The Factorial Validity of Three Frequently Employed Self-Report Measures of Self-Concept

Abstract

For a total sample of 280 seventh-grade students with predominantly Spanish surnames from a lower middle-class suburban community of Los Angeles and for each of two associated samples of 116 girls and 164 boys, the intercorrelations of 10 empirically derived factor scales of the Barksdale Self-Esteem Test (BSET), of 6 empirically based factor scales of the Piers-Harris Children's Self Concept Scale (PHCSCS), and of 15 a priori factor scales of the Tennessee Self Concept Scale (TSCS) were factor analyzed. The principal factors analysis was followed by varimax rotation. The identification of six factor dimensions that were essentially invariant across the three groups (Sense of Effective Agency; Integration of Moral and Ethical Behavior; Integration of Self-Reliant Behavior; Personal and Social Inadequacy Accompanied by Self-Depreciation; Projection, and Hostility; Personal/Social Satisfaction; and Personal/Social Self-Confidence) gave support to the multidimensionality of the self-concept construct. Evidence of convergent validity of the three self-report measures appeared to be lacking, as most of the test score variance was unique to each measure (method). The essential specificity of variance to each of the three measures (or the subscales of each test), which was apparent from the relatively low intercorrelations of the three tests as well as of their subscales, could not point to an inference of the presence of discriminant validity, as an alternative interpretation of the existence of methods variance associated with the properties of a given instrument (i.e., item format, response sets, or conditions of administration unique to each test) would appear quite plausible.