[FieldTrip] cluster>nothing to plot

[Stephen Politzer-Ahles]

Zita,

The sensitivity of the cluster test varies depending on the time window you
analyze. Specifically, the sensitivity is greater when testing a small time
window; see the discussion of cfg.latency at
http://fieldtrip.fcdonders.nl/tutorial/cluster_permutation_timelock#the_configuration_settings.
But to properly control the type I error rate, it seems the time
window of
analysis should be chosen based on a priori predictions about where the
effect should appear (i.e., not chosen based on your own data after you've
looked at it).

Best,
Steve Politzer-Ahles


Message: 5
> Date: Fri, 7 Oct 2011 10:18:01 +0100
> From: Zita Eva Patai <eva.patai at psy.ox.ac.uk>
> To: fieldtrip at donders.ru.nl
> Subject: [FieldTrip] cluster>nothing to plot
> Message-ID:
>        <CACPtmWPJAuVsRRJuKt55P9cMqxPc4LyoNw8f6gi-NPVUuLiUag at mail.gmail.com
> >
> Content-Type: text/plain; charset="iso-8859-1"
>
> Dear All
>
> I am running the Cluster-based permutation tests on event related fields,
> and for the same dataset, with all settings identical, when i run:
> 0-200ms: i get one significant cluster
> but:
> 0-500ms: no significant clusters
>
> why does my significant cluster disappear depending on the time window i
> use?
>
> thank you!
> z
>
>
> --
>
> Zita Patai
> DPhil Candidate, Experimental Psychology
> University of Oxford
> bcl.psy.ox.ac.uk/people/zita-eva-patai/
> eva.patai at psy.ox.ac.uk
>
-- 
Stephen Politzer-Ahles
University of Kansas
Linguistics Department
http://www.linguistics.ku.edu/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20111007/446faba5/attachment-0002.html>