 |
 |
Hi, The answer to this very much depends on what you are doing (e.g. first-level FMRI, second-level FMRI, VBM, etc.) and what tool you are using (FEAT, randomise). It is true that for first-level FMRI you should not remove the mean from the data before passing it into FEAT. At the first level you cannot sensibly get anything meaningful from first-level mean values and so they are automatically removed inside the statistics within FSL (but it still needs them in the input to find the brain, etc, as you've found). It is at second level (or higher) that it is the case that you want to model the mean (this is often the most interesting thing). So what I've been talking about previously has very much been focused on second level issues, either within FEAT or randomise. I hope this helps. All the best, Mark On 1 Sep 2012, at 12:29, Ib <[log in to unmask]> wrote: > Hi again, > > "If you do not have regressors that model the mean in the design matrix then it is important to demean all regressors and the data" > > I learned recently that for instance in SPM (and also FSL) the mean shouldn't be removed from the data, or if I detrend, I should simply add the mean back into the data (since SPM thresholds the data at 80% of the maximum voxel, mean removal makes the software say that there are no voxels with data). > > Is this correct? > at least since I added the mean back to the data, my analysis was giving many errors. > > Ib > > ----------------------- > > Hi, > > Scaling of regressors (multiplying by some constant) in the GLM is unimportant for the statistical values such as t-stats or z-stats (they are unaffected by this scaling) but the actual parameter estimates (PEs or COPEs) would change. > > If you have a set of regressors that model the mean (either a single regressor or several that add together to give a column of all ones) then it is not necessary to demean any _other_ regressors in the model (over and above the ones used to model the mean). However, this is true only for statistics associated with contrasts that do not include any of the regressors that model the mean (the entries for these regressors are zeros in the contrast vector). > > If you do not have regressors that model the mean in the design matrix then it is important to demean all regressors and the data (as otherwise the mean affects the parameter estimates, the contrasts and the statistics). > > I hope this makes things clear. > > All the best, > Mark >