AEA RCT Registry currently lists 12370 studies with locations in 171 countries.
This pre-analysis plan describes a randomized controlled trial evaluating an AI-powered mental health chatbot among undergraduate students in India’s public universities. Using baseline data from 4,489 students across seven colleges in Delhi, we document a high prevalence of depression, anxiety, and loneliness, and low rates of mental health care-seeking. Classrooms are randomized to a control group or to receive four weeks of access to the chatbot. Within treated classrooms, students are cross-randomized to alumni endorsements and habit-formation streaks. The plan pre-specifies intent-to-treat estimates of the effects of chatbot access and the additional interventions on adoption and engagement, care-seeking behavior, stigma (both own and perceived), willingness to pay for therapy, and...
Our experiment involves participants designing a product (a web game) with an AI coding agent to sell to other participants.
This randomised evaluation studies whether low-cost information interventions and financial incentives can increase voluntary adoption of privately beneficial, emissions-reducing motorcycle maintenance in a weak enforcement context. Air pollution disproportionately affects developing countries, where 91% of related deaths occur, with the urban poor — those most exposed to airborne particulate matter — facing the greatest health risks. Despite its urgency, little is known about how to effectively reduce pollution in weak enforcement contexts. In Viet Nam, motorcycles account for the vast majority of transport-related emissions, yet on-road emissions remain largely unregulated and uptake of emissions-reducing maintenance is low, despite potential private fuel cost savings. The study is...
Despite significant gains in financial access across low- and middle-income countries, financial health and resilience remain critically low among informal sector workers, largely due to exposure to uninsured health shocks. In Nigeria, only about 3% of adults aged 15-49 have any form of health insurance, leaving the majority vulnerable to catastrophic out-of-pocket expenditures that undermine business viability and loan repayment. This study provides causal evidence on the impact of bundling health insurance with microfinance loans on the financial outcomes of micro and small informal business owners in Edo State, Nigeria. In partnership with a microfinance institution and the Edo State Health Insurance Commission, eligible clients are offered a voluntary bundled product combining a mic...
Do decentralized markets efficiently aggregate supply decisions when firms must commit inventory under correlated uncertainty? We study street vendors selling perishable produce in Patna, India, who first commit to inventory at wholesale markets then set retail prices during the trading day. Demand uncertainty is driven by weather and marketplace-day footfall affecting all vendors simultaneously. Waste from spoilage provides a direct measure of costly supply mistakes. We develop a two-stage model identifying three inefficiency channels: business-stealing through prices, availability externalities through stockout spillovers, and coordination failure under correlated shocks. Our empirical strategy combines vendor-level panel data with randomized price experiments. Vendors are notified of...
This project examines how beliefs about local public services shape policy preferences, perceptions of public service deprivation, political attitudes, and behavioral intentions. To this end, I implement an information-provision experiment in Germany that exogenously shifts beliefs about local hospital accessibility. Respondents are randomly assigned to receive information about their municipality’s actual position in terms of hospital accessibility relative to all other municipalities within their federal state, based on a novel dataset covering all German municipalities. The study addresses three questions. First, do respondents hold systematic misperceptions about their municipality's relative position in terms of hospital accessibility? Second, how do respondents update their belief...
This study explores how zero-interest loans, in the form of inventory, shapes sales performance and economic empowerment of rural female entrepreneurs. The study also explores the relative efficacy of such loans versus general marketing support to make them better salespersons. Beyond asking whether inventory-on-credit helps, we unbundle two mechanisms: the ability to demonstrate a product to customers, and the ability to offer instant delivery from stock on hand. The study uses a four-armed randomized controlled trial in Bihar (India): full inventory advance (demonstration and instant delivery), sealed stock (instant delivery, no demonstration), demo unit only (demonstration, no instant delivery), and a business-as-usual control. The results are expected to inform scalable models for r...
This project conducts a field experiment at a large-scale shrimp farm in China to examine how LLMs affect compliance in organizations. Workers receive AI-generated harvest recommendations, and in one treatment they can also interact with a LLM chatbot to explore the rationale behind the recommendations. Using daily pond-level data, we study how workers respond when AI recommendations conflict with managerial directives, focusing on whether LLMs improves compliance or instead provides a justification for deviation. We further test whether engagement with the AI system strengthens its behavioral impact and whether forecast errors weaken trust in AI. By linking behavioral, engagement, and economic outcomes, the study provides causal evidence on how LLMs may reshape coordination, authority,...
We study whether the Complex Instruction Program (CIP) improves the achievement of disadvantaged students and lowers their risk of dropping out. CIP restructures classroom group work to break entrenched status hierarchies and raise the standing of low-status students, on the premise that dropout is as much rooted in how students are positioned in the classroom as in what they are taught. We test this in a school-level cluster-randomized trial across 84 Hungarian primary schools serving disadvantaged populations, using a staggered treatment assignment design: treatment teachers are trained and mentored over a school year and deliver CIP lessons to students in grades 5–8, while control schools adopt the program two years later. Our primary outcomes are standardized mathematics and reading...
China's "parallel volunteer" (平行志愿) college-admission mechanism is strategically simple in theory — it is safe to rank a reach school first, because being rejected there does not forfeit later choices — yet many students apply it conservatively and end up "high score, low placement" (高分低就). We run a three-arm randomized field experiment with high-school students in the window between Gaokao score release and rank-ordered-list (ROL, 志愿) submission, to test whether experiential and AI-assisted learning improve their real college applications. All participants receive standard information about the mechanism and an incentivized comprehension quiz. Two treatment arms additionally play a six-round incentivized simulation of the parallel-volunteer game; one of these also receives round-specif...