I am in the process of identifying the differences in the brain network before and during the onset of microsleeps. I have few questions regarding the usage of NBS in statistically analyzing the epochs in EEG.

I am considering 2 adjacent epochs - one 0.25 seconds before the onset of microsleep (group 1) and one 0.25 seconds after the onset of microsleep (group 2). Each of these groups consists of 64 effective connectivity matrices, this is for one event of one subject. Like this, I have several events for a particular subject (all have both the microsleep and non-microsleep events) and a total of 8 subjects in my study. I think I can't take a two-sample test between the groups since they belong to the same subject. The overall objective is to find the differences in the brain network before and during microsleeps irrespective of the event and/or the subjects.

It will be of great help if you can you please throw some light on how I can use NBS for my purpose.

Thanks for your time and help.
Warm Regards
Uma

Hi Uma,

you will need to use a repeated measures design, given that you have two measurements in each subject. The NBS manual has an example of repeated measures t-test.

Suppose you have 2 subjects (for example) arranged in the order of: S1 (group 1), S2 (group 1), S1 (group 2), S2 (group 2). Your design matrix will look like this:

1 0 1
0 1 1
1 0 0
0 1 0

and your contrast will be either [ 0 0 -1] or [0 0 1].

You will also need to add exchange blocks. See manual or other posts on this forum for details.

Andrew

Originally posted by Uma Venkat:

Thanks for your immediate response.. will definitely try that. 

In line with the same thing I have one more question. I actually have 64 measurements in group 1 and 64 in group 2. Should I take the average and have only one measurement per subject ??? Or can I have any number of measurements for each subject ??? Also, these measurements in group 1 and group 2 are not aligned temporally or in any manner. So will permuted paired t-test be better than paired t-test ????

Thanks for time and help.

Hi Uma,

this really depends on the hypothesis that you want to test. If you seek to test for differences between the two groups, averaging across the two groups would not be beneficial. I don't follow your second question.

Andrew

Originally posted by Uma Venkat: