[Eeglablist] Tukey or Bonferroni ?

[Makoto Miyakoshi]

Dear Alexandre,

My thoughts:

   - If you have three items to compare, using LSD is most powerful and
   justified. Note that using LSD for comparison across more than 4 items is
   wrong (i.e., insufficient suppression for Type I error)
   - Bonferroni overcorrects. You'll suffer from Type II error (i.e.
   missing the true effect.) Use Bonferroni-Holm since it is slightly more
   reasonable.
   - Using FDR is preferable IF you have many significant results.
   - There are whole bunch of multiple comparison correction methods which
   I haven't used. It's not very exciting to learn them all because in most
   cases you'll end up with using one or two methods repeatedly.
   - For me maximum statistics to correct omnibus hypothesis is currently
   most reasonable and powerful for time-frequency data. See Groppe et al. and
   Mike X Cohen's handbook for this method. I used it in Miyakoshi et al. 2010
   which is based on Trujillo and Allen (2007). This method is however
   computationally more demanding and if you are not testing time-frequency
   data maybe it is overkill.

Makoto

On Tue, Feb 9, 2016 at 8:40 PM, Alexandre Obert <obert.alexandre at gmail.com>
wrote:

> Dear all,
>
> I've got a simple question but I failed to find paper responding to:
> Is there a reason to prefer one the two post-hoc pocedures: Tukey or
> Bonferroni when computing ANOVA on ERP data?
>
>
> Regards,
>
> Alexandre Obert
> _______________________________________________
> 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/20160322/1a4bd5d5/attachment.html>