Permutation test

[#1]

Dear reader, 

I'm looking at the correlation between peripheral metabolism (Insulin, glucose, HbA1c) and brain metabolites (NAA, Cho, mI) in T2DM and control group. I choose spearmans rho because there was no normal distrbution. Now i'm intressted in looking if there is a difference between the correlation. So for example is there a difference in the correlation insulin-NAA in T2DM compared to the correlation insulin-NAA in the control group. I think permutation test is the best way to go. But I find no good way to do it. There was an add in: : https://community.jmp.com/t5/Abstracts/A-JMP-Add-In-for-Teaching-Statistical-Inference-Using-Resampl...

But this one is archieved. So my thought was to use a script but i'm not the best coder and can't find an example script. I'm using Jmp 17 pro

I hope you could help me, with kind regard Teun.

[#2]

Hi,

I am unsure whether I understand your question correctly.

Instead of calculating the correlation and then comparing them between subject groups, have you considered using a Mixed Linear Model (Analyze > Fit Model) where Y would be your INSULIN data (not the correlation values), and the effects would be NAA data, DISEASE STATUS (T2DM vs. CONTROL), and an INTERACTION Term of NAA x DISEASE STATUS.

Let us know if this is going in the right direction or if it is not what you are looking for.

Best regards,

TS

[#3]

Hi,

It could have been a suitable option, but the assumptions are violated — for example, the residuals are not normally distributed for insulin.