Improving the Validity of Crucial Decisions in Pattern Analytic Methods

THE concept of reciprocity is fundamental to many numerical methods of hierarchical classification (McQuitty, 1967). Reciprocity is a measure of internal consistency which is applied to pairs of individuals or obj ects to determine whether or not the individuals or objects qualify as members of the same statistical type. In a matrix of interassociations between individuals, a, b, c,-, i and j are reciprocal if and only if i is highest with j and j is in turn highest with i (McQuitty, 1957). We have used other concepts in referring to the internal consistency of larger categories of individuals or objects and in determining whether or not these individuals or objects all qualify as members of the same statistical type. What is said in this paper with respect to reciprocity of pairs applies also to the internal consistency of larger categories.