Original title: Modeling Network Dynamics with Overlapping Cohesive Subgroups
Stepan Zaretckii, Tom A.B. Snijders, Marijtje A.J. van Duijn, Christian Steglich · September 2026
People make friends and join social groups at the same time, but tracking how both change together is hard. To fix this, researchers built a new computer model. It treats social groups as hidden and evolving over time. In this model, people get frequent chances to update which groups they belong to. A built-in rule called membership cohesion pushes people to pick groups that match where their friends are, which keeps the groups and the friend network aligned. When testing the model on real student friendships, the researchers found that students are more likely to be friends with peers who share the same subgroup. Adding these shared groups also improved how well the model captured intermediate social structures.
You are organizing a study group for a class. You assign people to groups based on who they study with. But as the group works together, people become friends, and then they want to study together more. It is hard to plan because the groups and friendships keep influencing each other. Here, the model tracks both the friendships and the group memberships as they change over time, and it shows that people end up in groups with their friends more than you would expect just from knowing who studied with whom initially.
Understanding how friendships and group memberships change together matters for predicting social influence, the spread of ideas, and even epidemics. Schools and workplaces often try to shape groups to improve outcomes. This model gives researchers a way to track and test those dynamics in real networks.
The model is applied to only one friendship network, so generalizability is uncertain. It assumes group memberships are latent and that the cohesion effect is uniform. It does not address whether membership changes cause tie changes or vice versa. External factors like school policies or demographics are not included.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Modeling Network Dynamics with Overlapping Cohesive Subgroups — Stepan Zaretckii, Tom A.B. Snijders, Marijtje A.J. van Duijn et al. — SocArXiv — doi:10.31235/osf.io/7ey4a_v1
Abstract licence: CC BY
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.