Dear Anderson,
Thank you so much for the time and effort, I really appreciate it!
I have attached screenshots of the EVs and Contrast matrices that were build according to your recommendations. Could you be so kind and have a look at these, just to be sure we are on the right track. EV 8 (Q1*Q2*G) is obviously most relevant to our research question. All EVs are mean centred, except for intercept and G (EVs 1 and 2). Is this correct?
Further, we are pondering about the following:
a) Why should one also build terms that are not of primary interest (e.g., Q1*Q2; Q1*G; Q2*G)? Relatedly, if one nonetheless wants to assess how these terms might relate to connectivity measures, what contrasts should be build?
b) Why is important to have an intercept in this model?
c) For a better appreciation of a significant Q1*Q2*G interaction term would you suggest plotting the results? More specifically, what would a significant positive or negative Q1*Q2*G interaction term actually suggest?
d) Would it be valid to test the interaction terms (i.e., Q1*Q2) in each group separately (i.e., separate group analysis) and compare the effects?
Our apologies for all these questions, we really hope you could help us out!
Regards,
Mauro
|