Hi Rick,
Thanks for your clarification.
I am using an AM2 method to examine word characteristics affect signals. I am currently using two ABI values when I do the AM2 analysis.
The problem is that if an effect of a regressor A (the first ABI value) has positive activation in a certain region, an effect of a regressor B (the second ABI value) has negative activation in the same (or very close to the) region. Of course, two ABIs are not correlated each other. Is it a real effect? or could be an error of the AM2 method when two or more ABI values are included?
A related question is what the negative activation means? Is it like a negative slope in a regression analysis?
Thanks,
Cerouno