cluster-based permutation tests on single subjects

[Guillaume Rousselet]

Hey Tim,

trimming may improve the power of your statistics. However, after  
trimming, the degrees of freedom and the standard error of the T  
statistics need to be adjusted. This is because after trimming the  
remaining data points are no longer independent. The standard error of  
the trimmed mean is related to the winsorised variance, not the  
variance. The Yuen t-test on trimmed means is described here:

Wilcox, R. R. (2005). Introduction to Robust Estimation and Hypothesis  
Testing (2nd Ed. ed.): Academic Press.
Wilcox, R. R., & Keselman, H. J. (2003). Modern robust data analysis  
methods: measures of central tendency. Psychol Methods, 8(3), 254-274.

Wilcox has validated a  bootstrap trimmed mean t-test technique in  
which [1] the data are centred so that each condition has a trimmed  
mean of zero, [2] trials are sampled with replacement from each  
condition independently, [3] a t-test to compare trimmed means is  
computed, [4] the t-test obtained for the original data is compared to  
the distribution of t values obtained under the null hypothesis.

Best,

GAR
Note

On 6 Jan 2010, at 01:18, Tim Curran wrote:

> I have a few related questions about Maris and Oostenveld's (2007)  
> method for cluster-based permutation tests, in particular using an  
> indepsamplesT statistic in a between-trials design on single subjects.
> 1. Is it problematic to have very different numbers of trials within  
> each condition?
> 2.  Any advice for calculating statistical power for different  
> numbers of trials within each condition?
> 3. Might it be advisable to somehow trim the trials (e.g.,  
> Leonowicz, Karvanen, & Shishkin, 2005) to remove outliers prior to  
> analysis?
>
> Any advice would be much appreciated.
> thanks
> Tim
> ----------------------------------
> The aim of this list is to facilitate the discussion between users  
> of the FieldTrip  toolbox, to share experiences and to discuss new  
> ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html 
>  and http://www.ru.nl/neuroimaging/fieldtrip.





************************************************************************************
Guillaume A. Rousselet, Ph.D.

Lecturer, associate editor for Frontiers in Perception Science

Centre for Cognitive Neuroimaging (CCNi)
Department of Psychology
Faculty of Information & Mathematical Sciences (FIMS)
University of Glasgow
58 Hillhead Street
Glasgow, UK
G12 8QB

The University of Glasgow, charity number SC004401

http://web.me.com/rousseg/GARs_website/

Email: g.rousselet at psy.gla.ac.uk
Fax. +44 (0)141 330 4606
Tel. +44 (0)141 330 6652
Cell +44 (0)791 779 7833


“Computers in the future may weigh no more than 1.5 tons.”
	
	Popular Mechanics, 1949
************************************************************************************




----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20100106/8314eeeb/attachment-0002.html>