The Swedish education system underwent several phases of de-regulation and liberalisation since the early 1990s. Starting out with almost exclusively state-run schools with geographically defined catchment areas, many independent (i.e., private and for-profit) yet 100% municipality-funded schools have opened, and parents’ school choice today is more constrained by resources (information bias, travel distance) than by governmental administrative regulations.
Working with yearly individual level population data from Statistics Sweden (“the register”) over the period 1990-2012 and taking a network approach, we plan to study the impact of liberalisation on the educational sector. One outcome to study will be the national labour market mobility of secondary school teachers after the introduction of publicly funded voucher schools. Another will be the birth and death of schools in the population.
Like other “big” data, the register is a treasure chest with a curse upon it. On the one hand, it is huge and complete: we will not have noteworthy missing data or power problems to deal with. On the other hand, it is not very deep: we have to work with those variables that are available, which all have been collected for other, in this case mainly administrative purposes. Administrative purpose and sociological inquiry are, luckily, often not too far apart. Register data also allows us to follow life events of individuals over a long period of time, and, perhaps more importantly, we will have panel data on all teachers, students, and all schools which allows us to examine the dynamic feedback processes between labor market decisions of teachers and school compositions.
The nodal entities that can be involved in our analyses have a clear multilevel structure: students and their parents, teachers, schools, school boards, municipalities, workplaces, et cetera. The ties between them are of a straightforward affiliation nature only, as we typically do not have primary one-mode data on any node set. Each school belongs to a municipality, by its geographical location, and it receives its main funding from the municipalities that its student reside in (independent schools are also allowed to additionally accept private donations). Schools are not allowed to discriminate or require admission examinations, and we know for all students which school they went to. We know for all teachers which school they worked in, in addition to their personal educational histories. We also know who sat on the board of which school. Last but not least, we know workplaces and employers of the students’ parents. One-mode data will be obtained by projection of such affiliations, whenever this makes sense. For example, the sequence of yearly networks mapping teacher affiliations to schools can be used to construct a sequence of valued networks mapping the labour market moves between schools.