 |
 |
Dear SPMers, I am seeking some clarification on how to set up an ANOVA within subjects analysis of fMRI data with a covariate in SPM2. I have used ANCOVA and am not sure whether I am using it appropriately as I have 4-within-subjects (fMRI) conditions as opposed to 4 independent groups. When running ANCOVA, one is not able to enter number of conditions, rather number of groups is asked for and if I indicate that I have four groups, I must enter my covariate four times. (Note that I am selecting correction for non-sphericity). Although ANCOVA completes without error I am concerned that this may not be a valid analysis. Similar concerns were raised by someone else on the list, though this was with respects to VBM, not fMRI: http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0502 <http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0502&L=spm&P=R27702&I=-1> &L=spm&P=R27702&I=-1 Is there are rationale for why the fMRI models allow for a within-subjects ANOVA but not a within-subjects ANCOVA or is this something that will be updated in future versions of SPM? Although this type of analysis may be achieved using PET/SPECT models using a "multi-subjects: conditions and covariates" selection, is it valid to be running a 2nd level fMRI analysis under the PET/SPECT options. (Or is this again something that will be updated for fMRI models in SPM5?) I am aware that I could also run this as multi-session analysis but cringe at the thought of re-doing pre-processing (particularly if there is another option). Thanks in advance, Andrew