[Eeglablist] regardiing merging datasets

[Jung Ki-Young]

Dear all

I am planning to analyze drug effect on ERPs using auditory oddball
paradigm.
As I am making datasets, some questions are raised in my mind.
The datasets are consisted of pre-Tx standard and deviant, post-Tx standard
and deviant from 19 subjects, giving birth to four datasets.
Here are possible methods I am considering for my study.... but I have no
idea which method is the best.

1) make pre-Tx dataset using merging both std and dev data, and post-tx
dataset as the same way from all subjects, then do ICA each dataset

2) make pre-tx-std dataset, pre-tx-dev dataset, post-tx-std dataset, and
post-tx-dev dataset from all subjects, then do ica each dataset 

3) make individual dataset merging std and dev data from each subject for
each condition (pre- and post-tx), then ica -> cluster analysis in STUDY

4) do ica to each individual std and dev dataset without merging from each
subject for each condition (pre- and post-tx, then ica -> cluster analysis
in STUDY

I'd like to use cluster analysis to compare two conditions (pre-tx vs.
post-tx).
though I think, method 1 and 2, in fact, is not appropriate for cluster
analysis.

Any help would be much appreciated.


Ki-Young