Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 4 - Recent Advancements in Prior Elicitation and Computational Tools for Bayesian Design and Analysis
Type: Invited
Date/Time: Sunday, August 7, 2022 : 2:00 PM to 3:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #320335
Title: Practical Considerations in Power Prior: Interpretation and Software Advancements
Author(s): Fang Chen* and Frank G. Liu
Companies: SAS Institute Inc. and Merck & Co., Inc
Keywords: power prior; Bayesian; MCMC; software; SAS
Abstract:

The power prior provides a dynamic approach to translating data information to distributional information about the model parameters. Since its introduction, the power prior is playing an increasingly prominent role in many disciplines. Yet, a persistent question remains: how to best interpret the specified power weight value a0? This talk addresses the interpretation issue and links the value of the weight to the amount of information contained in the historical data set. The procedure can be extended to variations of the power prior, such as the partial borrowing power prior and the power prior for multiple data sets. The talk will also discuss software advancement in power prior analyses, which often relies on clever programming solutions that are problem-specific and difficult to generalize. We illustrate new features in SAS' BGLIMM procedure that enable you to fit the power prior to many models (hierarchical generalized linear models, repeated measurement models, missing data problems, etc.) with the simplest setup. We also discuss model selection and choosing a0 in single and multiple historical data sets settings.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program