Hello Jeanette,
Thank you for the clarification.
My model is the way you described it - perfect. I haven't mentioned it because I thought that the two ways of modelling (cells means and factor effects approach) are equivalent.
As to the interaction, I mean interaction between 2 factors "AB" and "xy". Depending on the value that xy takes, the difference between A and B is specific for the particular value of xy. [1 -1 -1 1] [-1 1 1 -1].
I realize that:
2x2 F-contrasts analysis
2x2 t-contrasts analysis
1x4 F-contrasts analysis
refer to different hypotheses.
However, are they so independent in results?
The data underlying them consist of the same 4 groups. The differences in significance are:
2x2 F-contrasts - not significant even on p=0.999 level (FWE corrected)
1x4 F-contrasts - significant on p=0.01 level FWE corrected.
And if what you're saying was the case (F-values significant on p=0.06 level for two-group contrasts), wouldn't they appear in tfce_corrp image and just didn't survive thresholding?
All the best,
Danusia
|