[Eeglablist] Average the F and p-values?

[Stephen Politzer-Ahles]

Hi David,

If you use cluster-based permutation statistics, you will get a single
p-value for each (channel,timepoint) cluster that the algorithm identifies,
and you can report those p-values directly (by saying something like "there
was a significant negative cluster (p<.##) of centro-parietal channels from
X ms to Y ms"). I only have experience doing this in Fieldtrip, but it has
also been implemented in the latest version of EEGLAB so it should be
possible. You can also do something similar with the Mass Univariate
Toolbox (an add-on for EEGLAB); both these methods correct for multiple
comparisons.

Otherwise, you can use some of the methods discussed in the statistics
chapter of Luck (2005). What is commonly done is to group electrodes into
several 'regions' a priori (for suggestions see Dien's chapter on ANOVA in
high-density datasets in Handy, 2004) and average across electrodes within
a region, and also to average across time periods within a time window of
interest (which is typically identified either a priori based on your
hypothesis, or is chosen based on visual inspection of your waveforms).
Then you perform your stats on those mean voltages. If you try a large
number of time windows and/or electrode regions you should also take care
to adjust your alpha-level to account for multiple comparisons.

I haven't checked the math myself, but I don't think it would be correct to
average F- or p-values across electrodes or timepoints.

Best,
Steve

On Mon, Feb 11, 2013 at 5:12 AM, david grahms <david.grahms at gmail.com>wrote:

> Hello list,
>
> I have a question about reporting statistics derived from EEGLAB, or
> rather how to reduce and report relevant statistics. One of the many
> strengths of EEGLAB is the possibility to perform statistical testing over
> all your electrodes and samples when performing a simple ERP experiment. I
> do, however have trouble understanding what to do with huge amount of F and
> p-values once I have performed the testing.
>
> One problem is deciding which electrodes to include when reporting the
> statistical significance another is, if I choose one (or a cluster of
> relevant electrodes), do I still have to average relevant time period of
> that electrode and test it again, or can I average the F and p-values
> within that period for that specific electrode (or cluster)?
>
> Best regards,
> David
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>



-- 
Stephen Politzer-Ahles
University of Kansas
Linguistics Department
http://people.ku.edu/~sjpa/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130211/3e299d9c/attachment.html>