Not Merely a Matter of Drawing Arrows: The Empirical Consequences of Measurement Model Specification

Abstract

Understanding measurement model specification is especially important for hospitality research due to its cross-disciplinary nature and the prevalence of measures used in the field which are often central to the formative versus reflective debate (e.g., SERVQUAL, socioeconomic status). The current study contributes to this topic by providing empirically based prescriptive advice to drive better measurement model specification. Specifically, the decision-making procedures developed by this study can complement theoretical reasons for a model choice as well as help determine a correct model choice when theories are equivocal or non-existent. This study combines actual and simulated data to show that model fit statistics alone cannot determine which model specification is correct, but also that a correct measurement model will generate more accurate predictions within a model which in turn will offer more accurate managerial recommendations.