THE HOLISTIC APPROACH TO CONSTRUCT VALIDATION IN IS RESEARCH: EXAMPLES OF THE INTERPLAY BETWEEN THEORY AND MEASUREMENT 1

The paper presents two empirical examples of how the reliability and validity of a set of measures can change contingent on the theoretical model (or lack thereof) in which they are to be applied. In particular, measures developed atheoretically where the constructs are tested in separate components analysis are shown not to work well in the context of a theoretical model. Furthermore, measures that are demonstrated to work well in one theoretical model are shown not to generalize to another context. Finally, the multi-dimensionality of a set of measures can often be masked using first generation analytic techniques whereas second generation methods such as Partial Least Squares can provide additional information. It is suggested that researchers consider using a "one step" approach for theory and measurement development. At a minimum, this paper recommends that final validation of the psychometric properties of a set of measures for a construct should be made in the context of the theory in which they are to be used.