significance of coherence differences
e.maris at DONDERS.RU.NL
Mon Aug 9 18:58:57 CEST 2010
If you get confused about the literature then I am partly to blame.
Things are different for single subject studies (units-of-observation are
trials) and multiple subject studies (units-of-observation are subjects).
You describe a multiple subject study of the within-subjects type (every
subjects participates in both experimental conditions). For a single channel
pair and a single frequency bin, the appropriate statistic is the dependent
(paired) samples t-statistic or, in a nonparametric framework, the Wilcoxon
signed rank sum test. (You mention the Wilcoxon rank sum test, but this is
for a study of the between-subjects type.) You apply this test statistic to
the condition-specific coherences, obtained by summing and normalizing the
trial(taper)-specific cross-spectra. This statistical test works fine
(asymptotically, for the dependent samples t-statistic) if the number of
trials in the two conditions are equal. The requirement of an equal number
of trials follows from the fact that coherence is a biased measure, with the
bias being dependent on the number of trials.
The situation is different for a single subject study. You may want to have
a look at our paper (Maris, Schoffelen, Fries, 2007), which also deals with
the multiple comparisons problem.
dr. Eric Maris
Donders Institute for Brain, Cognition and Behavior
Center for Cognition and F.C. Donders Center for Cognitive Neuroimaging
P.O. Box 9104
6500 HE Nijmegen
T:+31 24 3612651
Mobile: 06 39584581
F:+31 24 3616066
E: e. <mailto:e.maris at donders.ru.nl> maris at donders.ru.nl
From: FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL] On Behalf
Of Jan Hirschmann
Sent: donderdag 5 augustus 2010 15:50
To: FIELDTRIP at NIC.SURFNET.NL
Subject: [FIELDTRIP] significance of coherence differences
Dear fieldtrip experts,
I am getting more and more confused about the literature of significance
tests of coherence differences. The one essential point I did not quite
understand is why conventional tests do not seem to apply. Lets say I have
two vectors of mean alpha band coherences, relating to two experimental
conditions (continuous data, no trials). The vector entries stem from
different subjects: C1=[meancoh_cond1_subj1, meancoh_cond1_subj2,.] and
C2=[meancoh_cond2_subj1, meancoh_cond2_subj2,.]. What keeps me from
comparing the two vectors, say, with a Wilcoxon ranksum test?
Thanks for any comments,
The aim of this list is to facilitate the discussion between users of the
FieldTrip toolbox, to share experiences and to discuss new ideas for MEG and
The aim of this list is to facilitate the discussion between users of the FieldTrip toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip