[FieldTrip] General matlab script for cluster based permutation testing?
markus.gschwind at gmail.com
Wed Sep 9 13:37:18 CEST 2015
Thank you for your lines.
I have now written a matlab script, which uses the bwconncomp.m function
In this context, I have the following question concerning the initial
T-value threshold (which allows binarization and subsequent clustering).
I have bivariate histograms of two conditions, which I want to compare for
differences (N=50). When doing point-wise paried T-tests, my T-values have
a cdf-like shape and go from -8 to +8, meaning, that I have potentially
significant positive and (slightly more) negative differences.
In the paper it says :
(1) For every sample, compare the MEG-signal on the two types of trials
> (...) by means of a t-value (or some other number that quantifies the
> effect at this sample).
> (2) Select all samples whose t-value is larger than some threshold. (This
> threshold may or may not be based on the sampling distribution of the
> t-value under the null hypothesis, but this does not affect the validity of
> the nonparametric test; see further.)
> (3) Cluster the selected samples in connected sets on the basis of
> temporal adjacency.
And then later :
Also, for a two-sided test, the clustering in step 3 is performed
> separately for samples with a positive and a negative t-value.
=> So is it right to choose the threshold for the positive T-values and the
negative T-values separately?
=> Would you then choose the outer 97.5 percentile as the threshold (i.e.
T>7.2 for the positive and T< -6.3 for the negative values)?
However this is drastically more restrictive than a T-value of >2.0097 in
case of a single two-sided paired T-test with df=49 (N=50), which would be
applied without any correction.
The choice of this initial threshold defines the cluster size!
I would be grateful for any hint! Thank you in advance!
2015-09-02 14:14 GMT+02:00 Robert Oostenveld <r.oostenveld at donders.ru.nl>:
> Hi Markus,
> That is not something which is not is easily implemented in a single
> script. It requires a whole variety of functionality which is part of the
> FT toolbox (shuffling, clustering, bookeeping), but which cannot easily be
> used if your data is in another format. Perhaps you can reformat your data
> to make it FT-like.
> best regards,
> On 31 Aug 2015, at 13:56, Markus Gschwind <markus.gschwind at gmail.com>
> > Dear all,
> > I wonder if someone could guide me how to use the cluster-based
> permutation testing after (Maris & Oostenvidle, 2007) on non-fieldtrip
> data, for example on simple matrices in matlab.
> > Or is there even a matlab script around?
> > Thanks in advance,
> > Markus
> > _______________________________________________
> > fieldtrip mailing list
> > fieldtrip at donders.ru.nl
> > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the fieldtrip