[Eeglablist] ANOVA permutation test across all channels

[Makoto Miyakoshi]

Dear Myriam,

GREAT question! I also looked into this question myself some time ago.
Because I could not remember the answer off the top of the head, I checked
code.
So, if you read statcond() line 412-423, you can find the trick. It looks
like you perform ANOVA with surrogate data (i.e., shuffle subjects across
conditions--this represents the null hypothesis that the three or more
groups of data come from the same distribution hence no mean difference)
for let's say 5,000 times. You store these 5,000 F statistics to make a
distribution, against which you test which percentile your actual F
statistics falls within. I did not run this process this time so I can't
give you absolute guarantee, but 95% I'm sure this is how it works. This
explains why you can apply 'permutation' (which is one of the methods for
generating surrogate data to perform non-parametric test).

> Am I doing something wrong or is it possible to use permutation based
ANOVAs?

You are doing fine. But you may want to explain to your reviewer how it
works, since I don't hear this approach very often though it seems totally
sound.

Makoto

On Fri, Jan 25, 2019 at 5:35 PM Myriam Taga <myriamtaga at msn.com> wrote:

> Dear all,
>
> I have an experiment with 5 different conditions in which I want to
> compare ERP amplitudes at 100 ms across all channels.
> I have used permutation based ANOVAs using EEGLAB.
>
> I got a question from a reviewer stating that he is not familiar with this
> approach and not even sure if permutation statistics can be used with ANOVA
> statistics and not only paired comparisons.
>
> Am I doing something wrong or is it possible to use permutation based
> ANOVAs?
>
> Kind Regards,
>
> Myriam
> _______________________________________________
> 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



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20190125/bb87f6a4/attachment.html>