ABSTRACT
In contrast to the frequentist approach (sometimes called classical, Neyman–Pearson, or non-Bayesian), Bayesian inference is inherently conditional on a specific observed sample rather than the result of averaging over the entire sample space. In this sense some consider it as having a more “natural” interpretation and as more “conceptually appealing.” The main criticism with respect to Bayesians, however, is due to the use of a "prior", which is the core of the approach. Another approach for choosing a suitable prior is through so-called noninformative or objective priors. Bayesian approach to interval estimation is conceptually straightforward and simple. A posterior distribution of the unknown parameter of interest essentially contains all the information available on it from the data and prior knowledge. Similarly to interval estimation, Bayesian hypothesis testing is conceptually quite straightforward. The ideas for testing simple hypotheses can be straightforwardly extended to composite hypotheses.
