Hi Anastasia,
> if I would like to go a regression analysis,
> where I would
> like to correlate GM changes with some behavioral variable
> in this case this
> variable would be one of the main EVs, correct?
Yes
> In another
> words, if my
> design matrix looks something like this
>
> 1 value
> 1 value
> 1 value
>
> where the participants are no longer separated into groups,
> and the second
> column contains the values of the behavioral variable. In
> this case I did
> demean the behavioral variable, but I did not run randomise
> with the -D flag
> based on the previous suggestion from the FSL experts. So
> would this be a
> correct set up in this case?
Yes...
Doing
1 demeaned_value
1 demeaned_value
...
+ no -D flag
is equivalent to
demeaned_value
demeaned_value
...
+ -D flag
Cheers,
Gwenaelle
|