Optimal experimental design
Allocation strategies that maximize statistical power, minimize variance, or protect performance against uncertainty in model parameters.
Research
Designing experiments for complex evidence.
My work develops principled allocation strategies for studies where dependence, ordering, heterogeneity, and uncertainty matter.
Cluster-randomized trials
Design methodology for trials with correlated observations, heterogeneous cluster sizes, and longitudinal or stepped implementation.
Crossover studies
Bayesian and locally optimal crossover designs for binary, Poisson, and other generalized responses.
Ordered experiments
Efficient designs and tests when treatments have a natural ordering, including intersection–union procedures and constrained alternatives.
Multiple comparisons
Max–min and game-theoretic designs for pairwise treatment comparisons under normal and binary outcomes.
Robust and Bayesian design
Designs that account for parameter uncertainty through priors, robust criteria, and least-favourable configurations.