Abtract

Experimentation is often called the strongest tool for defining causality in the proverbial scientific toolbelt. But, the claim of “causality” comes with the burden of many explicit and implicit experimental assumptions. In this research, I investigate the role of measurement invariance across experimental conditions – something implicitly assumed to hold, and as a consequence previously ignored in experimental research. When manipulating treatment X (e.g., watching funny versus neutral videos), that manipulation is assumed to cause a change in a relevant latent construct (e.g., happiness). This construct is then measured using some operationalized dependent variable (e.g., a 5-item happiness scale). If happiness in one condition is higher than in the other, we usually assume that to mean that the treatment caused happiness. That is true under the tacit assumption that the dependent variable is measuring the same construct in both conditions. Does, for example, a happiness scale measure the same latent factor of “happiness” after watching a funny video as it does after watching a neutral video? This may be the case in many instances, but need not always be true. In this talk I introduces the potential issues that occur if there is not measurement invariance across conditions and introduce some initial simulations for defining a test for measurement invariance across conditions. I will demonstrate the implications in the context of a recent hot topic in consumer research: the “experiential advantage” whereby consumers derive more “happiness” from consuming experiences compared to consuming material goods.

Please contact dip.mkt@unibocconi.it if you wish to attend.