clusterrandanalysis questions

Eric Maris maris at NICI.RU.NL
Thu Aug 16 11:11:10 CEST 2007


Dear Sameer,





> 1)   cfg.onetwo = 'onesided_2<1'  vs  cfg.onetwo = 'onesided_1<2

> >From my understanding, it seems that one sided tests only look for

> positive clusters i.e. the test statistic higher than cfg.alphathresh.

> Thus clusters showing up as posclusters for coherence difference in one

> case should not show up at all in the other. Yet, both the tests seem to

> give me same clusters.



If this is the case, then there is an error in the code. Can you send me
your config and screen output of clusterrandanalysis?



>

> 2)   using cfg.onetwo = twosided I should get clusters which in some

> way are union of the two sets above. Yet, I seem to get completely

> different clusters, sometimes with  nothing in common between twosided and

> onsesided tests.



The clusters with the twosided-option should NOT be the union of the
clusters with the two onesided options. This is because the thresholding
values with the twosided-option are at cfg.alphathresh/2 and
(1-cfg.alphathresh/2), whereas for the two onesided options they should be
at, respectively, cfg.alphathresh and (1-cfg.alphathresh).





> 3)   Most clusters I find seem to have a high p-value. Would the

> confidence increase if I use a different alphathresh which I think is used

> for thresholding the Z-statistic for defining clusters. (I just realized

> that I neglected to define cfg.alphathresh and though I could not find a

> default value used in clusterandanalysis, the program is obviously using

> some).



On the basis of my experience, I do not expect a substantial influence of
the cfg.alphathresh value on the p-values.



> 4)   What is the motivation behind  cfg.minnbchan ? Should it be based

> upon the neurobiological effect I am investigating?



The motivation behind cfg.minnbchan is that some samples (i.e.,
channel-frequency-pairs or channel-frequency-time-triplets) with a big
sample-specific statistic may have a few neighbouring samples with also a
big sample-specific statistic, just by chance. To "prune out" these chance
connections, set cfg.minnbchan high (e.g., 3).





Kind regards,



Eric Maris









dr. Eric Maris

NICI/Biological Psychology and

F.C. Donders Center for Cognitive NeuroImaging

University of Nijmegen

P.O. Box 9104

6500 HE Nijmegen

The Netherlands

T:+31 24 3612651 (NICI)

T:+31 24 3610754 (FCDC)

F:+31 24 3616066 (NICI)

E: maris at nici.ru.nl

MSc Cognitive Neuroscience : <http://www.ru.nl/master/cns/>
www.ru.nl/master/cns/




----------------------------------
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/fcdonders/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20070816/cca945ca/attachment-0001.html>


More information about the fieldtrip mailing list