Figure 4

Dependence of the posterior probability on study power. Posterior probability of the null hypothesis plotted against the (pre-study) power at the 5% significance level for P = 0.05, 0.01, and 0.001 and a prior probability of q = 90%. The calculation is based on a normal prior with standard deviation τ = 1 (left plot) and τ = 2 (right plot) under the alternative, assuming that one unit corresponds to the minimum clinically important difference. The dashed lines indicate the minimum posterior probability as obtained from the BS (short dashed) and SBB (long dashed) approach, respectively.