ASSESSING CONSTRUCT-VALIDITY IN PERSONALITY-RESEARCH - APPLICATIONS TO MEASURES OF SELF-ESTEEM

This article addresses the assessment of convergent and discriminant validity in personality research. Four approaches are compared and contrasted for the analysis of classic multitrait-multimethod data, where three or more traits are measured with indicators derived from three or more methods. The approaches are the Campbell and Fiske criteria, the confirmatory factor analysis model, the correlated uniqueness model, and the direct product model. Pros and cons of the approaches are pointed out through a reanalysis of data originally collected by Van Tuinen and Ramanaiah, where global self-esteem, social self-esteem, and orderliness were each measured by true-false inventories, multipoint inventories, and simple self-reports. The Discussion considers guidelines for choosing among the approaches and further addresses procedures for nontraditional multitrait-multimethod data. © 1993 Academic Press. All rights reserved.