Research
Bayesian methods for public health, and helping the next generation of statisticians get the details right.
Different regions record deaths differently and see different causes, so a model built in one place transfers poorly to another. I develop Bayesian methods that let a target region borrow strength from many separate sources, combining their evidence without ever pooling individual records, to estimate its cause of death breakdown from verbal autopsy data.
Ensuring the next generation of statistical talent is rigorously correct. I encourage undergraduates to take apart and dissect published statistical methods through careful reproducibility practices.