ABSTRACT
This chapter summarizes the current Bayesian sample size calculations into two categories. One category considers making use of the Bayesian framework to reflect the investigator's belief regarding the uncertainty of the true parameters, while the traditional frequentist testing procedure is still used for analyzing the data. The other category considers determining the required sample size when a Bayesian testing procedure is used. Three different criteria are proposed for sample size estimation. They are the average coverage criterion, average length criterion, and worst outcome criterion. Lee and Zelen proposed a procedure for sample size calculation based on the concept for achieving a desired posterior error probability. The chapter shows how Lee and Zelen's Bayesian approach can be used to determine the sample size for comparing means. To evaluate the finite sample performance of the bootstrap procedure, a simple simulation study is conducted.
