Re: [Faster-eeg-list] outlier analysis

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
 <meta content="text/html; charset=ISO-8859-1"
 http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Hi Brian, 
 
My feeling would be to look for outliers separately in each group, as
you would do for ERP peak analysis because amplitude range &amp;
variation may be different between groups. We have run FASTER on data
from multiple sclerosis patients. The MS patients' data have low
amplitude relative to controls. If MS and control data were mixed then
an outlier in the MS groups (e.g., someone with a large amplitude
range, relative to other MS patients) would not get detected because it
would be masked by the control data. 
 
It is possible that some parameters, such as EOG data, would not differ
across groups. In that case all data could be combined. However, that
presupposes that EOG does not differ across groups (perhaps it could
because of medication effects, etc.). 
 
I think the safer approach would be treat the groups as separate for
the purposes of artifact rejection on a group level. 
 
All the best, 
Rob 
 
On 15/11/2010 19:09, Brian Roach wrote:
<blockquote cite="mid:4CE...@nc..." type="cite">
 <meta http-equiv="content-type"
 content="text/html; charset=ISO-8859-1">
 Hi All, 
 
We're using FASTER to analyze data from control participants and
patients with schizophrenia.&nbsp; We have two auditory N1 conditions.&nbsp; I
read about the application of FASTER to mmn data from Parkinson's
subjects.&nbsp; I am interested in recommendations for running the grand
average outlier test in multi-condition, multi-group studies.&nbsp; Would
you put all subjects and conditions in one test, run a separate test
for each condition, or run separate tests for every condition and
group?&nbsp; In a normal ERP peak analysis, we would look for outliers
separately in each group, but in that case we expect group
differences.&nbsp; With measures such as amplitude range, variance, and
channel deviation, I don't really know what to expect yet.&nbsp; I am
interested in what others think or have tried. 
 
Thanks, 
Brian 
 
</blockquote>
 
<pre class="moz-signature" cols="72">-- 
Robert Whelan, PhD
Senior Research Scientist

Trinity Centre for Bioengineering
Trinity College Dublin

Department of Neurology
St. Vincent's University Hospital
Elm Park, Dublin 4

webpage: <a class="moz-txt-link-freetext" href="http://www.mee.tcd.ie/~neuraleng/People/Robert">http://www.mee.tcd.ie/~neuraleng/People/Robert</a>

</pre>
</body>
</html>