Bayesian inference |

Then, based on the prior density functions and the likelihood functions of the data conditioned on each of the hypotheses, the prior odds ratio is modified to form a posterior odds ratio. In contrast to the classical approach, it is not necessary to accept or reject each hypothesis. If needed, such a decision can be made by minimizing the expected loss from making a wrong decision, using some specified loss function, where the expectations are calculated with respect to the posterior probabilities on each hypothesis.

Reference: Oxford Press Dictonary of Economics, 5th edt. |