Dear Christopher, > I am running SPM96, using a covariate for global normalization. If I > choose grand mean scaling to 100 and save normalized data, what will > units of response reflect: percentage of the global covariate? The units in this instance will be the same as the original image values (e.g. counts/sec or T2* signal) scaled so that the grand mean is 100. They can be thought of as percentages of the grand mean. > I am assuming that under these circuamstances I can still use SE{Y} for > one group to calculate an effect size for the mean effect in the other > [for a given SPM]- yes? Absolutely. I hope this helps - Karl %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%