Main effect without significant post-hocs?
In a factorial ANOVA is it possible to get a main effect for a particular factor but when post-hocs are run no significant pairwise differences shows up? If it is, what do you do then in terms of reporting it?
posted by Nathaniel Fernandez @ 10:37 AM
1 Comments:
At 5:43 PM,
Mari Clements said…
We talked about this in class, but it was in the context of interaction effects. The same reasoning, however, holds for main effects.
The F tells you that there is a systematic difference when you look at all the groups together. The post-hocs are examining individual pairs of means, and tells you if the members of that particular pair differs significantly from each other.
It is possible for there to be an overall effect--whether it be interaction or main effect--that does not appear in pairwise comparisons. In such a case, you would note the overall effect, including all required inferential statistics, and then you would note that post-hoc comparisons using the XYZ procedure did not reach significance.
Post a Comment
<< Home