Processing Stream for 3dFWHMx and 3dClustSIm

I was looking at the suggested processing stream for using 3dFWHMx and 3dClustSim and just wanted to check and see if I’ve got it straight.

  1. Using -errts in 3dDeconvolve (for instance) get residuals for each subject.
  2. Run 3dFWHMx on EACH TR in the errts file for each subject, excluding trails that are censored.
  3. Append the 3 output parameters (a,b,c) to a .1D file for each subject.
  4. Get the mean of these 3 parameters in the .1D file for each subject.

So: at this point, I will have 3 parameters for each subject. Now - do I average these across subjects to get a final a,b,c to supply as arguments in the -acf parameter in 3dClustSim?

thanks, mb

OK - Reading the Cox, Reynolds, & Taylor draft - in Figure 1, it seems to say the MEDIAN a,b,c values across subjects was used??