Hello,
If each column is a single 1, then each volume can be completely accounted for in the model, putting multiple ones in the same column would fit the "mean" outlier instead and so not fully account for the confound variance.
Hope this helps,
Kind Regards
Matthew
> Dear all,
>
> By default fsl_motion_outliers outputs a "confound matrix" in which a new column is created for each volume identified as a motion outlier. Can anybody please explain to me why this would be preferable to collapsing all columns into one, containing multiple 1's, corresponding to each of the outliers? Wouldn't the latter approach have a lower impact on the degrees of freedom?
>
> Thanks!
> Marco
|