RE: Character State Choice - We Have a Long Way to Go

[Mike Milbocker <mmilbocker@psdllc.com>]

David Peters wrote:

>I found in my cladistic analysis of the Pterosauria similar problems. Try
statistics. Take all the data and ratios, then place the figures on a
continuum. Let them fall, like pachinko pinballs, into natural groupings
into a jagged sort of bell curve, (usually). At the 'breaks' you can place
upper and lower limits on the various states.<

This is actually a good idea, but to make the choice more objective you
might calculate the standard deviations for the various clusters, and
require that any two clusters or potential states be separated by sigma, or
some multiple of it. If you do this for all the relevant characters, then
you can claim that there is no character state that overlaps another
character state by more than the sigma measure. That has true statistical
meaning.

Regards,

Mike Milbocker