Statistics for coherence in a within-subject experiment

Jurij Dreo jurij.dreo at GUEST.ARNES.SI
Sat Jun 17 12:18:26 CEST 2006


Hello,

I am using BESA to perform coherence and source coherence 
calculations for a multiple-subject/multiple-condition experiment. In 
the experiment I have 11 subjects, 6 conditions and around 30 sets 
per condition per subject.

In BESA the data was preprocessed and the coherence values in the 
Time-Frequency domain were calculated for each person.

As BESA has no capabilities to calculate statistics in a multiple-
subject study, their support personnel directed me towards FieldTrip.

I´ve been looking over the documentation and tutorials on your web 
page and also downloaded the scrip and ran it in matlab.

The tutorial on Cluster-Lever randomization Tests for the Time-
Frequency domain is quite straight-forward. I am however not quite 
sure which statistic to use for the cfg.statistic variable to analyze 
coherence in a within-subject study such as the one I have.

The documentation on the cfg.statistic variable says it can have 
these values:
1. indepsamplesT' (independent samples T-statistic) 
2. 'indepsamplesregrT' (independent samples regression coefficient T-
statistic)
3. 'indepsamplesZcoh' (independent samples Z-statistic for coherence)
4. 'indepsamplesTsqcrs' (independent samples T-square-statistic for 
the cross-spectrum)
5. 'depsamplesT' (dependent samples T-statistic)
6. 'actvsblT' (activation versus baseline T-statistic)
7. 'depsamplesregrT' (dependent samples regression coefficient T-
statistic) 
8. 'indepsamplesF' (independent samples F-statistic)
9. 'depsamplesF' (dependent samples F-statistic)

I am not sure if any of these options are compatible with a within-
subject multiple condition coherence statistical analysis.

option 3. - indepsamplesZcoh is obviously only suitable for a between-
trial, two-condition experiment

and there are no other options that explicitally say they are used 
for coherence. 

So how can one compare 6 conditions for 11 subjects with regards to 
coherence if one imports the Time-Frequency domain data for each 
person into FieldTrip?

Thank you for your help!

Regards,
Jurij Dreo



More information about the fieldtrip mailing list