Measuring Association Between Two Traits
Author: Gregorius H-R.
Source: Acta Biotheoretica, Volume 46, Number 2, 1998 , pp. 89-98(10)
A measure of association is introduced that is based on a conceptual rather than a model approach in order to ensure its broad applicability. The basis of the concept involves two variables or traits and of members of a population. The association of the -state with the -state is measured by the degree to which members of given -state share their -state. This formulation yields an index of association, which is applicable to all categories of traits, including discontinuous and continuous traits as well as combinations of these. Complete association of one trait with the other is equivalent to the existence of a functional relationship of the second to the first trait. Therefore, the degree of association can be understood as the closeness of the relations between two variables to a non-specified functional relationship. This feature in connection with the asymmetry of the index attests its suitability for cause-effect analyses. In fact, the conceptual approach to the measurement of association yields a conclusive method of detection and description of functional relationships between variables together with a method for quantification of the strictness of these relationships. The legitimacy of the correlation coefficient and of disequilibrium indices as measures of association is briefly addressed.
Document Type: Regular paper
Publication date: 1998-06-01