As such, the unobserved variable is linked to one that is observable, thereby making its measurement possible” ( Byrne, 1998, p. Instead researchers “must operationally define the latent variable of interest in terms of behavior believed to represent it. Since they cannot be observed directly, those variables cannot be assessed directly either ( Jöreskog and Sörbom, 1979). They provide a degree of abstraction that permits us to generalize relationships ( Bollen, 2002). These concepts are theoretical in nature. For instance, social psychology deals with “attitudes” or “intention,” work and organizational psychology is often interested in “motivation” or “commitment,” and educational psychology tries to figure out the effect of “teacher expectancy” on “verbal ability,” to name just a few. In psychology, we are often interested in concepts that cannot be observed directly. A simulation study demonstrates the usefulness of this “manifest Fornell–Larcker criterion” in providing an easily assessable method for vetting existing instruments, whereas a systematic literature review shows the necessity to do so even for instruments published in well-received journals. Adapting the Fornell–Larcker criterion, we propose estimating distinctiveness between dimensions by using nothing but the most commonly reported statistics: Cronbach’s alpha and the correlation matrix between the manifest composite scores of the dimensions in question. This leaves researchers and practitioners who are interested in a certain instrument in a precarious situation: Before starting their own data collection, they do not know whether dimensions adequately discriminate from each other and thus whether they can have confidence in any interpretation of these dimensions. However, validity evidence supporting the proposed internal structure is often missing.
Multidimensional instruments posit several components that are each expected to be homogeneous but distinct from each other. The assessment of an instrument’s conceptual framework as prerequisite for conducting further analyses has been advocated for decades. 3Independent Researcher, Hamburg, Germany.1Psychology School, Fresenius University of Applied Sciences, Hamburg, Germany.Frederic Hilkenmeier 1*, Carla Bohndick 2, Thomas Bohndick 3 and Johanna Hilkenmeier 2