. While the most believable alternative hypothesis is not always going to be obvious, Johnson’s (2013) results show that, if we can assume that one-half of the possible alternative hypotheses should yield a positive result, then this suggests ‘‘between 17% and 25% of the marginally significant results are false.’’