Bayesian Method: Powerful, but Not Magical

• Activity Number: 81
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Freda Cooner*+
• Companies: FDA/CDER/OTS/OB
• Keywords: Bayesian ; Priors ; Clinical Studies

Abstract

Bayesian methodology, a once burdensome method, has become a powerful, sophisticated and efficient analytic tool. Bayesian scientists are now able to solve issues that have challenged many frequentists. Clinical investigators have begun to recognize the many benefits of Bayesian methodologies. Hence, it is used not only in analyzing data, but also in clinical trial design for both safety and efficacy. However, researchers often seem oblivious to some of the methodological pitfalls, or, perhaps, they may have some misconceptions regarding what the Bayesian approach can truly provide. It is important to understand the Bayesian framework before applying it to a clinical study and expecting that it will "magically" resolve potential analysis issues. This presentation will focus on the application of this adaptive methodology, starting with a brief discussion of its advantages in the clinical settings. Then we will discuss some important limitations. We note that some advantages attributed to Bayesian methodology are, in fact, neither advantages nor of Bayesian origin. The presentation will conclude with some points to consider for advocates of the Bayesian method.