 |
 |
Dear Paola, It could be that in voxel v1 group 1 has a positive effect, but groups 2 and 3 have a small negative effect. The average effect could therefore be zero in which case [1 1 1] will show nothing at this voxel. The null hypothesis here is that the average is zero. Whereas [1 0 0; 0 1 0; 0 0 1] looks for voxels where any of the effects are significantly non-zero (ie where any of the multiple null hypotheses, each tested by a row of the contrast vector, can be rejected). This is more likely to pick up voxel v1. But voxel v2 which has small positive effects for all conditions will be detected by both contrasts. Best wishes, Will. From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Paola Valsasina Sent: 04 June 2014 09:48 To: [log in to unmask] Subject: [SPM] full factorial / average effect of condition Dear all, I have a question about testing average effects of condition in full factorial models. Let's say I have three groups in a full factorial model. An F contrast called "average effect of condition" will be automatically created by SPM with these weights: [1 1 1]; it displays any effect (positive or negative) in the three groups. If I build a F contrast in this way: [1 0 0; 0 1 0; 0 0 1], it seems to me that the average effects of condition are also displayed, because the pattern of activity is quite similar to the previous one, although not identical (my impression is that this latter contrast is more noisy). Which is the conceptual difference between these two contrasts? Thank you in advance! Kind regards Paola [cid:image001.jpg@01CF7FDB.2675E660] <http://www.5xmille.org/> Disclaimer added by CodeTwo Exchange Rules 2007 www.codetwo.com<http://www.codetwo.com>