- This event has passed.
Shrinkage Estimation for Causal Inference and Experimental Design (Evan T. R. Rosenman)
February 7 @ 4:15 pm - 5:30 pm
Title: Shrinkage Estimation for Causal Inference and Experimental Design
Speaker: Evan T. R. Rosenman, Assistant Professor of Statistics, Claremont McKenna College
Abstract: Passive collection of observational data — in settings such as medicine, insurance, and e-commerce — is a ubiquitous feature of modern life. For statisticians, these ever-proliferating datasets are both promising and perilous. Observational data often contain rich information about the causal effects of novel treatments, such as a new vaccine or drug regimen. Yet, because assignment to treatment is not randomized within these data, one can never guarantee that treated and untreated units are comparable. Consequently, causal effects derived from observational studies often suffer from bias. The applied literature contains myriad examples of treatments that seemed promising in observational data, only to be overturned by later, higher-quality studies.
Evan Rosenman is an Assistant Professor of Statistics in the Claremont McKenna Department of Mathematical Sciences. His research focuses primarily on problems in data science and causal inference, with applications to political science and public health. He is particularly intrigued by problems involving hybridizing observational and experimental data to better estimate causal effects, and by applications in modern electioneering, such as ecological inference and prediction calibration. He earned his PhD in Statistics from Stanford University and completed a postdoctoral fellowship at the Harvard Data Science Initiative.
