Dear Lars,
There is no functionality to regress out some user-specified part of
the variance while averaging. The line you are referring to has to do
with weighting based on the original number of trials when 'averaging
the averages'. There is functionality called 'robust averaging' which
makes it possible to downweight outliers while averaging. This
functionality is accessible via the 'Artefact' button.
In addition if you convert your data to images trial-by-trial you can
do the same kinds of analysis as are done in fMRI so you can indeed
specify any custom regressors. Right now, however, there is no way to
convert images back to ERPs, although in principle it's doable and can
be done with some custom code. There might be a way to do it in the
future in SPM8, within a year or so.
Best,
Vladimir
On Mon, Dec 29, 2008 at 10:54 AM, Lars Kuchinke
<[log in to unmask]> wrote:
> Dear All,
>
> is there a possibility to compute weighted averages (or something similar)
> at the first level in the eeg analysis, e.g. to incorporate further
> regressors or to regress out the variance explained by a continous variable
> across trials like it is done in spm designs in the fmri analysis?
> I found the line
> [115] c = c./sum(c); % vector of trial-wise weights
> in spm_eeg_average and think this might be something I'm looking for, but
> I'm not sure if this is correct. Could anyone help?
>
> Thanks,
> Lars
>
> Dr. Lars Kuchinke
> Freie Universität Berlin
> Allgemeine Psychologie
> Habelschwerdter Allee 45
> 14195 Berlin
>
> [log in to unmask]
> www.fu-berlin.de/allgpsy
> www.fu-berlin.de/loe
>
|