Dear freesurfer experts:
We are using freesurfer to preform a cortical thickness analysis between patients and controls. We plan to use the mean_thickness and subject age as covariates. It's fine to analysis as group x age or group x mean_thickness, respectively. But when we set design as group x age x mean_thickness, an error report appeared in the terminal: -------- Matrix condition is 11355.6 ERROR: matrix is ill-conditioned or bad scaled, condno = 11355.6 Possible problem with experimental design: Check for duplicate entries and /or lack of range of continuous variables within a class. (continuous with values in Xg.dat) ------
Does the problem course by the different "range" from age value (between 18-28) and mean_thickness (between 2-4) with in the same subject? if so, how can we solve it? Does proportional scaling on age or mean_thickness help?
Any comments are welcome and very appreciate.
you can divide the continuous variables by the standard deviation.
doug
lordowen wrote:
Dear freesurfer experts:
We are using freesurfer to preform a cortical thickness analysis between patients and controls. We plan to use the mean_thickness and subject age as covariates. It's fine to analysis as group x age or group x mean_thickness, respectively. But when we set design as group x age x mean_thickness, an error report appeared in the terminal:
Matrix condition is 11355.6 ERROR: matrix is ill-conditioned or bad scaled, condno = 11355.6 Possible problem with experimental design: Check for duplicate entries and /or lack of range of continuous variables within a class. (continuous with values in Xg.dat)
Does the problem course by the different "range" from age value (between 18-28) and mean_thickness (between 2-4) with in the same subject? if so, how can we solve it? Does proportional scaling on age or mean_thickness help?
Any comments are welcome and very appreciate.
Dear Professor Greve:
Thanks for your advisement. After re-value the covariates, the same error report still popup. The value of age goes between 6.5-10.5 (origin SD =2.9), and mean_thickness goes around 31-35 (origin SD = 0.07). Can you give us some suggestions?
Any other comments also welcome and appreciate..
在 2010年2月24日上午12:00，Douglas N Greve greve@nmr.mgh.harvard.edu 寫道：
you can divide the continuous variables by the standard deviation.
doug
lordowen wrote:
Dear freesurfer experts:
We are using freesurfer to preform a cortical thickness analysis between patients and controls. We plan to use the mean_thickness and subject age as covariates. It's fine to analysis as group x age or group x mean_thickness, respectively. But when we set design as group x age x mean_thickness, an error report appeared in the terminal:
Matrix condition is 11355.6 ERROR: matrix is ill-conditioned or bad scaled, condno = 11355.6 Possible problem with experimental design: Check for duplicate entries and /or lack of range of continuous variables within a class. (continuous with values in Xg.dat)
Does the problem course by the different "range" from age value (between 18-28) and mean_thickness (between 2-4) with in the same subject? if so, how can we solve it? Does proportional scaling on age or mean_thickness help?
Any comments are welcome and very appreciate.
-- 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
Can you send your fsgd file and your command-line?
doug
lordowen wrote:
Dear Professor Greve:
Thanks for your advisement. After re-value the covariates, the same error report still popup. The value of age goes between 6.5-10.5 (origin SD =2.9), and mean_thickness goes around 31-35 (origin SD = 0.07). Can you give us some suggestions?
Any other comments also welcome and appreciate..
在 2010年2月24日上午12:00，Douglas N Greve greve@nmr.mgh.harvard.edu 寫道：
you can divide the continuous variables by the standard deviation.
doug
lordowen wrote:
Dear freesurfer experts:
We are using freesurfer to preform a cortical thickness analysis between patients and controls. We plan to use the mean_thickness and subject age as covariates. It's fine to analysis as group x age or group x mean_thickness, respectively. But when we set design as group x age x mean_thickness, an error report appeared in the terminal:
Matrix condition is 11355.6 ERROR: matrix is ill-conditioned or bad scaled, condno = 11355.6 Possible problem with experimental design: Check for duplicate entries and /or lack of range of continuous variables within a class. (continuous with values in Xg.dat)
Does the problem course by the different "range" from age value (between 18-28) and mean_thickness (between 2-4) with in the same subject? if so, how can we solve it? Does proportional scaling on age or mean_thickness help?
Any comments are welcome and very appreciate.
-- 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
freesurfer@nmr.mgh.harvard.edu
-
Douglas N Greve -
lordowen