Hi,
> I am having a bit of a problem trying to figure out whether
> or not to use the -D option in randomise. I have found
> several emails regarding this issue (06/17/2010, 10/30/2010,
> 02/04/2008), but I am still confused. Currently, I have 3
> nuisance variables (age, gender, and SES) and one variable
> of interest (RT). In this design I enter RAW values for ALL
> variables and include the -D option when running randomise.
> My contrasts looks like this (0 0 0 0 1) and (0 0 0 0 -1).
> Is this correct?
Not quite, please see below...
> randomise -i all_FA_skeletonised -o tbss -m
> mean_FA_skeleton_mask -d design.mat -t design.con -D -n 5000
> --T2 -V -x
>
> Additionally, if I want to see if there is an interaction
> between 2 variables of interest and I have already demeaned
> them and created an interaction term for each subject do I
> need to use the -D option when running randomise? Example, 3
> nuisance variables (age, gender, and SES) and 1 interaction
> term (demeaned and multiplied). Contrasts (0 0 0 0 1) and (0
> 0 0 0 -1). Is this correct?
>
> Does -D demean each EV or does it depend on my contrasts?
> In other words, does it only demean the 1's in my contrast?
-D demeans your data, not your EVs. That's why, if the mean is not modelled in your design.mat, you need to use the -D option. In your example, you can demean each of your EVs and your randomise command should be fine.
Cheers,
Gwenaelle
--------------------------------------------------------------------
Gwenaëlle Douaud, PhD
FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington OX3 9DU Oxford UK
Tel: +44 (0) 1865 222 523 Fax: +44 (0) 1865 222 717
www.fmrib.ox.ac.uk/~douaud
--------------------------------------------------------------------
|