Fig. 1.

Fictitious example. Two normal posterior distributions following N(0,1) (dashed) and N(0.5, 2²) (continuous) with credible intervals. The probability that treatment A, corresponding to the flat distribution, is better than treatment B, corresponding to the steep distribution, is 59 %