Dear FSL experts,
I just wanted to take a deeper look at how FSL performs a paired t-test, with COPE/VARCOPE files.
With one-sample t-test, this seems to make sense, as the VARCOPE file for each subject contains information about the within-subject variance. FSL then would perform Bayesian approach to estimate the between-subject variance (If I understand the concept correctly).
With the paired t-test, I imagine the COPE could be calculated easily by taking the difference between the two COPE files within the subject. But how do we utilized VARCOPE information in this setting? What would be the corresponding "within-subject" variance in this case?
Any help or reference would be appreciated - Thanks!
|