webinar register page

Webinar banner
Fan Li - Propensity score in Bayesian causal inference: why, why not and how?
A major debate in Bayesian causal inference is the role of propensity score. On one hand, under the standard ignorability assumption, the propensity score drops out of the likelihood and thus does not factor into Bayesian inference of causal effects. On the other hand, numerous empirical evidence suggests the importance of including propensity scores in outcome modeling (Bayesian or not). We weigh in on this debate. We argue for the benefit of separating design and analysis, the former of which critically depends on propensity score. We re-examine and expand the concept of double robustness. We also discuss a few specific examples of the popular Bayesian nonparametric models for causal inference, including BART, Gaussian Process, and Dirichlet Process.

Mar 3, 2022 05:00 PM in Paris

Webinar logo
* Required information


Fan Li
Professor @Departments of Statistical Science, and Biostatistics and Bioinformatics at Duke University
Fan Li is a professor in the Departments of Statistical Science, and Biostatistics and Bioinformatics at Duke University. Her primary research interest is statistical methods for causal inference, with applications to health and social sciences. She has developed the overlap weighting method. She also works on Bayesian analysis and missing data. As everyone else nowadays, she is interested in the interface of causal inference and machine learning. She is an associate editor of Journal of the American Statistical Association, Bayesian Analysis, and Observational Studies.