With Jacob V. Spertus, PhD, Postdoctoral Researcher, UC Berkeley Statistics, Children's Hospital of Philadelphia
Abstract: Recent developments in permutation tests and E-processes support trustworthy inference by controlling Type I error without asymptotic arguments, parametric models, or pre-fixed sample sizes. These tools extend to stratified designs by maximizing P-values over nuisance parameters – representing unspecified within-stratum means – and are often much more efficient than competing approaches. They can be used to draw rigorous inferences about average treatment effects in clinical trials and extended to other important statistical tasks, like online change-point monitoring and treatment policy optimization.
The Department of Epidemiology & Biostatistics welcomes all participants to our events. If you need a reasonable accommodation to participate in this event because of a disability, please contact Liz Buggs ([email protected]).