If you know that the activity is changing from baseline, then you
should model it if you can. Otherwise, the condition of interest will
be with respective to the constant+instruction. The problem can arise
though, that the instruction is too close to the condition of interest
or is not properly randomized, then it becomes hard to estimate the
effects of each.
For example, if the instruction is for 1s followed by your condition,
then its hard to separate the two events, especially if the
instruction is always the same before that type of condition. Its not
the orthogonality with the constant, its the lack of
orthogonality/collinearity of each of the condtions that becomes an
issue.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General
Hospital and Harvard Medical School
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, Feb 10, 2011 at 1:47 PM, maaike rive <[log in to unmask]> wrote:
> Hi all,
>
> I hope you can help me with a question regarding modelling regressors of no
> interest in a first level model.
> I assumed it is better to model all time intervals, including the periods of
> certain ''conditions'' you are not interested in (for example, an
> instruction period before a specific event), to reduce variance in the
> signal during the conditions that you will use for further analysis that is
> actually caused by conditions of no interest.
>
> However, now I was told that in doing so, one would reduce the degree
> of orthogonality of each regressor with respect to the constant regressor.
>
> Now I do not know what is the best option: model every condition or only the
> ones you will use for further analysis. I hope someone can advise me on
> this!
>
> Thanks,
>
> Maaike
>
>
>
|