Conference Program

A considerable body of scientific research has taken shape in recent years that examines how schools influence educational outcomes, against the backdrop of residential segregation, social inequalities, family preferences and individual attributes. This research explores the effects of socio-economic composition, ethnic composition, endogenous peer influence, school climate, the characteristics of teachers and other factors that are believed to influence educational outcomes. This field of inquiry often uses state-of-the-art methodological tools and statistical models to identify aggregate-level and interaction effects, yielding scientific advances with policy implications. The study of school/contextual effects has also become a fruitful area of multidisciplinary research, with debates taking place between sociologists, economists, statisticians, geographers, demographers and education researchers. It is arguably an appropriate time to take stock of these developments and to summarise the results of this research. In this paper, we present the results of a systematic literature review which synthesises the scientific literature of the past decade, with reference to schools in Europe, USA, Canada and Australia. This enables us to go considerably beyond all existing reviews, which are much more limited in scale and scope. Eligibility was confined to studies examining the relationship between the socio-economic or the ethnic composition of schools or classrooms, on the one hand, and either academic performance or choice of upper secondary school, on the other. A four-concept search strategy was used to search Scopus, Web of Science, ProQuest, EBSCO, Google Scholar and Semantic Scholar, including both quantitative and qualitative studies. The first concept identifies the context (secondary schools), while the second confines the search to studies which focus on school composition; the third describes the specific form of composition (socio-economic or ethnic) and the fourth captures the main outcome variable (performance or choice of upper secondary school). The resulting 12,462 records were subjected to rapid screening in order to exclude obviously ineligible studies, and the remaining 6,519 publications were blind screened in parallel on title and abstract by at least two researchers using the Rayyan software system, and any relevant documents were then screened on their full text. Disagreements were resolved through discussion, based on a close examination of the records in question. We then carried out data extraction and constructed a data extraction sheet which summarises the key characteristics of the 232 studies identified as relevant. A narrative synthesis of the research results was completed, summarising the principal findings of the included studies using headings developed from the thematic analysis of eligible research reports.