Hi,
Changing the cluster-forming zstat threshold (assuming that's the
option you're using) means that you form larger clusters, and GRF
theory adjusts for that threshold lowering by requiring that clusters
are larger before they are considered significant (the p-value final
test). In general changing the cluster-forming threshold may make a
given cluster more or less significant, depending.....
So IF GRF theory was perfect, it wouldn't matter how low you took the
cluster-forming threshold, the final inference on cluster size would
still be fine - but it's not that easy - GRF approximations break
down for cluster-forming thresholds much less than (say) 2.3, so I
wouldn't recommend it.
(Of course if you're using cluster thresholding in randomise the
inference remains "perfect" even at very low cluster-forming
thresholds).
Cheers.
On 9 May 2007, at 23:42, Peter Fried wrote:
> Hi,
>
> I have noticed that lowering the Z threshold in the post-stats tab
> increases the group differences (A>B and B>A). How would I go about
> choosing an appropriate Z threshold so that the differences I see are
> meaningful and not exaggerated? Thanks.
>
> -Pete Fried
> Center for Magnetic Resonance Research
> University of Minnesota
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|