Dear Anderson,
Thank you so much for providing these examples, we truly appreciate it!!
I had two question though:
1) Which of the contrasts you suggest will actually help disambiguate the 3-way interaction? Could you specify those for us please so we can look into them? Many thanx.
We would like to know if the 3-way interaction occurs because of the opposing relations X1 and X2 may have with Y in patients relative to controls, or is due to the fact that only one of the predictors (X1 or X2) is significantly associated with Y while the other is not in patients relative to controls.
2) How do I check the sign of the test statistics?
Using FSLview, do I load in the zstat/tstat image of the main effects, load the thresh_zstat image related to the 3-way interaction on top of this, and simply point the cross on regions that show interaction effects and read the test statistic sign within those regions for the main effects?
Regards,
Mauricio
|