In AlphaSim, I can find a balance between cluster size and individual voxel p thresholds to obtain an overall alpha value. A flag in the program (-fwhm) helps account for gaussian spatial blurring in this determination.
My question is this: at what level of data processing is spatial blurring accounted for? Does it consider blurring that occurs on the statistics, or on the 3d+time data before analysis? My gut tells me it's on the statistics obtained after data analysis, but I'd appreciate a less visceral source.
Thanks,
Eric