Dear all,
Does qdec allow analyses of covariance, and if so, how? is it possible for example to test the effect of a regressor while controlling for the effect of a particular measure?
thanks, best, Narly.
Narly,
qdec uses mri_glmfit to perform its analysis, and that is based on the General Linear Model, which is a type of analysis of covariance. Refer to these slides for more info:
https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial? action=AttachFile&do=get&target=freesurfer.groupanalysis.ppt
In the case of qdec, the measure is typically 'thickness', and you can select up to four regressors: two discrete (say, gender and diagnosis), and two continuous (say, age and left-hippo volume).
I think I answered your question, but Doug is the stats wizard, so he may chime-in, or you can restate your question.
Nick
On Thu, 2008-03-13 at 10:25 +0000, Narly A Golestani wrote:
Dear all,
Does qdec allow analyses of covariance, and if so, how? is it possible for example to test the effect of a regressor while controlling for the effect of a particular measure?
thanks, best, Narly. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
OK thanks, here's more:
So is it the case that if say I'm interested in the effect of diagnosis on cortical thickness, but want to *control* for the effect of age, I can just enter age as a variable, but by not selecting it for the analysis I would effectively be controlling for it? Or do I have to select it for the analysis, and assume that looking for the effect of diagnosis automatically controls for age?
thanks, Narly.
On Thu, 13 Mar 2008, Nick Schmansky wrote:
Narly,
qdec uses mri_glmfit to perform its analysis, and that is based on the General Linear Model, which is a type of analysis of covariance. Refer to these slides for more info:
https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial? action=AttachFile&do=get&target=freesurfer.groupanalysis.ppt
In the case of qdec, the measure is typically 'thickness', and you can select up to four regressors: two discrete (say, gender and diagnosis), and two continuous (say, age and left-hippo volume).
I think I answered your question, but Doug is the stats wizard, so he may chime-in, or you can restate your question.
Nick
On Thu, 2008-03-13 at 10:25 +0000, Narly A Golestani wrote:
Dear all,
Does qdec allow analyses of covariance, and if so, how? is it possible for example to test the effect of a regressor while controlling for the effect of a particular measure?
thanks, best, Narly. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
==================================== Narly Golestani Institute of Cognitive Neuroscience 17 Queen Square London WC1N 3AR United Kingdom Tel: +44 (0)20 7679 7529 ====================================
Narly, I don't think simply including a variable in the qdec.table.dat file will automatically control for that variable. It is my understanding that selecting a variable will control for that variable in all effects of the analysis. Someone please correct me if this is wrong. Cameron
OK thanks, here's more:
So is it the case that if say I'm interested in the effect of diagnosis on cortical thickness, but want to *control* for the effect of age, I can just enter age as a variable, but by not selecting it for the analysis I would effectively be controlling for it? Or do I have to select it for the analysis, and assume that looking for the effect of diagnosis automatically controls for age?
thanks, Narly.
On Thu, 13 Mar 2008, Nick Schmansky wrote:
Narly,
qdec uses mri_glmfit to perform its analysis, and that is based on the General Linear Model, which is a type of analysis of covariance. Refer to these slides for more info:
https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial? action=AttachFile&do=get&target=freesurfer.groupanalysis.ppt
In the case of qdec, the measure is typically 'thickness', and you can select up to four regressors: two discrete (say, gender and diagnosis), and two continuous (say, age and left-hippo volume).
I think I answered your question, but Doug is the stats wizard, so he may chime-in, or you can restate your question.
Nick
On Thu, 2008-03-13 at 10:25 +0000, Narly A Golestani wrote:
Dear all,
Does qdec allow analyses of covariance, and if so, how? is it
possible
for example to test the effect of a regressor while controlling for
the
effect of a particular measure?
thanks, best, Narly. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
==================================== Narly Golestani Institute of Cognitive Neuroscience 17 Queen Square London WC1N 3AR United Kingdom Tel: +44 (0)20 7679 7529 ====================================
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Yes, this is what the continuous variables are.
doug
Narly A Golestani wrote:
Dear all,
Does qdec allow analyses of covariance, and if so, how? is it possible for example to test the effect of a regressor while controlling for the effect of a particular measure?
thanks, best, Narly. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Cameron Ellis -
Doug Greve -
Narly A Golestani -
Nick Schmansky