For thickness studies, if you have just one group, then the only thing to test is the slope, and it should not matter (but you might need DeMean to make it well conditioned). doug
Knut J Bjuland wrote:
Hi
Should i also add demean flag if I had just one group an one continus variable like iq. I get in this case significant findings on the other side of the brain in Fs 5.1 than in Fs 4.30
Thank you for you time Knut j
Date: Thu, 1 Dec 2011 12:09:52 -0500 From: greve@nmr.mgh.harvard.edu To: knutjbj@hotmail.com Subject: Re: [Freesurfer] Four Groups (Two Factors/Two Levels), One
Covariate with very strange results
I think the problem is just a scaling issue with the ages being 20x
more
than the offset (which is 1). The next version will handle this internally, but you can go ahead and use the DeMean flag in the
FSGD. If
your goal is to test for a difference in covariate slope across the groups, then adding the DeMean flag will not affect the results. However, if the continuous variables (covariates) are nuisance variables, then you should first test your contrast of interest on the slopes (using the DODS model). If there is no difference, then you should re-run the analysis with DOSS. doug
Knut J Bjuland wrote:
Hi Nick
Thank for the replay.
I have one group with premature that I compares to a group with term born controls. However when I control for gender and age I get this error message.
ERROR: matrix is ill-conditioned or badly scaled, condno = 11065.5
Possible problem with experimental design: Check for duplicate entries and/or lack of range of continuous variables within a class. If you seek help with this problem, make sure to send:
- Your command line:
mri_glmfit --y lh.60.mgh --fsgd group_diff_gender_age_old.fsgd --C Controls_VLBW.intercept.mtx --cortex --glmdir new --surf fsaverage lh 2. The FSGD file (if using one) 3. And the design matrix above
But when I added those subject that are commented I am able to run mrl_glm fit without problems. I have checked for the usual errors and there are none duplicated entries. The variables are 4 fixed variables and one continuous variable named age in years. When I add the subjects that are commented out I am able to produce an design matrix. Having said that I am able to produce an matrix that works by adding demean flag, but I am uncertain if it will is good idea. Could you please take a look at my fsgd file and error message.
I know that there is difference between Freesurfer versions in results. But should the difference be huge or small. I am able to reproduce results in previous Freesurfer when using group difference in thickness, but not when I am adding continues variables for instance in correlation analysis. I am enclosing two jpg files of the results.
Best regards Knut J
-- 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: www.nmr.mgh.harvard.edu/facility/filedrop/index.html
The information in this e-mail is intended only for the person to
whom it is
addressed. If you believe this e-mail was sent to you in error and
the e-mail
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to
you in error
but does not contain patient information, please contact the sender
and properly
dispose of the e-mail.