Hi Harris,
Currently, there is no option in swe that allows the user to directly get the residuals or an estimate of smoothness. The main reason for this is that parametric inference based on random field theory has, to the best of my knowledge, not been validated for the Sandwich estimator method or, more generally, for longitudinal or repeated measures data. Indeed, classic inference based on random field theory assumes that the errors are independent, which is not the case in the Sandwich estimator method. Thus, the only way to currently make FWE inference is to use a non-parametric wild bootstrap procedure using the option --wb.
All the best,
Bryan
> On 4 Jan 2020, at 23:46, Harrison Fisher <[log in to unmask]> wrote:
>
> Hi Bryan,
>
> This info is very helpful and much appreciated! I'll add in the --wb parameter to get the non-parametric outputs as well.
>
> We are also curious as to how the SWE compares to a more standard analysis where we use a cluster forming threshold of 2.3 (or 3.1) and the FWE p < 0.05 on the clusters in a parametric approach. I know that I can run smoothest and cluster to achieve this same thresholding on a zstat output of SWE, for example, but I am missing how to obtain the degrees of freedom and residual image from swe, which are recommended over just using the zstat as an input to smoothest.
>
> Best,
> - Harris
>
> ########################################################################
>
> To unsubscribe from the FSL list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|