[Eeglablist] bootstrap and group analysis?

[wu xiang]

hi all

I want to know that how use bootstrap in group analysis.

-----------------------------------------------------
For example:
Given a memory study. 2 conditions, A and B. 8 subjects.
The aim was to investigate whether the coherence between C1 and C2 is
greater in condition A than in condition B.

Analysis includes 2 steps:
step 1:
For each subject, we get the coherence values of condition A and B.

step 2 (group analysis):
Here t-test can be used to compare coherence between condition A and B
across subjects.

The above is the usual steps in EEG analysis. In step 1, we get average
values (across trials) for each subject, no statistical analysis. In
step 2, a statistical analysis is used.

--------------------------------------------------------
Question:
To test the coherence is true coherence (not caused by volume
conduction), bootstrap may needed to be used.

question 1: I know how use bootstrap for each subject. However, how do
bootstrap for all subject? Is it to catenate the data of all subjects
(so like as one subject) and do bootstrap? Or, if do bootstrap for each
subject (for each subject, a P value can be got), how get a P value for
all subject?

question 2: bootstrap and t-test
bootstrap is to test whether the coherence is true coherence. t-test was
used to compare two conditions. How combine the two tests? For example,
if bootstrap results show that coherence of subject 2 is not true
coherence, than subject 2 is not included in group analysis?

-- 
Wu Xiang
the cognitive neuroscience lab of life school of University of Science and Technology of China
Hefei, Anhui, P.R.China. 230027
web-page:http://202.38.64.10/~rwfwu
You can mail to: rwfwu at ustc.edu or rwfwuwx at gmail.com