 |
 |
Hi, 0- Don't use the grp column. In randomise, it sets the exchangeability blocks. See Jeanette Mumford's recent post on this. 1- I suspect the model should be the same in SPM or FSL, as both programs use the GLM. It sounds like you need four columns: group1, group2, confoundev1, confoundev2. You should demean the confoundevs. 2- Yes, the DOF will be N-p, where N is the number of subjects and p is the number of columns. 3- I don't think so, but I'm not sure... 4- No. Cheers, David On Dec 5, 2012, at 1:24 PM, Saeed Masouleh wrote: > Dear Fsl users, > I am a new FSL user. I am very much confused in making Design matrix for the dualregression procedure. > > I am using GLm gui to make the design.mat, .con. > > I have two groups and 2 nuisance covariates and one covariate of interest. I have changed the column GRP (manually in to 1 & 2s) and have assigned two Evs for groups . > Now, with the described set-up, if I don't expect any interaction between the groups and covariates and don't want to look at each covariate separately in different groups, could I just have one column for each co-variable? > I did so, and got a warning which meant that I had to define each covariate in a seperate column with regard to the group (2 columns for each covariate). > > 0- Is there any better way than manually changing GRP column in to 1 and 2s? > > > 1- I am wondering if it is a "must" in FSL desings to have different columns (EVs) for each group for one covariate? (I know in SPM it is related your research question but the FSL- GLm tutorial, made me to doubt if it could be the same in FSL as well) > > 2- Does it affect the DOF? > > 3- is there any other way to make these desing files and the design.fsf files? (apart from using FEAT or Glm.gui) > > 4- Is there any way to use randomise for question 3? > > > thanks a lot in advance , > M. Saeed > > > > -------------------------------------------- David V. Smith, Ph.D. Postdoctoral Fellow, Delgado Lab Department of Psychology Rutgers University Newark, NJ 07102 --------------------------------------------