Pulling the field out of a “One Variable, One Role” mindset: maximizing the theoretical value of interaction terms in communication’s mediation models

Recent analytical work reveals the need to assess mediated interactions (independent variable-by-mediator multiplicative terms) in mediation models to ensure the proper reporting of indirect effects. Besides their analytical value, mediated interactions can aid theory development. This study adds a theoretical support structure to this emergent analytical imperative and provides a theory-driven decision tree for incorporating mediated interactions into communication models. More broadly, mediated interactions are used as a basis to encourage the field to move beyond a “one variable, one role” approach to model building. Monte Carlo simulations reflecting common communication research practices were constructed and 1,920,000 datasets were analyzed to reveal the relative upsides and minimal risk incurred from assessing mediated interactions. In addition, the analyses elucidate the downsides incurred from not exploring these relationships when they are present in a population. The implications of these findings for future research and theory development are explored.