Hi All,
I have a couple of questions regarding how FSL comes up with the variance estimated for the
varcopes. On your site you say that the variance for EV is calculated from the residuals. For a
given contrast of two EVs, the variance is calculated by combining the varienced.
Why are there different variences for different EVs if there is only one residual term (general linear
model states that the data is made up of the sum of EVs*Betas + a single residual)?
If the error comes from what is left over after estimating a single EV, then doesn't activity due to
the other EVs end up in your error? Say in a given dataset, you have two tasks, one task of interest
and a control task. If the control task colocolizes with the task, will activity due to the control task
add to the error term and vice versa? If this is the case, you would be more powered to see a
difference between these two tasks if you stimulated in seperate runs.
Thanks as always.
Jennifer Bramen
UCLA Brain Mapping Center
|