I am trying to find out if there are any specific rules when setting up contrasts in SPM2 for an fMRI analysis where different conditions have differing number of samples. Suppose I have a de-meaned fMRI data set with three conditions A, B and Rest with unequal number of scans say 3, 2 and 4 respectively. If my design matrix were as shown below (orthogonal, full rank, simple box car) TaskA TaskB Rest 1 0 0 1 0 0 1 0 0 0 0 1 0 0 1 0 0 1 0 0 1 0 1 0 0 1 0 My questions are: Do the regressors need to be normalized in some way (de-meaned/other) considering the fact that the actual dataset (Y) is already de-meaned ? My concern here being the differing number of samples for the conditions. Also, is it important to have the sum of the contrast weighted regressors equal to zero i.e sum(c1*X1 + c2*X2 + c3*X3) = 0 ? In other words is it ok to have t-statistic contrasts such as [1 1 -2], [1 0 -1] and [0 1 -1] in the case of the above design matrix ? I did run an analysis comparing both ways i.e regressors with zero mean and non-zero mean regressors and got results which were very close but am trying to verify whether it is ok to do so. ....Raj ____________________________________________________ Yahoo! Sports Rekindle the Rivalries. Sign up for Fantasy Football http://football.fantasysports.yahoo.com