分散分析における単純効果の検定について

概要

論文の詳細を見る
The analysis of variance technique was applied, and it was found that the interaction term between two factors was significant. Thereafter, the tests on the simple effects for each of the factors must be made. The appropriate use of the error term is different in four cases. It is as follows: Case 1: p×q or p×q×r factorial experiment with no repeated measures. Case 2: p×q factorial experiment with repeated measures on one factor. Case 3: p×q×r factorial experiment with repeated measures on last two factors. Case 4: p×q×r factorial experiment with repeated measures on the last factor. As for the Case 1, the usual error term which has been resulted from the analysis of variance is used. As for the Case 2, the error(w) term is used for the tests of the simple effects on the repeated factor, and the composite error term which is consisted of error(b) and error(w) is used for the tests on non-repeated factor. As for the Case 3, the error(w1) or the error(w2) term is used; or the one of composite error terms which are consisted of error(b) and error(w1); error(b) and error(w2); error (w1) and error(w3); and error(w2) and error(w3) is used. Or the composite error term which is consisted of error(b), error(w1), error(w2), and error(w3) is used. As for the Case 4, the error(b) or the error(w) term is used; or the composite error term which is consisted of error(b) and error(w) is used.
奈良学芸大学の論文