[FieldTrip] between group statistics of coherence

[Johannes Tuennerhoff]

Dear Eric,
dear other experts,

We conduct a
 study that involves corticomuscular coherence analyses and I am still a bit new to the field.
Even though I
 read several papers and searched the fieldtrip mailing list, I am still
 not sure how to do the statistical testing correctly. I read your 
message "coherence and group analysis" from April 21 2015 with great 
interest and I was hoping you could maybe give me some advice.
There are two groups and each group has the same condition (a motor movement task).
I
 have the coherence spectra for the range 5-55Hz (binned at 0.3Hz) of an
 EEG signal x and an EMG signal y  in 24 healthy controls and 19 
patients and I would like to test if there is a significant difference 
between the two groups. The coherence of the individual subjects was 
computed from around 120 segments of each 3s.

Specifically I have the following questions:

1.
 Can I use the fieldtrip function ft_freqstatistics to calculate 
Monte-Carlo estimates of the significance probabilities and setting 
cfg.statistic = 'indepsamplesT' using the coherence spectra as data? Or 
would I need to do a manual z-transformation like 
z1=atanh(Coh)-1/(dof1-2) beforehand? If I understood it right cant use 
cfg.statistic = 'indepsamplesZcoh' because it is just meant for within 
subject designs. Would I need to take tapers and segments into account 
for the degrees of freedom?

2. Would I need to correct for 
multiple comparisons when I use a Monte-Carlo aproach? If I would like 
to, could I just use the cfg.correctm='bonferroni' or 'fdr' option in 
fieldtrip?

3. I think I saw other groups just using a parametric chi-square test, would you consider this aproach correct too?

4.
 Would it be valid to conduct a repeated measures ANOVA first with GROUP
 as between-subject effect and frequency bins as within-subject effect?

5. Could I use the parametric tests (3. and 4.) after a z-transformation?

I will be very happy about any suggestions and opinions. Thank you very much.
Best regards
Johannes
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160201/34bbbfae/attachment-0001.html>