Dyadic Data Analysis: Current Issues and Directions

In this chapter, we provide an overview of modern methodological approaches to relationship science that is both practical and accessible. We start with a “status report” on the field, outlining three popular methodological trends. First, we discuss the application of machine learning techniques, specifically random forests, to the field of relationships science. Second, we elaborate on the importance of multimodal data, describing studies incorporating physiological, neurological, and linguistic measures. Third, we briefly discuss nonstandard dyadic designs, such as the round-robin design and network analysis, which enable the examination of multiple dyads within larger groups. Throughout these sections, we provide recommendations for scholars who wish to implement these methods in future work. Next, we provide a description of two analytical approaches used frequently in dyadic data analysis – structural equation modeling (SEM) and multilevel modeling (MLM) – and discuss the strengths and weaknesses of these approaches. We also present a brief discussion of differences between statistical software programs and suggestions for when to use each.