Evidence
Data, experiments and evidence
We strengthen how institutions collect, analyse and apply data, so that policies and services are designed on evidence that is useful, timely and decision-ready: A/B tests, field trials, discrete choice experiments and behavioural diagnostics.
The evidence
Agricultural market intelligence for an irrigated valley: weekly price monitoring, storage and forecasts
A worked weekly market-intelligence system for paddy rice and onion in a Sahelian irrigated valley: seasonal norms and alert bands, valley-to-capital price transmission, marketing margins, a storage break-even calculator, and backtested forecast bands, accompanied by the bulletin template and data-collection design required for operational deployment.
A four-arm randomised trial of savings reminders
A pre-analysis-plan-grade protocol evaluating whether SMS or voice reminders—plain, goal-framed, or loss-framed—increase the regularity of mobile-wallet savings, accompanied by a reproducible and verifiable power calculation.
A behavioural diagnostic, from holding a mobile wallet to saving in it every month
A worked diagnostic for one precisely defined behaviour: a monthly deposit into a mobile wallet's savings feature. The behaviour is selected against alternatives and mapped using the COM-B model onto an end-to-end user journey with a synthetic funnel. Its 14 candidate barriers are scored and ranked via an explicit decision rule, and each priority barrier is systematically linked to a behavioural mechanism, an intervention, a theory of change, and an evaluation protocol.
A cluster-randomised field trial through savings groups
How to test a financial-capability intervention delivered in community savings groups, from the feasibility test to a powered trial, featuring reproducible stratified randomisation, baseline balance diagnostics, cluster-robust ANCOVA estimation, subgroup heterogeneity analysis, and implementation fidelity tracking.
A discrete choice experiment on a digital savings product
Which features induce users in a low-literacy, mobile-first market to adopt a digital savings product, and what economic value do they attach to them? A fully reproducible, D-efficient experimental design, pictorial choice cards, conditional and mixed logit models estimated via simulated maximum likelihood, willingness-to-pay distributions with confidence intervals, and a market-share simulator.