[FieldTrip] Cluster based permutation post-hoc after ft_statfun_indepsamplesF
Martin Gallagher
martin.j.p.gallagher at gmail.com
Fri Nov 20 20:07:11 CET 2020
Hello,
This is my first post to the group. I've Googled this question and have
seen it asked, but I'm still unsure of the definitive answer.
I have time frequency analyses of three groups (three seizure types --
absence, myoclonic, and tonic). I am performing independent comparisons
tests of the three groups. The data are organized as subjects X channel X
frequency with 15-20 subjects in each group, 19 channels, and 19
frequencies (2-20 Hz) and only 1 time point.
I am using ft_statfun_indepsamplesF to determine differences among the
groups. After performing this initial cluster-based permutation test, what
is the appropriate post-hoc. I think if I did three pairwise
comparisons and made a Bonferroni correction it would be too strict.
Would it be appropriate to use the electrode/frequency clusters identified
in the initial 3 way comparison as channels and frequencies of interest and
then perform three pairwise comparisons and multiply the resulting pairwise
p-values by 3?
Thank you!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20201120/53377915/attachment.htm>
More information about the fieldtrip
mailing list