As widely used as p-value in basic science and clinical trials, it is also greatly criticized. Alongside with p-value is Bayesian posterior probability. One idea that is widely spread is that p-value could be very close to Bayesian posterior probability for a one-sided hypothesis test, while those two are dramatically different for a two-sided test. In this talk, we argue that it is the choice of prior distribution other than the set up of a hypothesis testing that impacts the reconcilability or lack of it between p-value and posterior probability. The practical meaning is to emphasizes the importance of careful evaluation and choice of the prior distribution.
   
Presenter: Dr. Xiting Cindy Yang, PhD
   
Dr. Yang is a Statistical Subject Matter Expert at Biomedical Advanced Research & Development Authority (BARDA) under U.S. Department of Health and Human Services (HHS). She has about 20 years clinical trial experience, in both regulatory and industry settings, including 15 years in the FDA. Her statistical expertise includes both frequentist and Bayesian methods. Other interests include adaptive designs and propensity score analysis. She is Associate Editor of for Biostatistics & Epidemiology and 2018 Chair of Medical Devices and Diagnostics (MDD) Section of American Statistical Association (ASA). She has a Ph.D. on statistics from Carnegie Mellon University.