Modeling Self-Selection Bias in Interactive-Communications Research

Interactive media use is a key issue in contemporary and future communication research. However, when users can interact with messages, new sources of variation emerge that make generalization difficult. This article introduces methods that communication researchers can use to increase the external validity of studies investigating the effects of interactivity versus noninteractivity. For this area of research, the article recommends the use of forced-interaction studies with random assignment to maximize internal validity, and free-interaction studies with self-selection bias modeling to maximize external validity. The article concludes with examples of self-selection bias modeling, using data from real studies, and a discussion of their implications for communication researchers.