Elementary Linkage Analysis for Isolating Orthogonal and Oblique Types and Typal Relevancies

Abstract

SEVERAL of the terms of the above title are analogous to key concepts in factor analysis; types correspond to factors, and typal relevancies correspond to factor loadings. These terms have similar meanings to the corresponding ones in Thurstonian factor analysis except for one fundamental difference ; the terms of elementary linkage analysis are here applied to a theory of psychological structure which is basically different from simple structure as developed by Thurstone (16, p. 181). However, data are sometimes so interrelated that the difference in theory does not express itself in empirical results; linkage analysis sometimes yields results very similar to, or even identical with, those of rotated factor-analytic solutions.Advantages of elementary linkage analysis are its provision for investigating a particular theoretical position, its speed, and its objectivity. A fifteen variable matrix can be analyzed into objectively determined types in five to ten minutes, all operations with only pencil and paper. The original solution of a linkage analysis gives the desired structure; no rotation is required. Need for the MethodElementary linkage analysis is a method of clustering. It can be used to cluster either people or items, or any objects, for that matter, which have distinctive cluster-characteristics.In order to make our discussion somewhat concrete, rather than abstract, we shall talk of cluster methods as procedures 1 This article was completed while the author was on the Faculty of the University of Illinois.

Identity change and the human dissection experience over the first year of medical training

“…To examine the relation between roles at different points in time, two non-parametric correlation analyses (Spearman's rho) were conducted on the scores obtained for the five roles in the identity grids at t1 and again at t2. The resulting two correlation matrices were then explored using McQuitty's (1957) elementary linkage analysis; a simple, objective clustering technique which provides a visual depiction of the strength of the relationships among roles. In this procedure, the roles are linked by arrows such that the role at the tail of an arrow has its largest correlation with the role at the head of the arrow.…”

The Methods and Problems of Cluster Analysis

“…This method, first described by Florek, Lukaszewica, Perkal, Steinhaus, and Zubrzycki (1951) and later by McQuitty (1957) and Sneath (1957), is the oldest of the hierarchical agglomerative methods which is still commonly used. All hierarchical agglomerative methods start the clustering process by searching the similarity matrix for the two most similar entities.…”

Critiquing the critics: an approach for the comparative evaluation of critical schemas

“…As stated earlier, the agglomerative hierarchical methods can be distinguished by the criterion used to assess similarity between clusters. Under single-linkage (McQuitty, 1957;Sneath, 1957), the similarity between clusters is measured by the distance between the closest pair of objects from each group. Average-linkage (Sokal and Michener, 1958), by contrast, measures similarity as the average distance between every pair of objects between each group.…”