students. We also report positive effects on cognition for males. We find no improvements for labor market outcomes, yet positive effects on well-being, driven by recipients of the merit-based scholarships.