I can't tell you what FSL does; however, I can tell you that you
always want to include a constant term in the model.
In SPM, Y is not modified, but you can demean/remove the mean from X.
This changes the interpretation of the constant. See the following
page for more details: http://mumford.fmripower.org/mean_centering/
I am not sure why Y would need to have its mean removed, although it
might be something unique to randomise.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Postdoctoral Research Fellow, GRECC, Bedford VA
Website: http://www.martinos.org/~mclaren
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.
On Thu, Jan 10, 2013 at 6:22 AM, chenchunhuichina
<[log in to unmask]> wrote:
> Dear experts,
>
> I am confused by demean during GLM model setup and would like someone here
> can help me. Thanks!
>
> Suppose I have a continuous behavior data and I want to find out which brain
> region(s) correlated with it. The simple way is to calculate correltation
> between brain data and behavior data, or in GLM to set up a model Y=AX+B+e,
> A is slope, B is constent and e is error, Y is brain data and X is beahvior
> data. So this design matrix have two collumns: behavior data and ones.
> or we can think of demean that dX=X-mean(X), dY=Y-mean(Y)
> Y=A'dX+B'+e', A' should be the same as A and B' differed from B, and B'
> maybe more interpretatable.
>
> I was told that -D opion in FSL randomise will demean X and Y at the same
> time. one can set up models like dY=A1dX+B1+e1, or dY=A2dX+e2. I simulated a
> data in SPSS and confirmed that A1,A2 is the same as A, B1 is 0. but t test
> for A1 A2 differed a little, A1 is the same as above models.
>
> So it seems to me that include a constent in the model is always right,
> right?
> My confusion is why Y also need to be demeaned and how was it done in
> randomise? when one says "demean" or "mean centering", he/she means demean X
> only or demean X and Y both?
>
> Thanks for any clarification!
>
>
> 2013-01-10
> ________________________________
> Chunhui Chen
> _________________
>
> State Key Laboratory of Cognitive Neuroscience and Learning
> Beijing Normal University
> Beijing, China 100875
|