PhD Defense M.L. Meijerink MSc

Who, when, and how long? Time-sensitive social network modeling using relational event data

Social interactions between people play a central role in society, and understanding social interaction behavior is thus an important area of study in the social sciences. The Relational Event Model (REM) is a statistical tool that helps us examine the factors that motivate individuals in a social network to engage with each other and the timing of these interactions. An essential aspect of this model lies in its ability to consider the past interactions among individuals in the network, leading to a time-sensitive analysis. The primary question it addresses is how patterns that have emerged from previous interactions explain social interaction behavior and predict when the next interaction is likely to occur and who will be involved. 

This dissertation contributes to the study of social interaction dynamics using REM in several ways. Firstly, it offers a clear introduction to REM for psychologists, demonstrating its application in uncovering trends in social interaction behavior over time among university freshmen. Three key research questions are explored: What motivates students' social interaction behavior? How do interaction processes change as students get to know each other? How do these evolving processes influence interactions in different contexts? The main findings indicate that patterns of interaction develop early in the acquaintance process, which play a significant role in predicting future interaction behavior. 

Moreover, this work introduces two methodologies that enhance the REM toolkit. One extends REM to explore changes in social interaction behavior over time. Another extension allows us to examine the role of the duration of interactions in explaining future interaction behavior. The proposed methods are evaluated through simulations and applied to real-world cases, including interactions between employees, interactions within a healthcare setting, and interactions amid a violent conflict. These applications highlight how the proposed methods can be applied to deepen our understanding of how interaction patterns develop over time, aiming to gain insight into when the next interaction is likely to occur, who will be involved, and how long it will last. 

Finally, the dissertation includes two tutorials for using REM and testing scientific theories related to REM parameters in R. These tutorials offer step-by-step explanations and examples for researchers interested in applying REM to their own social interaction research. This allows researchers to more easily utilize REM and contribute to the further development of knowledge regarding the dynamics of social interaction behavior.