[Eeglablist] Subjects number in ERP experiment

[Stephen Politzer-Ahles]

Hello Ahmed,

This depends on many factors, including the number of trials that each
participant contributes, the design of the experiment, and the nature of
the component under investigation. For example, a mismatch negativity
experiment---in which each participant [depending on the design] may
contribute hundreds of trials per condition, and the ERP component under
investigation is very robust, focal, and precisely time-locked to the
stimulus---is often possible with a pretty small number of participants
(there are many published papers with just around N=10 per experiment). On
the other hand, an experiment on certain sustained negativities in complex
language processing---in which each participant might only be getting 20-30
trials per condition because the stimuli are a pain to create, and in which
the component itself is often weak and jittery from one trial/participant
to the next---often requires more participants, maybe even 30+. The
recording environment may also matter.

Steve Luck's book (2005/2014) includes a section discussing signal-to-noise
ratio and the optimum number of subjects. It is probably possible to
conduct some kind of power analysis based on the strength of the signal you
expect to detect. But barring that, the best practice is always to review
the relevant literature that is similar to your planned experiment, and
make a decision about your N based on what is typical in related
experiments. (and, of course, once you have chosen a number, to stick to
that number, rather than peeking at the data as you go).

Best,
Steve



Stephen Politzer-Ahles
New York University, Abu Dhabi
Neuroscience of Language Lab
http://www.nyu.edu/projects/politzer-ahles/

On Sat, Jun 20, 2015 at 9:31 PM, Ahmed Almurshedi <ahmed19875000 at yahoo.com>
wrote:

> Dear All,
>
> I can I see i variety of ERP experiments some researcher used very wide
> range of participant may exceed 100 samples.
> While some others deals with limited number of samples which maybe around
> 12 subjects.
> My question is:
> How much the minimum size of subject is required in an ERP experiment?
> What is the criteria to choose the number of participants?
> If so how to calculate it statistically?
>
>  I have goggled it and find some suggest around 30-40 subjects. as in the
> following link
>  Sands Research - White Paper: Sample Size Analysis for Brainwave
> Collection (EEG) Methodologies
> <http://www.sandsresearch.com/WhitePaperSampleSize.aspx>
>
>
>
> [image: image] <http://www.sandsresearch.com/WhitePaperSampleSize.aspx>
>
>
>
>
>
> Sands Research - White Paper: Sample Size Analysis for B...
> <http://www.sandsresearch.com/WhitePaperSampleSize.aspx>
> View on www.sandsresearch...
> <http://www.sandsresearch.com/WhitePaperSampleSize.aspx>
> Preview by Yahoo
>
>
>
> Please share your comments..
>
>
> *Ahmed Almurshedi*
>
>
>
> _______________________________________________
> 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
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20150627/aeb6ca24/attachment.html>