Structural choice modelling : theory and applications to combining choice experiments

We propose and describe a comprehensive theoretical framework that integrates choice models and structural equation models. Referred to as "structural choice modelling" the framework easily combines data from separate but related choice experiments. We describe the mathematical properties of the new framework, including goodness-of-fit and identification and we illustrate how to apply the framework with three empirical examples. The examples demonstrate new ways to evaluate choice processes and new ways to test substantive theory using choice experiments. We show how to combine choice experiments within the same model where there is a common research question, yet the designs and nature of the experiments differ. The seemingly simple notion of combining two or more choice tasks for the same people offers considerable potential to develop and test theory, as illustrated with the new framework. -- choice Models ; discrete choice experiments ; latent variables ; structural equation models