Many collective outcomes, such as residential segregation, the diffusion of ideas and social practices, the strength of cultural norms, and swings in public opinion, are not simply a reflection of the microlevel characteristics of the involved individuals. When individuals interact and influence one another, macro-level outcomes are also driven by the structure of social ties linking the individuals to one another, and the existence of complex tipping points. Such processes not only affect how behaviors spread in society, but also which behaviors become seen as valued, interesting, or otherwise desirable. Understanding how such social dynamics unfold is of considerable importance because they have the capacity to construct highly path-dependent realities that form the context of much of our day-to-day living.
Research at SweCSS analyzes the dynamics of social, cultural, and discursive influence using data on entire populations of interacting individuals, large text corpora capturing how people talk about the world they live in, and fine-grained digital trace data from online interactions. This research agenda combines powerful data-science methods—machine learning, natural language processing, and agent-based simulation—with precisely articulated social-science theories to improve our understanding of how individuals collectively and often unintentionally reshape the world we inhabit.
SweCSS seeks to establish reliable, valid, and efficient computational practices to improve our ability to assess causal relationships in the social world. Our key aim is to make CSS tools both more theoretically informed and more interpretable, so they can take center stage in social science inquiry, where explanation is of utmost importance.