Hi Rick,
Thanks again! The info on TENTzero was very helpful!
I had run the analysis using TENT(12.5,42.5,7), which looks very good. Another option I considered was TENT(10,45,8) but I don't think I'm really gaining anything (probably rather the opposite).
Things I may have not appropriately considered:
- There is not much of "rest" left
- There are 6 "stimuli" (i.e. 6 bh blocks), each "stimulus" modeled with 7 tents bc I thought sampling every other TR would be sensible with respect to DOF etc
Scaling is done on a voxel-wise basis (in those of significant checkerboard-related activity):
scaled checkerboard beta = checkerboard beta / (1+ bh beta / avg bh beta across all voxels in the mask)
I had planned on using an average (across all tents) of the reml or iresp betas to be used as the voxel-wise bh beta in the scaling function though your post suggests that this may not be the way to go? If I understand you correctly, then averaging across betas 2-4 or 1-4 (out of 0-6) from the reml output would be reasonable?
Does it make sense to consider the iresp in favor of the reml betas to be used in the scaling operation?
Thank you!
Nic
-------------------
The task consists of 6 bh blocks each 15 sec and 6 rest blocks each 30 sec (I said 27 in an older post). The bh blocks have a 3 sec countdown before them.