On Oct 1, 2018, at 9:57 PM, David V. Smith <[log in to unmask]> wrote:
Hi Erik,
I'd be curious to hear what others say, but I think you'd want to correct the cluster-level p-value. An analogous situation arises with multiple networks and dual-regression analysis (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/DualRegression/UserGuide#The_need_for_correction.2C_and_correction_via_Bonferroni).
So, something like below should work:
cluster <other_options> --thresh=3.1 --pthresh=.025
Note that I think the default of the cluster-forming threshold is now a bit higher (z = 3.1) after some papers suggested z = 2.3 might be too low (e.g., Woo et al., 2014, NeuroImage; Eklund et al., 2016, PNAS).
Alternatively, you might be able to use PALM work these contrasts into the same group-level model and correct for multiple comparisons with that.
Best,David
Hi FSL Experts,
We planned an anatomically constrained apriori PPI analysis, and have an interesting question after looking at the data: the functional contrast that we want to use has two clearly defined clusters within the anatomical region mask. To get a sense of which of them may be driving an effect, we would like to use them both as seeds, but because we didn’t expect two we want to be careful about multiple comparisons.
Since there are only two clusters, I think we’ll have to use bonferroni correction. Typically, we could threshold with `cluster`, using the standard `--thresh=2.3 --pthresh=.05`. To bonferroni correct you’d divde the p by N (2), but we aren’t sure if this is appropriate for the cluster-masking threshold (2.3 is a p-to-z of .01, so take a p-to-z of .005 or z=2.58), the GRF p threshold, or both. E.g. do we use
cluster <other_options> --thresh=2.58 --pthresh=.025
Is it more appropriate to only correct the GRF? The cluster mask? For our goal of correcting ourselves, are both or neither appropriate? Thanks in advance for any insight,
Erik
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
--
David V. Smith, Ph.D.Assistant Professor of PsychologyTemple UniversityWeiss Hall, Room 8251701 North 13th StreetPhiladelphia, PA 19122Office Phone: 215-204-1552Lab Website: https://sites.temple.edu/neuroeconlab/
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1