Working Papers
Can Schools Teach Innovation? Experimental Evidence from India (See Working Paper)
As automation substitutes for routine cognitive tasks, returns to creativity and problem-solving are rising. But can these skills be taught at scale through public education? I evaluate a two-year innovation curriculum using a cluster randomized trial across 80 government schools and 5,000 marginalized students in India. I develop a measurement toolkit combining an incentivized exploration task, large-scale creativity and problem-solving assessments, and an externally evaluated team idea competition. The curriculum increases idea generation by 0.31 SD, problem-solving by 0.14 SD, fluid intelligence by 0.11 SD, and team-level idea quality by 0.22 SD. Students labeled low-performing academically develop above-median creative problem-solving at 7 percentage points higher rates, showing us talent that test scores alone would never reveal. For the highest-ability quintile, training raises both creative problem-solving (0.26 SD) and mathematics (0.22 SD), evidence of cognitive transfer. Identity shapes what ideas emerge: girls produce health solutions boys overlook; rural students generate agricultural innovations urban students miss. At $40 per student annually, the program demonstrates that these skills can scale even in the most constrained settings.
Partners: Inqui-Lab Foundation, Telangana Social Welfare Residential Educational Institutions Society, Telangana Tribal Welfare Residential Educational Society
Coverage: The World Bank's Development Impact Blog
Funding: The Weiss Fund, NAEd/Spencer Dissertation Fellowship, Teachers College Grants
Going All In: Simultaneously Breaking Down Barriers for Women in the STEM Workforce
(with Ashutosh Bhuradia)
This research aims to measure the impact of a program for women engineers in India, an 18-month STEM training initiative designed for first-generation women engineering students. Deployed nationwide by an education start-up, the program employs a holistic strategy to overcome multifaceted barriers faced by women in STEM fields. By fostering a women-only environment, providing online accessibility, and emphasizing self-directed learning, the initiative seeks to address cultural, institutional, and psychological challenges hindering women's success in STEM. We aim to evaluate the WE program's efficacy in enhancing participants' technical and higher-order skills, ultimately influencing their labor market outcomes. With an underrepresentation of women in STEM persisting globally, this research contributes valuable insights into designing targeted interventions for breaking down barriers and fostering inclusivity in STEM education and careers.
Funding: Digital Harbor Foundation
Scaling Up Personalized Adaptive Learning in India
(with Wendy Wong, Alex Eble, Guthrie Gray-Lobe, Michael Kremer, Emily Cupito, Sabareesh Ramachandran)
This research examines the implementation and impact of a Personalized Adaptive Learning (PAL) program across 480 schools in Andhra Pradesh, India. Working in partnership with the Government of Andhra Pradesh, our study investigates a large-scale educational technology initiative that began in October 2023 and will conclude in February 2026. While educational technologies show promise in controlled environments, their effectiveness when scaled through government systems remains understudied. This project bridges this critical research gap by combining rigorous impact evaluation with implementation science. Using advanced machine learning and econometric techniques, we analyze student-level data on PAL usage and learning outcomes across diverse school contexts. Our research questions explore the range of learning benefits PAL generates at scale, and identify predictors of success or failure at both school and student levels. By studying implementation patterns across hundreds of schools, we aim to provide evidence-based insights for policymakers on how to effectively scale educational technologies in developing countries. This work represents a crucial step in understanding the transition from research efficacy to practice at scale, with potential implications for education policy in India and beyond.
Can phones be used for measuring foundational literacy and numeracy? Experimental evidence from India
Using a crossover randomized design with 1603 government primary school students in Uttar Pradesh, I report two key results about phone-based assessments (PBA) for measuring foundational literacy and numeracy skills. First, PBA is valid, reliable, and equivalent to currently prevalent in-person assessments (IPA). Second, in the specific case of literacy assessments, reliability on WhatsApp for sending literacy prompts is better than using students' school textbooks. Consider these results along with two other important possibilities - PBA are more cost-effective and operationally easy and they are better representative of the student population today due to accelerated penetration of phones and internet in the rural areas. These outcomes offer policymakers an additional assessment mode to measure students' learning outcomes for formative purposes. This could also potentially address the principal-agent problem that is currently prevalent and leads to huge distortion in the reporting of learning outcomes.
Coverage: Ideas for India and Central Square Foundation:
Funding: Central Square Foundation