Abstract: A deep exploration of Analysis of Variance by Items (ANOVA by items) had been made in this paper, which is very popular in recent years both at home and abroad. The author first proposed that items in an experiment could also be seen as experiment units like subjects. Therefore, two kinds of basic experimental designs could be distinguished theoretically: Between-items design and Within-items design. Accordingly with ANOVA by subjects, ANOVA by items can also be used to the same experiment for data processing based on its experimental design by item. But at the end of this paper the author indicated that it is not always that ANOVA by items could be used for data processing because items in some experiments are constant rather than random variable. Implications of using ANOVA by items for data processing was also discussed.

Key words: experiment units, between-items design and within-items design, ANOVA by items