分散分析における誤差項に関する一考察(5)

概要

論文の詳細を見る
In the previous article, the author has considered the appropriate use of error term to test the statistical significance of the main effects, the interaction effects, and the simple effects, after the analysis of variance about 'three-factor' experimental design with repeated measure on one factor was applied. The purpose of the present article is to consider the appropriate use of error term as the denominator of the usual F ratio, and the quasi-F ratio when the usual F ratio can not be applied, to test the main effects and the interaction effects, and the appropriate use of the error term as the form of the simple mean square and the composite mean square to test the simple effects in 'three-factor' experimental design with repeated measures on two factors, for example, which are Factor B and Factor C. In such a 'three-factor' experimental design as described above, there are six cases which are described as follows: Case 1 : Factor A, Factor B, and Factor C are all fixed. Case 2 : Factor A and Factor C are both fixed, but Factor B is random. Case 3 : Factor A is fixed, but Factor B and Factor C are both random. Case 4 : Factor A is random, but Factor B and Factor C are both fixed. Case 5 : Factor A and Factor B are both random, but Factor C is fixed. Case 6 : Factor A, Factor B, and Factor C are all random.
奈良教育大学の論文
1976-12-25