Those are the same things. The way you control for age is to compute a value for both groups at the same age (eg, age=0). If you demean the age, then the difference will be at age=meanage. With DOSS, the difference is always the same regardless of age (which makes it more interpretable) doug
On 02/25/2014 03:00 PM, pfotiad@nmr.mgh.harvard.edu wrote:
Hi Doug,
Thank you for your response. Concerning scenario 2, I believe I have some hard time understanding the term intercept. If I'm not mistaken the intercept represents the cortical thickness at "age 0." The resulting clusters by using a design matrix of [1 -1 0 0] will give me the areas where the intercept is significantly different between the Diseased group and the healthy group.
Are those clusters the same clusters that have a different mean regional cortical thickness in Diseased versus Healthy subjects after controlling for age, or are those two completely different things?
Thanks again, Panos
On 02/25/2014 02:07 PM, pfotiad@nmr.mgh.harvard.edu wrote:
Hi FS community,
I had a few questions regarding my understanding of the group analysis in FS:
- If I use an fsgd file with Two Groups (One Factor/Two Levels), No
Covariates, and my contrast matrix is [1 -1], do the resulting clusters (if any) show the regional significant differences between the two groups by comparing just the means of the values that I am comparing? For instance, if I'm comparing cortical thickness between a diseased and a control group, is this going to give me the clusters where the Diseased subjects' (no age included in the fsgd file, just the two groups) mean/regional cortical thickness differs significantly from the controls'? In other words, is a 50 yo subject's mean/regional cortical thickness going to be treated the same as a 80 yo subject's?
Yes, just a simple two-group t-test.
- Now suppose that I would like to control for age. Then I'm assuming
that I need to create an fsgd file with two Groups (One Factor/Two Levels) and One Covariate (age). Similarly, if I would want to visualize the differences in the regional values of cortical thickness between the two groups but after they have been controlled/adjusted for age, what contrast matrix should I use? Normally I would use [0 0 1 -1] but this will give me the significant differences in the age slopes not the mean values, unless I am missing something here.
Correct, that will give you the difference in age/thickness slopes. You could do [1 -1 0 0] to give you the difference in intercepts. A better way, is to first look for differences in the slopes. If you do not find any, then rerun with DOSS with a contrast of [1 -1 0]. By using a DOSS, the difference in intercept is always constant regardless of age (as opposed to DODS). doug
Thanks in advance for your help and sorry for the lengthy message.
Best, Panos _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer