Dear Ahmed,
> I am conducting a second level analysis linear regression trying to
> correlate clinical markers of disability with contrast maps in fMRI data.
>
> I have entered my contrast images and the appropriate covariates (8 each)
> as asked for. I entered 'no grand mean scaling' and 'omit global
> normalistion'. I am not sure what to type for threshold masking and type
> omit.
>
Threshold masking is needed when you want to exclude other than GM
voxels, i.e. at a first level. At a second level, it should indeed be
omitted.
> After estimation, the design matrix has two columns, the first has the
> mean corrected covariates and the second is blank. What do I enter to look
> for positive or negative correlations. Is it just [1] and [-1]?
>
Yes. The second column refers to your condition effects, so [0 1]
and [0 -1] should show the regions were this is sign greater/smaller than
zero, with the effects of disability removed.
Good luck
Dick Veltman
.
> Thanks
>
>
> Ahmed Toosy
|