A note on the alpha-level: Not all alphas can be justified

In the past years, discussions about the ‘alpha-level’ in null hypothesis significance testing gained attention in psychological literature. Much has already been said in this debate. One point has, however, not been raised in the recent discussions: that sometimes it is illogical to fix alpha to a specific level, whether this is 0.005, 0.05 or some other level. In this note, I raise this point and offer recommendations.