Summary of what happened:
Running F-test one a one-sample design, the resulting output_vox_corrp_fstat1.nii.gz contains all zeroes, while the corresponding output_vox_corrp_tstat1.nii.gz has none-zero values.
I expected the two vox_corrp images to be similar since the fstat is just the square of the tstat. However, they should differ since the fstat is effectively doing a two-sided test while the tstat is one-sided.
Command used (and if a helper script was used, a link to the helper script or the command generated):
randomise -i diff_img.nii.gz -o output -d glm_gui.mat -t glm_gui.con -f glm_gui.fts -n 5000 -x
diff_img.nii.gz: 4D difference images between two conditions for 24 subjects
glm_gui.mat: 24x1 with all ones
glm_gui.con 1x1 with a single one
glm_gui.fts 1x1 with a single one
Version:
randomise v2.9
Environment (Docker, Singularity / Apptainer, custom installation):
Installed on Mac M1 using Python script