Research summary
With the rise of rich, large-scale datasets, there is growing interest in analyzing decision-making heterogeneity—the variety and diversity found within datasets—to inform welfare-improving policies. While empirical Bayes methods are effective for evaluating such heterogeneity, current approaches are generally limited to simpler statistical models. (Empirical Bayes methods estimate prior distributions from the data itself rather than relying on predefined or subjective priors.) Dr. Jiaying Gu, Canada Research Chair in Statistical Decision Making, is developing advanced empirical Bayes techniques.
She and her research team are designing a method to evaluate multidimensional decision heterogeneity and applying it to assess patent evaluations (which determine the economic worth of patents). Gu’s team is also evaluating the risk performance of traditional empirical Bayes methods in adaptively generated data and proposing improvements for more complex experimental settings. Their research will deepen our understanding of individual decision-making, with applications in innovation, health care and beyond.