Hi - this is OK but you do need to model the mean effect as well. Also, we would generally recommend doing a fixed-effects second-level analysis for each subject, and then a third-level mixed-effects to combine across subjects. Cheers. On 12 Oct 2011, at 18:35, Elisabeth Karuza wrote: > Hi all, > I am interested in modeling linear increases/ decreases in activation over the course of several sessions for a given contrast. I would first like to model linear inc/dec within individual subjects and then take the group mean. Is it correct for my second level to look like this: > > EV1 EV2 > Subject1 Subject2 > Input1 (session1) -1.5 0 > Input2(sess 2) -.5 0 > (sess3) .5 0 > (sess4) 1.5 0 > (sess1) 0 -1.5 > (sess2) 0 -.5 > (sess3) 0 .5 > (sess4) 0 1.5 > > Contrasts EV1 EV2 > Sub1 1 0 > Sub1 -1 0 > Sub2 0 1 > Sub2 0 -1 > > Do I need to orthogonalize the linear trend coefficients w.r.t. “multi-session mean” for each subject or is this unnecessary because the coefficients are already demeaned? > > The third level would just be a standard group analysis combined across subjects. > > Thank you so much, > Lizz > --------------------------------------------------------------------------- Stephen M. Smith, Professor of Biomedical Engineering Associate Director, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve ---------------------------------------------------------------------------