Abstract
Basic principles of analysis of variance (ANOVA) and its extensions to nonexperimental areas such as test theory and factor analysis were briefly reviewed. Then the distinct feature of experimental use of ANOVA, namely the sole dependence on significance testing, was pointed out together with its methodological limitations. The remaining part of the article was devoted to discussion on whether and how recent statistical reform, which advocates the use of effect size, confidence interval and power analysis, can overcome these limitations.
