<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<style type="text/css" style="display:none;"><!-- P {margin-top:0;margin-bottom:0;} --></style>
</head>
<body dir="ltr">
<div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;" dir="ltr">
<p style="margin-top:0;margin-bottom:0"></p>
<div>Subject: Clustering Statistics with Different Number of Observations Per Channel<br>
<p><br>
</p>
<p>Dear community, </p>
<p><br>
</p>
<p>My name is Dimitris Mylonas and I am working on coupled oscillations on EEG. Currently I am analyzing a high density EEG dataset where we measure the phase-amplitude coupling of specific oscillations (slow waves (.5-2 Hz) and sleep spindles (12-15 Hz)) at
each electrode, for two different groups. The outcome measure is the vector length of the coupling (coupling strength) and the question is whether the two groups show significant difference in coupling strength.
<br>
</p>
<p>In order to estimate reliably the coupling strength for each electrode/subject we need to have a minimum number N of coupled oscillations. If for a given subject and a given electrode there are less than N detected oscillations there will be a missing value
in the outcome measure. As a result each electrode has different number of subjects where the coupling strength can be "measured reliably".</p>
<p>My question is if I can implement the non-parametric statistical testing (not necessarily as implemented in Fieldtrip) (Maris and Oostenveld, 2007) to test for a group difference under this condition (different degrees of freedom at each electrode).
<br>
</p>
<p><br>
</p>
<p>Thank you,</p>
Dimitris </div>
<br>
<p></p>
</div>
</body>
</html>