How to set the threshold in 'permutation_cluster_test'

External Email - Use Caution

Hi all,
    When I run 'permutation_cluster_test', I do not know how to set the
threshold. some example with the value threshold=6.0, what is the meaning
of this value? I want to test whether there is a power difference between
the two groups. So, my data is (n_individual,t_freqs,t_timepoint) with one
channel.
   Wish you could give me some suggestions.
   Thanks a lot.
   Tao
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20181213/27cbd91a/attachment-0001.html

External Email - Use Caution

hi,

if you use an F-test you can use: scipy.stats.f.ppf
to get a threshold based on a proba such as 0.05

We should improve our doc of the stats module. I hope
we can do this for the next release.

HTH
Alex

External Email - Use Caution

Dear Tao,

for your purpose you should define one groups power values as the control and therefore you can get your threshold from this group. Remember the test will give you clusters, which are defined significant because it's different to your applied threshold.

Greetings

Daniel

--- Please consider the environment before printing this email.

        External Email - Use Caution

hi,

if you use an F-test you can use: scipy.stats.f.ppf
to get a threshold based on a proba such as 0.05

We should improve our doc of the stats module. I hope
we can do this for the next release.

HTH
Alex