Figure 5. Illustration of the different hypotheses under investigation [18]. (A) The classic two-sided significance testing aims to reject a point null hy-pothesis (here an effect size of exactly zero). (B) In equivalence test, the H0 of no equivalence is tested (grey region), so the white area is the rejection region. (C) For the Bayesian estimation approach, the 95% highest density interval of the posterior is compared against the Region of Practical Equiv-alence (ROPE) between ∆L and ∆U . (D) For the Bayes factor, two models are compared that differ in their prior distributions: TheM0 prior is a point mass of 1 at an effect size of 0, the alternative model M1 is here plotted as a Normal distribution as an example. Note, that other alternative models can be used, e.g. centered on a value derived from theory or previous studies (see Figure 4).
