Hi Michael
> Is my interpretation that the F-test assigning 1 "one" to the task and 0 "zero" to all
> control variables (such as motion regressors) addresses the question whether there is
> either activation or deactivation compared to baseline correct?
Yes, exactly.
> How do I take this question to a second level analysis? The F- test produces no ".con"
> images which are being used for a second level analysis as "scans".
You can perform a second level F test on con* images generated at the
first level. This is because the con* file itself has both positive
and negative values, even if a t-test will only give you a one-sided
result. So, in your example, you can simply perform a t-test at the
subject level testing for a positive effect of your task ([1 0 ...]),
producing a con* image. At the second level, you can select these
con* images, one per subject, and run an F test on them. This F test
will show you voxels where your task had a positive or negative
effect.
Hope this helps!
Jonathan
|