Dear Dan,
Each *individual* (i.e. subject) gets multiplied by a different value (depending
on the mean value across all spatial voxels in the mask and all timepoints).
So difference between subjects are accounted for, but no differences between
different voxels or different timepoints are changed (except by consistent
scaling). That is, the variations over space and time within an individual (a
subject) are all preserved, although there is a single scaling factor applied
across all of these (therefore the absolute values of the differences may
change, but the relative changes are all preserved). It is only the gross
differences between subjects that are changed by this scaling.
I hope this clarifies things.
All the best,
Mark
On 28 Nov 2011, at 11:24, Dan Shaw wrote:
> OK - so I had mistaken within- for between-timeseries variation. Thanks for clearing that up for me. Unfortunately, however, I'm still confused; but if I don't get it after this question then I shall leave you alone...
>
> If I multiple all individuals' timeseries by a single value, aren't I just re-scaling the inter-individual differences? If I want to normalise my data, shouldn't I multiple each individual's time-series by a timeseries-specific value. Only then would all subjects have the same grand mean, right?
>
> Once again, I appreciate your efforts on this one.
>
> Dan.
>
|