Figure 1. Different models for hypothesis testing
The different models for testing a theory predicting a difference via a smallest effect of interest (A and B), using inference by intervals (e.g. equivalence testing); or a rough scale of effect (C and D), using Bayes factors. The scale of effect, S, implies a rough range of plausible effects of -2S to +2S in C, and of 0 to +2S in D. The equivalence region in A is typically symmetric though need not be. Note for every test two models are specified, one for H1 and one for H0 (in the diagram, an interval H0 for inference by intervals and a point H0 for the Bayes factors).
