Sure, those t-thresholds appear to correspond to the P-thresholds. But there are a few things to clarify:

1. You have not performed a non-parametric test, and the fact that you used randomise is irrelevant. The t-values that you get out of randomise are identical to those that you get when performing a parametric t-test. What differs is whether or not those t-values are determined to be significant or not. What you are actually doing here is a parametric t-test at your chosen level of significance, without correction for multiple comparisons.

2. Given the above, you are also assuming that the noise in your data follows a gaussian distribution, which is generally considered to not be the case in VBM-type data, and is the reason why randomise is used in the FSL VBM pipeline.

Thank you Paul!

This is a figure for the supplementary.

I write in the method:"we perform a nonparametric 2-sample t-test using randomise. There was no correction for multiple correction."

I just want to make sure that I'm right:

I have 23 subjects in each group.

df=44

for p=0.001 I set t=3.28 (I set this value as "min" in fsleyes)

for p=0.0001 I set t= 4.057 (I set this value as in "min" in fsleyes)

Thanks again,

Tali

########################################################################

To unsubscribe from the FSL list, click the following link:

https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1