Thanks so much, Doug! I got the same results for both of these contrast matrices, so my original question is all set.

However, I was surprised when I ran mri_glmfit-sim that there were 47 clusters for the LH, some as small as 2.51mm^2 (and the max 382.23mm^3, which is way smaller than what I found in qdec with the 2 group analysis of these data). So, I’m thinking this can’t be correct (particularly with 1.3 threshold).

I followed the steps here https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/GroupAnalysis and copied my commands below.

1. mris_preproc mris_preproc --fsgd dx4subgroups_n146.fsgd --cache-in thickness.fwhm10.fsaverage --target fsaverage --hemi lh --out dx4subgroups_n146-lh-thickness.mgh

2. mri_glmfit mri_glmfit --y dx4subgroups_n146-lh-thickness.mgh --fsgd dx4subgroups_n146.fsgd dods --C group.effect.mtx --surf fsaverage lh --glmdir dx4subgroups_n146-lh-thickness_Fmatrix.glmdir

3. mri_glmfit-sim mri_glmfit-sim --glmdir dx4subgroups_n146-lh-thickness_Fmatrix.glmdir --cache 1.3 abs --cwp 0.05 --2spaces

Do you see any issues?

Thank you so much for your help!

Best, Andrea

Re: [Freesurfer] contrast matrix for one-way ANOVA with 1 factor, 4 groupshttp://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=subject:%22Re%5C%3A+%5C%5BFreesurfer%5C%5D+contrast+matrix+for+one%5C-way+ANOVA+with+1+factor%2C+4+groups%22&o=newest Douglas N Grevehttp://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=from:%22Douglas+N+Greve%22 Tue, 21 Feb 2017 12:53:08 -0800http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=date:20170221 That four group test will yield the same exact result as

1 -1 0 0

1 0 -1 0

1 0 0 -1

so I think you are ok

From: "Gold, Andrea (NIH/NIMH) [F]" andrea.gold@nih.gov Date: Tuesday, February 21, 2017 at 3:09 PM To: "freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu Subject: contrast matrix for one-way ANOVA with 1 factor, 4 groups

Hi,

I am trying to run a cortical thickness analysis comparing four groups – my goal is to conduct an omnibus F-test testing for any group differences. I have 0 covariates/nuisance regressors.

This would be equivalent to the “Contrast 4 group.effect.mtx” in this page, for 3 groups: https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf3G0V

But, instead of 3 groups I have 4 groups.

I thought the contrast matrix for 4 groups F-test would be: 1 -1 0 0 0 1 -1 0 0 0 1 -1

But for 3 groups it is: 1 -1 0 1 0 -1

Would you please clarify the correct contrast matrix for this test in Freesurfer mri_glmfit?

Thank you! Andrea

-- Andrea Gold, Ph.D. Section on Development and Affective Neuroscience National Institute of Mental Health 9000 Rockville Pike MSC 2670 Bldg. 15K Bethesda, MD 20892 Phone: 301-402-6955 Fax: 301-402-2010 Email: andrea.gold@nih.gov