s********9
发帖数: 74 | 1 Whne use ANOVA to test five groups's mean difference, the model shows
significant difference. While when use multiple comparasion method to
checked each group pairs, there is no difference between any two groups.
What is the problem? | s********9
发帖数: 74 | 2 Is the question stupid?
No one answer?!
Is the question not clear? I can not type in Chinese....
Anyone has the answer?
【在 s********9 的大作中提到】
: Whne use ANOVA to test five groups's mean difference, the model shows
: significant difference. While when use multiple comparasion method to
: checked each group pairs, there is no difference between any two groups.
: What is the problem?
| g**r
发帖数: 425 | 3 No problem, this can happen, can even happen just within ANOVA without MC
adjust
A simple example: three treatment, A, B, C. Model:
Y=A B
Global testing: A=B=C, reject
testing: 1. A=C failed to reject; 2. B=C failed to reject
The global testing is a much more powerful test, but the detailed pairwise
testing, even without splitting alpha or using family-wise alpha, is less
powerful. | s********9
发帖数: 74 | |
|
