Some properties on tests based on the Bayesian confidence interval

Abstract

In testing statistical hypotheses, quite generally, if we admit the result of Neyman-Pearson (apart from the interpretation of them) in case that we specify $n$ in advance and admit the likelihood principle, the stopping rule that “continue the experiments until rejecting the null hypothesis” is closed. As a matter of fact, a stronger phenomenon happens, and we shall show it with some examples.