On Use of Multiple Tests of Significance in Psychological Research