On Validating a Relationship Between Constructs

Publisher Website
Google Scholar
Abstract
An obtained relationship between two empirical variables may mean that (a) they both reflect the same theoretical construct or that (b) they reflect different constructs which are related. It is suggested that the demonstration of shifting, nonmonotonic, and/or asymmetric relationships provides evidence for rejecting hypothesis (a). Demonstrating that the variables are differentially related to a third variable, factor analysis, corrections for attenuation, and a multitrait-multimethod approach appear to be limited ways of generating information useful for choosing between hypotheses (a) and (b).