 |
 |
Dear Laura, > I have a study with 15 controls. One of the tasks is very simple, just hand movement and rest. > At the first level analysis I do a positive t contrast A = [1 0] (the 0 is for the constant) and a > negative t contrast -A = [-1 0]. > > At the second level analysis I take the con* images, smooth them again, and do a one-sample > t-test for all the A con* images and another one-sample t-test for all -A con* images. > > If I do an F contrast to see the results obtained with the A con* images I find that this is exactly > the same as an F contrast for the t-test with the -A con* images. > > I am a lot puzzled because I thought that once one had separated A and -A at the first level, > the separation would have been kept at the second level. yes,. that's correct. The t-test is only showing you postive effects, i.e. entering a contrast [1 0] you see those areas, which are positive in the ON-condition, entering [-1 0] will show you those areas more active in the OFF condition. The latter one are those areas, having a negative sign in the [1 0] contrast, which are therefore not displayed, but they are still computed. Or, with other words, everything which is negative in the [1 0] contrast will be positive in the [-1 0] contrast. So, everything is computed for the con images, i.e. also the negative values. The sign of the effects are not separated in the con-images, as you assumed. Therefore, it is just enough to put only the con images from the [1 0] contrast into the one-sample t-test. There, the [1] contrast will show you the positve effects the [-1] contrast will show you the negative effects. Since the F-test will show you the overall effect of your design, irrespective of the sign, both designs, that with the [1 0] con-images and that with the [-1 0] con-images are absolutly identical. Does that help? Karsten -- ------------------------------------------------------------- Karsten Specht, PhD Department of Biological and Medical Psychology & National Competence Centre for functional MRI University of Bergen Jonas Lies vei 91 5009 Bergen Norway Tel.: +47-555-86279 Fax: +47-555-89872 [log in to unmask] http://fmri.uib.no/