Figure 14.
Distribution of p-values under three different reporting strategies based on the p-hacking strategy ‘selective reporting of the dependent variable’: reporting the first significant, smallest significant or smallest p-value out of a batch of tests. Left side shows full p-value distribution, right side shows p-curves [34]. One dependent variable indicates no p-hacking, p-hacking aggressiveness increases with the number of dependent variables. Simulation with N = 100, ρDV = 0.3.