명목척도를 갖는 변수의 축약방법에 관한 연구

Reduction of Nominal Variables Using Factor Analysis Model

In this article, a reduction method for nominal variables is presented and its use illustrated. Factor analysis model (FAM) generally enables us to reduce variables having interval or ratio scale based on their correlation coefficients. We developed an extensive method that makes FAM applicative to the case of nominal variables which does not give correlation coefficients, but only the degree of association. Cramer's V coefficient is a well-established measure that provides the strength of association for nominal variables with a range of [0,1]. When Cramer's V coefficient can logically substitute for correlation coefficient, FAM would be extensively used for reduction of nominal variables.