[FieldTrip] Minimal threshold in TFCE?

[Nir Ofir]

Dear list members,

I'm playing with threshold-free cluster enhancement on data and I collected
and I noticed a detail in fieldtrip's implementation which I wondered
about: the smallest cluster-forming threshold (which is called h0 in the
TFCE papers I found). In fieldtrip I saw that it's set (implicitly) at 0,
while for example in LIMO_EEG it's set at the minimum of the data. The
literature isn't clear about this too: Smith & Nichols 2009 write that it's
typically zero, but Pernet et al. 2015 write it's typically the minimum of
the data. Does the minimum threshold affect the results of the statistical
test? Are there considerations for choosing either of the options?

Thanks for your time,
Nir Ofir
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20230906/2cbca83c/attachment.htm>