ABSTRACT
Although the consequences of incorrectly modelling formative latent variables as reflective constructs are serious, many researchers are still unaware of or reluctant to incorporate formative measurement in their models. The main reason may relate to the absence of a clear statistical rule to decide on the best causal relation between a latent variable and its indicators, as conceptual criteria can be ambiguous. This chapter follows the proposal of Bollen and Ting (2000) of using CTA as a statistical tool to unambiguously decide on the best configuration for latent variables. After presenting the CTA fundamentals, several empirical study cases are used to illustrate step -by-step its applicability in the marketing and management research field.
