Dear Alfonso and hello everyone,

I have a few concerns regarding my MVPA rsfMRI results. I would appreciate your thoughts.

Here is a brief overview:
We have conducted a rsfMRI with high and low worriers (21 subjects per group) and assessed state worry during the scan via ratings. We performed MVPA with 8 components, looking at associations with state worry. State worry ratings are not normally distributed, so we went with non-parametric stats. Initially we set the voxel-level threshold at p<0.001 and at cluster-size k>100 (following the procedure in Thompson et al., 2016, http://dx.doi.org/10.1016/j.nicl.2016.05...). This resulted in three significant clusters. Post-hoc seed-to-voxel analyses, however, only revealed correlations between state worry and the functional connectivity of the cluster seeds and other brain areas for the second and third cluster, and not for the first one (voxel-level threshold p<.001, uncorrected and cluster-forming threshold p<.05, cluster-size p-FWE corrected).

We then came across Alfonso's recommendations in this forum (for non-parametric statistics: height threshold p-unc<0.01, cluster-mass threshold p-FDR<.05). I repeated the MVPA applying these thresholds which led to no significant clusters.

I am unsure what to do about these results now. Are the results with the Thompson threshold trustworthy at all?

Thanks in advance for your thoughts and thanks for this great forum!
Best regards
Fanny