Covariates of no interest?

[jonny]

I'm interested in using NPAIRS to select the subject-by-subject model
for motion covariates. Is this possible with the Java version of
PLSNPAIRS?

All the best,
Jonathan

[nchurchill]

Hello Jonathan,
I can confirm that NPAIRS may be used for selecting motion covariates;
we have already been using this method to examine the effects of
regressing motion parameters out of fMRI data. Just to clarify, do you
intend to
(1) test some pre-defined set of motion covariates (e.g. the output
from image registration), or
(2) estimate motion effects directly from the NPAIRS output (e.g. in
data-driven fashion)?

Nathan

[Jonathan Ó Muircheartaigh]

Hi Nathan - thanks for getting back to me!

I'm looking to test option 1, is this possible with the Java implementation?

--
You received this message because you are subscribed to the Google Groups "plsnpairs" group.
To post to this group, send email to plsn...@googlegroups.com.
To unsubscribe from this group, send email to plsnpairs+...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/plsnpairs?hl=en.

[nchurchill]

Not a problem - the Java version of NPAIRS can definitely be used to
evaluate the effect of motion covariates on your results.
The most direct way to test their effects would be to run within-
subject analyses, for your different motion covariate models.
As part of the analysis, NPAIRS estimates (1) spatial reproducibility
and (2) temporal predictive accuracy of results, for each model. You
can then directly compare these metrics for the different models, and
choose the covariate set that optimizes reproducibility and/or
prediction of your results.
Hope this helps!

Nathan

as part of the analysis, NPAIRS outputs estimates of spatial
reproducibility and temporal prediction accuracy

On Sep 15, 3:42 pm, Jonathan Ó Muircheartaigh

<jonnymoria...@gmail.com> wrote:
> Hi Nathan - thanks for getting back to me!
>
> I'm looking to test option 1, is this possible with the Java implementation?
>

> On 15 September 2010 20:24, nchurchill <nchurchill.bac...@gmail.com> wrote:
>
> > Hello Jonathan,
> > I can confirm that NPAIRS may be used for selecting motion covariates;
> > we have already been using this method to examine the effects of
> > regressing motion parameters out of fMRI data. Just to clarify, do you
> > intend to
> > (1) test some pre-defined set of motion covariates (e.g. the output
> > from image registration), or
> > (2) estimate motion effects directly from the NPAIRS output (e.g. in
> > data-driven fashion)?
>
> > Nathan
>
> > On Sep 14, 10:44 am, jonny <jonnymoria...@gmail.com> wrote:
> > > I'm interested in using NPAIRS to select the subject-by-subject model
> > > for motion covariates. Is this possible with the Java version of
> > > PLSNPAIRS?
>
> > > All the best,
> > > Jonathan
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "plsnpairs" group.
> > To post to this group, send email to plsn...@googlegroups.com.
> > To unsubscribe from this group, send email to

> > plsnpairs+...@googlegroups.com<plsnpairs%2Bunsu...@googlegroups.com>
> > .