Hi Anita,
The full F statistic will tell you how well the fitts curve
actually fits the data. For a quick test, I often:
- set Overlay to stats dataset
- set Olay/Thr sub-bricks to full-F (sub-brick #0)
- set the threshold (removing most of the noise)
- Clusterize (rmm=0, vmul=20 (so 20 voxels are required))
- then show 'all_runs' dataset with fitts using Dataset #N
(where 'all_runs' is the scaled EPI input to 3dDeconvolve)
Note that the underlay and overlay values (so scaled EPI and
full F-stat) for the voxel at the crosshair center are shown
in the lower right corner of the "Define Overlay" side panel.
So as you jump around, the current F-stat is displayed.
Note also that once you have clustered, it is nice to open
the "Rpt" window (meaning "cluster report"). You can "jump"
to the peak activation point in each cluster, etc.
- rick