 |
 |
Can anyone tell me if there is a difference when running a GLM between 1) including a confound matrix in which each column in the matrix identifies only a single volume (e.g., the kind of output generated by feat_motion_outliers) and 2) including a confound matrix that is only a single column that identifies all confound volumes? I can't decide if option 1 is more appropriate for mathematical reasons over option 2, or if option 1 is what people use just because of the way feat_motion_outliers produces its output. To put it another way, if I already know which volumes I want to treat as confounds, can I simply specify those volumes in a single column, or is there a reason I need to transform the single column into a matrix in which each column identifies only a single confound volume? Thanks in advance, Ruskin