The orthogonally and the obliquely transformed component analysis in socimetric data.

概要

論文の詳細を見る
This article presents a new matrix method of analysis of sociometric data, which is more reliable and more efficient than the conventional methods in obtaining full informations relevant to the subgroup structure underlying the sociometric data. The method extracts the orthogonally and the obliquely transformed components, using the procrustean factor rotation in multiple factor analysis. First, a hypothetical target matrix is constructed by principal component analysis of the mutual choice data matrix, and then to approximate the target matrix the orthogonally and obliquely transformed components are obtained based on the choice and received choice data matrices. A tentative application to the results of a sociometric test given to junior high school students suggests that the method is adequate for extracting full informations from both types of sociometric data matrices, and for revealing the salient subgroups hidden in the matrices.
公益社団法人日本心理学会の論文