It is important to emphasize the major theoretical difference between the frequentist and the Bayesian approaches to hypothesis testing. In the frequentist approach, the decision to reject the null hypothesis of no association is typically based on controlling the probability of the type I error and this is done by imposing a threshold on the P-value. This procedure does not assess per se whether the null hypothesis is true or false. In the Bayesian approach, the decision to reject the null hypothesis is based directly on the probability that the null hypothesis is false, given the evidence provided by the data. Only Bayesian procedures allow for an explicit assessment of the likelihood of parameters and hypotheses [72].
