Towards a new paradigm of measurement in marketing

In measuring latent variables, marketing research currently defines measurement as the assignment of numerals to objects. On this basis, marketing researchers utilize a multitude of avenues to measurement. However, the scientific concept of measurement requires the discovery of a specific structure in the data allowing for the inference of a quantitative latent variable. A review of the quite diverse approaches used to measure marketing constructs reveals serious limitations in terms of their suitability as measurement models. Adhering to a revised definition of measurement, this paper finds that the Rasch model is the most adequate available. An empirical example illustrates its application to a marketing scale. Furthermore, this study investigates how instructions about response speed and the direction of an agree-disagree response scale impact the fit of the data to the Rasch model and to confirmatory factor analysis. The findings are diametrically opposed, with Rasch suggesting more plausible conclusions. Rasch favors well-considered responses and the agree-disagree scale, while factor analysis supports spontaneous responses and the disagree-agree format. The adoption of the Rasch model as the foundation of measurement in marketing promises to promote more advanced and substantial construct theories, is likely to deliver better-substantiated measures, and will enhance the crucial link between content and construct validity.