I calculated it from the residual fMRI time series. The fMRI data were analyzed in AFNI. In AFNI, you can get the residual time series during 3dDeconvolve by adding the option: -errts. The effective smoothness for x, y and z are then computed from the residual time series using 3dFWHMx. Then it’s just a matter of averaging the effective smoothness FWHM numbers across x, y, z directions, and then across subjects. Any ideas? Thanks. brian Brian T. Gold, Ph.D. Associate Professor Anatomy and Neurobiology MN 214 Medical Sciences Center, University of Kentucky, 800 Rose Street, Lexington, KY, 40536-0298 email: [log in to unmask] office: 859-323-4813 fax: 859-257-6700 From: [log in to unmask] [mailto:[log in to unmask]] On Behalf Of Michael T Rubens Sent: Monday, August 16, 2010 3:15 PM To: Gold, Brian T Cc: [log in to unmask] Subject: Re: [SPM] effective smoothness How did you calculate effective smoothness? -Michael On Mon, Aug 16, 2010 at 11:22 AM, Brian Gold <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hello, I am conducting a combined fMRI-VBM study. I want to compare the relative spatial extents of between-group differences in fMRI and VBM data sets. I thus want the effective smoothness of the functional and structural data sets to be as close as possible. I have computed the effective smoothness of my fMRI data to be 7.6 mm FWHM (averaged across x,y,z directions). What would be the simplest way in SPM 5 that I can determine the size of the FWHM smoothing kernel that, when applied to my VBM data, will result in the same effective smoothness as my fMRI data? Thanks in advance, Brian -- Research Associate Gazzaley Lab Department of Neurology University of California, San Francisco