Traditional hypothesis testing, which serves as the data analysis tool in psychological experiments, can be appraised according to two criterions, the first one is legality and the second utility. The logic foundation of traditional hypothesis testing, rooted in the frame of frequency school statistics, is, in fact, correct. But, in terms of utility, traditional hypothesis testing involves two shortcomings. One is that its alternative hypothesis can’t be denied, and the other is that it can only provide qualitative conclusion. Confidence interval is suggested to replace traditional hypothesis testing to report the results of psychological experiments, for it can modify and supplement traditional hypothesis testing in these two aspects. Clarification of the mistakes in the using of traditional hypothesis testing also makes the researchers pay much attention to the PSI problem, which would turn the focus of the design and data analysis of psychological experiments from emphasizing population to emphasizing individuals.