Advances and Applications in Statistics
Volume 9, Issue 2, Pages 247 - 260
(August 2008)
|
|
DETECTING AN UNCONDITIONALLY IDENTIFIABLE PATTERN IN TWO-WAY ANOVA
Wei Ning (U. S. A.)
|
Abstract: In the study of non-zero interactions in Two-way ANOVA model, the traditional F-test has been a most frequently used tool. However, as Neyman [3], Yates [7] and Wade [6] pointed out, it is more difficult to detect interaction effects than main effects due to relatively large standard errors of the interaction estimate, especially when the number of repetitions is small. Daniel [1] and Terbeck and Davies [5] proposed tests based on a recognizable pattern, which they defined as an unconditionally identifiable pattern (UIP). They showed their tests are more powerful than the traditional F-test under such a certain pattern. Daniel [1], suggested a procedure to identify an unconditionally identifiable pattern in unreplicated two-way data, however, his procedure is visual, and it may make some inappropriate conclusions. In this paper, we propose a new procedure to detect the existence of an unconditionally identifiable pattern statistically in replicated two-way data. |
Keywords and phrases: two-way ANOVA model, unconditionally identifiable pattern, replicated. |
|
Number of Downloads: 350 | Number of Views: 1323 |
|