The Convergence of the Results of Exploratory and Confirmatory Factor Analysis in the Latent Structure of a Standardized Affective Measure

Abstract

For a sample of 271 tenthand eleventh-grade students from middle-class families residing in a suburban setting, both exploratory and confirmatory factor analyses were carried out on a correlation matrix of 90 items and on a correlation matrix of 18 subtests, each consisting of five items. Both correlation matrixes were derived from an affective measure comprising six relatively homogeneous factor scales, each with 15 items. A comparison of the various orthogonal and oblique solutions for the item correlation matrix and for the subtest correlation matrix revealed comparable dimensions. Application of goodness-of-fit indexes in the confirmatory analyses revealed that the six-factor oblique solution for either the item or subtest correlation matrix accounted for the greatest amount of covariation in each matrix. It was concluded that in the instance of a multidimensional affective measure with homogeneous scales a high degree of convergence was realized in the solutions obtained by either exploratory or confirmatory factor analyses whether intercorrelations of items or of composites of items (subtests) were examined.