Fwd: RE: clusterrandomization for 2 epochs separately

[Nicolas Robitaille]

Dear Olga,

I have a quick answer for you, and would be please to heard people's comment on it. If you had a priori reason to belive something would happen for these electrodes and time-window, you don't need cluster analysis at all. All the solutions to the multiple-comparison problem are to be used in exploratory situation, when you don't know what to expect. If you had specific hypothesis, don't be shy: test them directy with a T-Test on the average voltage for a pre-determined time-window. 

Think about it, if you did expect these effects to be there, you could had create a montage with only one or two electrodes, and the multiple-comparison problem would not apply at all.

Hope this help,

Nic

************************************
Nicolas Robitaille, candidat Ph.D
Département de Psychologie
Université de Montréal
C.P. 6128, succursale Centre-ville
Montréal, Québec H3C 3J7
Tel.: 514-343-6111 x2631
Fax: 514-343-5787
************************************


> Date: Wed, 22 Oct 2008 12:10:01 +0100
> From: olga at GRAPHICMIND.INFO
> Subject: [FIELDTRIP] Fwd: RE: clusterrandomization for 2 epochs separately
> To: FIELDTRIP at NIC.SURFNET.NL
> 
> > Dear Eric, 
> > I would like to consult with you about some our ERP data and
> > application of the clusterrandomization to it. 
> > 
> > I am writing now the article about ERP signs of visual change
> > detection. 
> > 
> > We have ERP to the "deviant" and "standard" stimuli sampling rate
> > 512 Hz, 64 channels (Biosemi), 12 subjects. 
> > 
> > If we use the paired-wise t-test to compare the ERPs' amplitudes in
> > response to standard and deviant stimuli separately for each
> > electrode and time point (and plot it see the picture at the end of
> > the letter) we can see 2 (4) time regions of differences. But
> > application of clusterrandomization analysis to these data (latency
> > 0-500 ms) reveal only late cluster as significant, the early cluster
> > is marginally significant (p=0.1). But I really believe in this early
> > effect (120-160 ms), it is very important for our paper. When I run
> > the clusterrandomization analysis on the narrowed epoch (0-300 ms) I
> > got the early cluster significant. But the problem than that I loose
> > the late effect, which is also worth mentioning in the paper. 
> > 
> > Therefore, I would like you opinion about the possibility to use the
> > clusterrandomization analysis separately on the 2 epochs of interest,
> > let say first 250 ms post-stimulus (0-250 ms) and next (250-500 ms). 
> > 
> > In this case there would be two significant clusters...  
> > Best Regards, 
> > 
> > Olga Sysoeva 
> > P.S. 
> > 
> > I also used different parameters for clusterrandomization (ndis,
> > nchannels) but this does not change the results much.  
> > P.P.S. 
> > 
> > Additional material 
> > The paired-wise t-test was used to compare the ERPs' amplitudes in
> > response to standard and deviant stimuli separately for each
> > electrode and time point. 
> > 
> > The standard and deviant ERPs comparison reveals 2(4) intervals of
> > significant differences:  
> > 
> > Fig.1. At this weird J picture we see the boxes of significant
> > differences between standard and deviant ERPs for each time point (x
> > axis) and electrode (y axis). 
> > Olga Sysoeva, PhD
> > Institute of Higher Nervous 
> > Activity and Neurophysiology
> > Russian Academy of Sciences
> > 5a Butlerova str.
> > Moscow 117485
> > RUSSIA
> > tel.: (7-095)-3347011, 
> > fax:(7-095)-338-85-00.    
> > 
> > Links:
> > ------
> > [2] http://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.

_________________________________________________________________


----------------------------------
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/20081022/43ce93e9/attachment-0002.html>