NISTH Symposium: Modeling Social Complexity for Public Policy – the case of School Segregation

NISTH Think Out: Debate Series

Aug 16 2022 • 1 hr 5 mins

Complex Systems span a wide range of different application areas but exhibit common systemic behaviours that emerge through the interactions of simple elements. This systemic behaviour can only be understood by holistic analysis that necessitates viewing the system as a dynamic collective of individuals. In this talk I’ll explain and demonstrates ways in which complex systems models can be applied to reason about social systems, and how these models can be used to reason about policy.  In this talk I’ll focus on one area, that of school segregation. The issue of segregation in education can be (and has been) examined from both the individual level (e.g., parent surveys, choice analysis, etc.) or from macro-level statistics (e.g., changes in segregation level, region, city or national level). The uniqueness of a complexity science approach is the ability to connect these two levels and perhaps demonstrate that seemingly innocuous changes in individual behaviour or societal context can lead to drastic change in macro level dynamics. In this talk I will describe our approach to understanding segregation in the compass project (, working with the inspectorate of education and the city of Amsterdam, we are developing agent-based models to analyse the process of school segregation.


Michael Lees is an associate professor at the University of Amsterdam (UvA) where he leads the Computational Science Lab ( in the Informatics Institute. He is also a Principal Investigator for Complex Systems at the UvA Institute of Advanced Study (  His research is driven by scientific challenges in the area of social-urban complex systems. The fundamental challenge in this area is how to map social technical and natural phenomena into scalable and predictive computational models that can help develop and test interventions. In order to address this fundamental challenge my research aims to develop novel methods in agent-based modelling (modelling methodology) and discrete-event simulation (computation execution). This includes methods for semi-automatic model construction, modelling formalisms that are able to capture human behaviour and new ways to probe and measure social-urban systems to be able to validate and calibrate such models.