Hello Anderson,
Thank you for the helpful response!
Sorry for the confusion, but to clarify: Instead of using a block model in my feat analyses, I am modeling a linear signal increase - ie across a given block, is there activation that resembles an upward slope (increase in activation over time). This model mimics the idea of learning over time, as the participants are more exposed to our task.
For example, in a one column format, a block design will have 1s for each TR across each block. For my linear signal increase, I've modeled a block running from 1 to 47 for each TR in the block (47 TR of 3 sec each, 141 sec total).
So back to my earlier question, would you say that regardless of model design for a group analysis, if two or more scanners were used in the study, it would be wise to include a single binary regressor to control for scanners?
I am a bit worried about this because after checking for scanner variability in my block design, there is little variability (good!). But suddenly, when checking for scanner variability in my linear signal increase analyses, there is SO much activation (bad?) within conditions.
Do you have any thoughts on why this might occur?
Thank you again for your help.
Best,
Kevin Japardi
|