Adaptive Design of Clustered Surveys and Experiments, 2026+ (JMP). Cluster designs depend on the correlation of units within clusters. I propose two algorithms that implement the optimal design without knowing this correlation. Additionally, I show that adaptive sampling depending on variances leaves inference intact.
Signaled Bandits, 2026+. I present the signaled bandit game, a generalization of bandit and experts' problems through the notion of signal informativeness.
Sequential Bias Identification, 2026+. I propose a framework for adaptive experimentation in the presence of observational estimates with unknown biases.
Joint and Independent Sampling with Time-Correlated Outcomes, 2026+. I characterize best-group identification strategies when arms are time-correlated and the analyst can sample several arms at a premium.
Escaping Adverse Selection through Learning, 2026+(joint with Hubert Wu). In an Akerlof-style model, we show agents can escape market unraveling when they are constrained to learn through trade.
Sequencing and Information Provision in Search Games, 2026+. I characterize a search recommendation model where the designer can only learn from the search behavior of the agent given the recommendation.