Computational modelling of school choice and school segregation From theoretical to data-driven agent-based models

Many educational systems are afflicted by substantial levels of school segregation along various lines, such as race, ethnicity, household income levels, parental educational attainment, and ability. This means that pupils with similar characteristics cluster together in the same schools, which is widely acknowledged to reproduce, and even exacerbate, inequalities and result in unequal outcomes. Despite decades of research and policies aimed at counteracting school segregation, it is still an important societal problem. This dissertation conveys that a complexity perspective on the dynamics of school choice can contribute meaningfully to the current scientific understanding of school segregation. By conceptualising the dynamics of school choice at the micro-level (e.g., parental, household), meso-level (e.g., school, neighbourhood) and macro-level contextual factors (e.g., segregation, institutional rules), as interdependent and interacting elements of a broader complex system. A notable tool for studying such systems are agent-based models (ABMs). ABMs can explicitly model interactions between households, factors in school choice, and how levels of school segregation feed back into choices of households and schools. However, ABMs are rarely used in this field of study and could potentially complement our scientific understanding of the mechanisms underlying school segregation. Therefore, this dissertation develops theoretical and stylised ABMs to enrich our existing theories and provide new explanations of which mechanisms affect school segregation. In addition, combined with the development of methods for empirical calibration of ABMs, data on specific social and institutional contexts are incorporated for more empirically realistic ABMs to provide validation and a tool for policy analysis.