FW: QDEC 1.2 displays different results depending on variable selection in the design tab
Dear developers,
I recently found that if you have two discrete variables in QDEC, then the order in which you select them in the design tab drastically effects results. It appears that the selection dictates the definition of the covariate.
In particular, regardless of the comparison you choose to view in the display tab (i.e. "Does the average thickness, accounting for X, differ between 1 & 2" OR "Does ..., accounting for Y, differ between 1 & 2") the results depend on what you click in the design tab rather than these definitions.
Is this correct? If so, then how do we verify that the contrast that we think we're testing is correct? Should we just not use QDEC 1.2 and set up the models manually?
Many thanks for your help.
best,
Dominic
PS. I suspect that this problem has been fixed in the new version, but it wouldn't work with our data which has been processed with FS4.5. On this note, is there a way that around this to use FS4.5 images with the new QDEC?
Hi all,
It seems that the post below is the case for both QDEC1.2 and 1.4, but only when you have two discrete variables with cases labeled as "1" and "2" in both.
Cheers,
Dominic
On 28/1/11 1:58 PM, "Dominic Dwyer" dwyerd@unimelb.edu.au wrote:
Dear developers,
I recently found that if you have two discrete variables in QDEC, then the order in which you select them in the design tab drastically effects results. It appears that the selection dictates the definition of the covariate.
In particular, regardless of the comparison you choose to view in the display tab (i.e. "Does the average thickness, accounting for X, differ between 1 & 2" OR "Does ..., accounting for Y, differ between 1 & 2") the results depend on what you click in the design tab rather than these definitions.
Is this correct? If so, then how do we verify that the contrast that we think we're testing is correct? Should we just not use QDEC 1.2 and set up the models manually?
Many thanks for your help.
best,
Dominic
PS. I suspect that this problem has been fixed in the new version, but it wouldn't work with our data which has been processed with FS4.5. On this note, is there a way that around this to use FS4.5 images with the new QDEC?
Does this mean that if you label them something else, the problem goes away?
doug
Dominic Dwyer wrote:
Hi all,
It seems that the post below is the case for both QDEC1.2 and 1.4, but only when you have two discrete variables with cases labeled as "1" and "2" in both.
Cheers,
Dominic
On 28/1/11 1:58 PM, "Dominic Dwyer" dwyerd@unimelb.edu.au wrote:
Dear developers,
I recently found that if you have two discrete variables in QDEC, then the order in which you select them in the design tab drastically effects results. It appears that the selection dictates the definition of the covariate.
In particular, regardless of the comparison you choose to view in the display tab (i.e. "Does the average thickness, accounting for X, differ between 1 & 2" OR "Does ..., accounting for Y, differ between 1 & 2") the results depend on what you click in the design tab rather than these definitions.
Is this correct? If so, then how do we verify that the contrast that we think we're testing is correct? Should we just not use QDEC 1.2 and set up the models manually?
Many thanks for your help.
best,
Dominic
PS. I suspect that this problem has been fixed in the new version, but it wouldn't work with our data which has been processed with FS4.5. On this note, is there a way that around this to use FS4.5 images with the new QDEC?
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Dominic Dwyer -
Douglas N Greve