Hi AFNI experts,
In preparing a manuscript for a recent study, in which I used surface-based analyses...I have run into a question about how to estimate spatial smoothing to determine cluster correction. In volumetric data, I have read that it is preferable to calculate cluster correction using the calculated smoothness of individual-level residuals and then average that smoothness across participants. This is opposed to simply using the smoothness level that you enforced on the data.
Does the same rule apply with surface-based data? I used a target FWHM of 4mm during my preprocessing. I then ran individual level GLMs using 3dDeconvolve and a group analysis using 3dttest++. Then to determine my cluster threshold, given my uncorrected and corrected p-values of interest, I ran slow_surf_clustsim.py. It sets the blur to 4mm by default. Is there a reason this smoothing value should be different than the original target FWHM?
Also, I did go back and calculate the smoothness of the individual-level residuals (errts data) using 3dFWHMx for each participant and the average is essentially 4mm, if anything its lower (3.9mm).
Can someone please let me know if I am missing something? Or perhaps how best to justify my choices to reviewers?
Thanks in advance,
Kara