Optimal sample sizes for comparison of proportions using FBST

The estimation of sample size is a subject of great interest to statistical theory, as a way to study the feasibility of experiments. This procedure is called preposterior analysis in Bayesian theory, since it is usually done before collecting data and aims to minimize sampling cost and the decision loss [1]. Our proposal is to develop a method for determining the optimal sample size, n*, which minimizes the loss function related to the problem of testing sharp hypotheses [2]. In this work we present a method to test the sharp hypotheses of equality of two proportions using the FBST - Full Bayesian Significance Test.