Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Are you sure that it is saving t-values in F.mgh and not F-stats?
From: Glenn Lawyer d.g.lawyer@medisin.uio.no To: freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] t-value saved values Date: Thu, 20 Sep 2007 15:00:44 +0200
Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_________________________________________________________________ A place for moms to take a break! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us
The values in F.mgh are F dist under the null. But you should only have an F there if it is a multivariate contrast (ie, the contrast matrix has more than one row).
doug
Don Hagler wrote:
Are you sure that it is saving t-values in F.mgh and not F-stats?
From: Glenn Lawyer d.g.lawyer@medisin.uio.no To: freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] t-value saved values Date: Thu, 20 Sep 2007 15:00:44 +0200
Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
A place for moms to take a break! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
No, not at all sure. That is why I am asking. However, mri_glmfit says that the values should be t-values if there is only one row in the contrast matrix, which makes sense mathematically: "The F-ratio for the contrast is then given by:
F = G'*inv(C*inv(X'W'*W*X))*C')*G/(J*rvar)
The F is then used to compute a p-value. Note that when J=1, this reduces to a two-tailed t-test."
In my test data, J does equal 1.
Don Hagler wrote:
Are you sure that it is saving t-values in F.mgh and not F-stats?
From: Glenn Lawyer d.g.lawyer@medisin.uio.no To: freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] t-value saved values Date: Thu, 20 Sep 2007 15:00:44 +0200
Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
A place for moms to take a break! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us
But while an F with J=1 gives the same p value as an unsighed t, the actual statistic will be F = t^2. Could this account for the discrepancy?
Glenn Lawyer wrote:
No, not at all sure. That is why I am asking. However, mri_glmfit says that the values should be t-values if there is only one row in the contrast matrix, which makes sense mathematically: "The F-ratio for the contrast is then given by:
F = G'*inv(C*inv(X'W'*W*X))*C')*G/(J*rvar)
The F is then used to compute a p-value. Note that when J=1, this reduces to a two-tailed t-test."
In my test data, J does equal 1.
Don Hagler wrote:
Are you sure that it is saving t-values in F.mgh and not F-stats?
From: Glenn Lawyer d.g.lawyer@medisin.uio.no To: freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] t-value saved values Date: Thu, 20 Sep 2007 15:00:44 +0200
Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
A place for moms to take a break! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us
--
CARPE NOCTEMYou can sleep when you're dead
- Glenn Lawyer +
- +352 661 967 244 +
- Instituttgruppe for psykiatri +
- Seksjon Vinderen +
- PB 85 Vinderen +
- 0319 Oslo +
- http://folk.uio.no/davidgl +
+++++++++++++++++++++++++++++++++++:)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Yes.
Doug Greve wrote:
But while an F with J=1 gives the same p value as an unsighed t, the actual statistic will be F = t^2. Could this account for the discrepancy?
Glenn Lawyer wrote:
No, not at all sure. That is why I am asking. However, mri_glmfit says that the values should be t-values if there is only one row in the contrast matrix: "The F-ratio for the contrast is then given by:
F = G'*inv(C*inv(X'W'*W*X))*C')*G/(J*rvar)
The F is then used to compute a p-value. Note that when J=1, this reduces to a two-tailed t-test."
In my test data, J does equal 1.
Don Hagler wrote:
Are you sure that it is saving t-values in F.mgh and not F-stats?
From: Glenn Lawyer d.g.lawyer@medisin.uio.no mailto:d.g.lawyer@medisin.uio.no To: freesurfer@nmr.mgh.harvard.edu mailto:freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] t-value saved values Date: Thu, 20 Sep 2007 15:00:44 +0200
Hi,
I asked about this last week, and had a nice discussion with Doug, but somehow in that we never got the answer to my original question.
When FreeSurfer saves t-values (in F.mgh), what is it actually saving? The numbers in the file do not have a t-distribution (even when the source data for the glm is random noise). Is a transform applied to the t-values before the write, as is done with the p-values?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
A place for moms to take a break! http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us
--
CARPE NOCTEMYou can sleep when you're dead
- Glenn Lawyer +
- +352 661 967 244 +
- Instituttgruppe for psykiatri +
- Seksjon Vinderen +
- PB 85 Vinderen +
- 0319 Oslo +
- http://folk.uio.no/davidgl +
+++++++++++++++++++++++++++++++++++:)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu mailto: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 mailto:greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
In order to help us help you, please follow the steps in: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
freesurfer@nmr.mgh.harvard.edu
-
Don Hagler -
Doug Greve -
Glenn Lawyer