Analysis of circular data - cont.

Doherty Brian J brian at cc.brynmawr.edu
Wed Feb 7 09:18:31 EST 1996


Thnaks to all who have responded to my post about the analysis of
circular data.  It is nice to find out that there is some shareware
software (called Oriana) out there that will do much of what I need 
to do.

However, having briefly looked at that program - it does not appear to 
do the type of analysis with which I am having the most trouble - 
what Zar (1984) calls the parametric two-sample second-order analysis
of angles.

This test involves comparing two groups in terms of their respective
second-order mean angles.

For instance - we have phases of responses from 2 hemispheres (of the
brain) and are comparing two groups in terms of the second-order mean
(i.e. collapsing across hemisphere).  Thus, what is done is to calculate
the mean angle across the 2 hemispheres for each subject and then to
use these calculations to compare the two groups.

This is the analysis from which I am getting a small but disturbingly
_negative_ F value.  The fact that the F is negative makes me wonder
if I am miscalculating somewhere.

I am trying to get access to the 1993 Fisher book Statistical Analysis
of Circular Data, but does anyone know if this book contains a fully
worked out example of this type of analysis?  Zar only has examples
from after the point in which the second-order mean angles have been
calculated (i.e. when X, Y and r for each is known).

Any help would be appreciated.

Thanks,

Brian
--
brian at brynmawr.edu





More information about the Comp-bio mailing list