Weights

Frederick J. Peabody fpeabody at SUNFLOWR.USD.EDU
Thu Feb 27 11:58:28 CST 1997


Weighting involves attributing more "value" to certain characteristics
(morphological, genetic, etc.) than other characterictics when performing
a numerical analysis.  Justifications usually involve the notion that some
characterictics have a greater discriminatory ability for taxa in a
perticular group; or that they are "key" or "diagnostic" features.

I have always felt uncomfortable with character weighting.  By employing a
set of pre-conceived notions of character importance (an a priori
approach) one might arrive at a classification system that simple mimics
one's assumptions of the classification of the group.  The very process of
selecting certain characters over others to be included in the analysis
is, in itself, a form of character weighting.  In fact, one needs to be
very careful in selecting characters, or character states, so that
characteristics are not weighted more heavily than others in the analysis.
A good method to use to check if the characters selected are
inappropriately weighted is to perform a cluster analysis of the
characteristics, not the taxa, using an inverted data matrix.  This may
uncover any patterns of similarity among the characteristics, and may
alert the researcher to characteristics that are essentially measuring the
same attribute.

Frederick J. Peabody
Associate Professor of Botany
University of South Dakota
414 East Clark Street
Vermillion, SD  57069  USA
fpeabody at sundance.usd.edu

On Thu, 27 Feb 1997, Matt Davey wrote:

> Can anybody actually give me a brief definition/justification for
> reweighting/weighting morphological characters?
>
> Matt Davey
>




More information about the Taxacom mailing list