The field of marketing uses many statistically different methods for assessing discriminant validity. However, the discipline lacks an underlying standard as to which method(s) provides the most accurate and consistent results, especially given the varying conditions that exist within quantitative data and analysis. This lack of consistency creates the potential for inaccurate assessments of discriminant validity that either allows redundant constructs into the field or prevents the introduction of useful constructs, depending on whether discriminant validity truly exists or not. As such, it is likely that both redundant and missing constructs are present in our theories. This article seeks to help remedy this situation by comparing the effectiveness of six commonly used approaches for assessing the existence of discriminant validity.
Construct validity, in measures of constructs pertaining to human beings, was first examined by Campbell and Fiske (1959). Validation of these constructs requires assessing the validity of a measure (or set of measures) designed to accurately capture an opinion, trait, or tendency of a person or organization - such as a consumer's product category involvement (e.g., Zaichkowsky, 1985), an organization's market orientation (e.g., Jaworski & Kohli, 1993), or a buyer's commitment to a supplier-buyer relationship (e.g., Moorman & Podsakoff, 1992). As such, construct validity plays a pivotal role in the advancement of theory by verifying the extent to which a conceptual entity, or construct, can be accurately utilized in research.