Hello, If you are only interested in the group difference then demeaning makes no difference here. Also as the group regressors contain the global mean, there is no need to use the -D option in either case. Kind Regards Matthew -------------------------------- Dr Matthew Webster FMRIB Centre John Radcliffe Hospital University of Oxford > On 4 Oct 2017, at 07:56, Manuel Blesa <[log in to unmask]> wrote: > > Hi all, > > I have a similar doubt, I've been reading about when to demean or not and when to use the -D option. But I'm a little bit confused. > > In my case I want to compare two different groups, adjusting for some covariates (some of them are binary), the design matrix looks something like that: > > 1 0 284 0 1 > 1 0 296 1 1 > ..... > 0 1 290 1 0 > 0 1 293 0 0 > > Should I use this or should I demean my covariates? > > 1 0 -9 -0.7 0.4 > 1 0 1 0.3 0.4 > ... > 0 1 -5 0.3 -0.6 > 0 1 -2 -0.7 -0.6 > > From what I read, I think that in this specific case there is no difference to use one of the other, but demeaning is harmless. Am I correct or should I use the demeaned version? I also have to use the -D in the randomise command? Thanks in advance? > > Regards, > > Manuel > ...