Dear Aser,
For cluster level inference, you first need to specify a height
threshold that will define clusters (hence it being called a cluster
forming threshold), see e.g.:
http://www.fil.ion.ucl.ac.uk/spm/course/slides17-oct/05_ParamThresh.pptx
One assumption when using the random field theory is that clusters are
defined by a sufficiently high threshold which is usually satisfied by
using a cluster forming threshold of p<0.001 (uncorrected).
Best regards,
Guillaume.
On 31/10/17 18:06, fmri2013 wrote:
> Hi all
>
> what does mean when people use for example p 0.001 uncorrected at the
> vocal level but then corrected at the cluster level?
>
> what if we use a p of 0.01 or 0.0001? How does this going to affect the
> cluster corrections ?
>
> Many thanks
>
> Aser
>
>
>
> Sent from my Samsung Galaxy smartphone.
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|