This project aims to understand the inter-related dynamics of school composition, neighbourhood composition and school attainment. We will use Echenique and Fryer’s network-based index of segregation. Little is known about changes in school and neighbourhood composition from large-scale representative data, or the drivers for change. By 2012, 10 years of the pupil-level census PLASC will be available for this analysis. We will ask how the patterns of families moving through the school age years changes school and neighbourhood segregation and composition. We will model specific cohorts as they pass through compulsory schooling and track the evolution of school and neighbourhood composition in given areas.
The statistical framework for measuring and modeling segregation is surprisingly underdeveloped, particularly given the importance of segregation in a number of academic and policy debates. Our previous work set up such a framework based around the concept of an assignment mechanism. We plan to extend this work, focusing on the issues involved in dynamic modeling and understanding changes or differences in segregation.
Using custom-written software to model the ‘core catchments’ of schools, the objective of this project is to look at the spatial configurations of schools’ admissions spaces and to identify the local markets within which the schools compete. This geographical knowledge can then be built into models of ethnic segregation and social polarization. In addition, working with the Centre for Multilevel modeling at Bristol, it can be used to assess the evidence for school competition affecting pupils’ learning progress.
Our previous research has demonstrated the link between the nature of school assignment and the social segregation of schools and neighbourhoods. Our understanding of this process can be greatly enhanced by studying it given different assignment rules in different, through comparable, countries. Whilst we will look into a role for the OECDwide databases such as PISA, it seems likely to be more fruitful to focus in depth on a small number of countries with datasets similar to the PLASC/NPD we have used in the UK. Given a head start through our existing contacts, we will seek out such databases and make new international links to facilitate this research. The most promising avenues are Norway, Sweden, the Netherlands and some states in the US.