Standard hypothesis testing methods are nor robust when differences exist. The practical result is that in applied work, many nonsignificant results would have been significant if a statistical method developed after 1960 had been used