Dear AFNI minds,
I have been interested in comparing task activation between human subjects at different life stages, where I can compare activation between discretely-defined groups using 3dttest or 3dANOVA.
However, there is some literature out there that dichotomizing a continuous variable into discrete categories compromises statistical power.
What I would like to do is find a way where I can model the combined data of all my subjects to identifiy voxels where:
1) There is an interaction between subject age and the event-related beta correlation of "A"
2) There is an interaction between subject age and the event-related beta correlation of "B"
3) There is an interaction between subject age and the beta correlation of the linear contrast between events "A and B"-- which is generally how "activation" is determined in the behavioral paradigm.
For now, perhaps just assuming a linear relationship between age and activation, though being able to plug in some other hypothesized age function would be desirable.
To be sure, I could dump out hemodynamic responses in VOI, calculate peak signal change, and export into some standard statistical package, but I would rather model age into the statistical map itself.
What is the most straightforward way to accomplish this? Is there a tutorial somewhere?
I see several instances in the literature where data are modeled with some other non-timing-related individual difference variable, such as behavioral performance.
Jim Bjork