Hi all,
This is a bit of a silly question but I haven't yet seen this explicitly
stated.
From what I have read about the single-session ICA implemented in MELODIC
(as a biologist I found the literature quite tough!), p-values are
irrelevant to ICA (or at least I have hardly ever seen p-values stated
near any ICA data). The technique separates components from noise, and
that is all there is too it- there is nothing to assess a p-value on- the
components are simply there, with no/little question as to their validity
(this last assertion seems a bit much to me). I assume this would also be
the case for group/tensor-ICA too. Is this, or something similar, true?
Sorry if I have completely got the wrong end of the stick
|