Why do you use an effect size of d=0.40 as a cut-off point and basically ignore effect sizes lower than 0.40?