Projects

Partners: Stanford University (Anirudh Sankar), Princeton University (Robert Dulin)

Location: Eastern Uganda

Working Paper: How Mechanistic Explanations Reshape Learning and Behavior: Evidence from a Fertilizer Choice Experiment in Eastern Uganda

Target Group: Smallholder Tomato Farmers (408 farmers used the tool)

Research Question & Context:

How does mechanistic learning about soil macronutrients (Nitrogen (N), Phosphorus (P), and Potassium (K)) influence agricultural investment decisions and fertilizer application efficiency among smallholder farmers?

Design Challenge:

• Low Connectivity & Literacy: High reliance on visual communication in rural field settings with zero cellular connectivity.

• Complex Payoff Functions: Translating nonlinear chemical soil responses and dynamic crop growth curves into intuitive visual feedback without overwhelming participants.

The Solution & Prototyping:

• Developed a mobile simulation game allowing farmers to experiment with N-P-K nutrient ratios on virtual crops in real time.

• Embedded a custom optimization algorithm suggesting efficient fertilizer combinations based on selected inputs.

• Built an offline-first practice mode to allow participants to master game mechanics before treatment data collection began.

Data & Impact:

• Field Execution: Deployed in field experiments across Eastern Uganda.

• Data Pipeline: Logged granular trial-and-error input sequences, parameter choices, and final decisions offline.

Partners: International Food Policy Research Institute (IFPRI) / AICCRA (World Bank); PI: Berber Kramer and Carly Trachtman

Working Paper: Information as a Source of Empowerment: The Role of Climate Information Services

Location: Zambia

Target Group: Individual Farmers & Intra-Household Couples

Research Question & Context:

How do male and female farmers (individually and jointly within household decision-making units) use Climate Information Services (CIS) and weather forecasts to guide agricultural investments and adapt to climate risk?

Design Challenge:

• Intra-Household Dynamics: Designing synchronized interfaces for both solo players and joint-decision couples playing simultaneously on shared or adjacent devices.

• Information Frictions: Communicating probabilistic weather forecasts and economic risk without technical jargon.

The Solution & Prototyping:

• Engineered a streamlined multi-player/couple simulation app operating entirely offline.

• Integrated an enumerator admin panel for local session setup, treatment assignment, and seamless post-experiment data synchronization once connected to Wi-Fi/cellular networks.

• Designed visual weather scenarios and investment choice trees to model household bargaining.

Data & Impact:

Research Outcome: Published in an IFPRI Working Paper demonstrating that weather forecasts significantly increase farmer welfare and help spouses align on agricultural investment decisions.

Partners: University of San Francisco (Prof. Bruce Wydick, Prof. Andrew Hobbs)

Field Context: Multi-Country RCTs in Uganda, India, and Peru

Research Question & Context:

How does experiential, game-based learning impact financial decision-making, numeracy, and business aspirations among female microcredit borrowers in low- and middle-income countries (LMICs)?

Design Challenge:

• Low-Literacy & Multilingual Settings: Delivering abstract financial concepts (interest rates, working capital, cash flow management) to non-literate or semi-literate participants across multiple language contexts (Luganda, Kannada, Spanish).

• Cognitive Load & Behavioral Control: Designing an intuitive interface that teaches through repeated micro-decisions without sacrificing game-theoretic consistency or introducing enumerator bias.

The Solution & Prototyping:

• Engineered a progressive mobile simulation game that replaces text-heavy instruction with iconic UI, visual tokens, and localized voiceovers.

• Designed a progressive difficulty curve: players navigate basic household cash management before unlocking multi-period business investments and loan repayment mechanics.

• Built offline-first state persistence to ensure unbroken gameplay and session recovery on low-cost Android devices in remote field sites.

Data & Impact:

• Logged every in-game decision, timestamp, choice latency, and trial-and-error sequence to track learning curves and behavioral shifts over time.

• Field Deployments: Field-tested and deployed across three randomized controlled trials: Uganda and India (2023) and Peru (2024).

• Academic Outcomes: Produced empirical data demonstrating statistically significant increases in financial literacy scores and entrepreneurial agency among participants.

Partners: University of San Francisco (Prof. Alessandra Cassar)

Target Context: Controlled Lab & Field Experiments on Competition

Research Question & Context:

How do individuals navigate competitive pressures, time constraints, and strategic steals in a controlled game-theoretic setting?

Design Challenge:

• Sub-Second Precision: Managing real-time competitive interactions (pen stealing, rapid number transcription) across multiple devices with zero state desynchronization.

• Experimenter Control: Giving researchers fine-grained, live admin control over game rules, round timers, and session parameters.

The Solution & Prototyping:

• Built a two-part app system: a client gameplay app and a real-time administrator command app.

• Integrated fast participant onboarding via QR codes and local socket/database state synchronization to guarantee latency-free competition logging.

• Designed an interactive digital interface where players compete to control a virtual pen while copying target numbers under time pressure.

Data & Impact:

• Data Telemetry: Captures micro-second decision latencies, steal frequency, accuracy under pressure, and risk-taking behavior.

• Research & Teaching: Serves as a flexible digital instrument for academic research and experimental economics instruction.