The combined weights are whatever you tell the program. If you use 1s
and -1s, then your estimate will be N times as great as 1 run. I'd
recommend using 1/N and -1/N. This will create the average contrast
across runs. You could also decide to weight each run based on the
number of trials.
In FSL, the weighting is based on the variance - not the beta. It will
deweight betas with high variance estimates. It's only used with FLAME
and other models, not the OLS model.
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 Wed, Feb 20, 2013 at 8:14 AM, Glen Lee <[log in to unmask]> wrote:
> Dear SPM experts,
>
> For the 1st level GLM, I've been modeling each run separately as opposed to
> concatenating all the runs.
> Obviously I get more beta values that are generated by each run. However,
> when I create the contrast images (condition A vs. condition B), I then
> combine betas corresponding to the same condition across the entire runs.
>
> I've been assuming that this fixed analysis would be just as simple as an
> independent t-test for comparing betas in two different conditions without
> any other complicate procedures such as giving a greater weight to the
> higher beta value (I heard that this is the way how FSL does).
>
> If anybody knows how this works out, please help me.
>
> thanks in advance,
> Glen
>
>
>
>
>
>
>
|