Teacher turnover as inter-school mobility networks
Marc Sarazin, Xavier DumayTeacher turnover has become an important, cross-national policy problem. It has been framed as an inability of schools and educational systems to retain teachers both in schools and in teaching, resulting in teacher shortages in schools, administrative burden for schools needing to hire new teachers, and disruptions in students’ learning. Reflecting this perspective, much research has looked at teacher and school characteristics that influence teachers’ decisions to leave schools. Key among these characteristics are teachers’ working conditions, understood to be influenced by accountability pressures, the student make-up of schools, and teachers’ social relations with their colleagues, a factor much explored in recent SNA research.
This previous research fundamentally sees teacher turnover as a phenomenon occurring between individual teachers and their schools: if teachers deem working conditions in their school to be unsatisfactory, then they will be more likely to leave that school. While this conventional perspective has borne many fruits, it ignores a basic feature of instances where teachers move schools, which constitute an important part of teacher turnover: that, in those cases, turnover is a cross-school mobility event, whereby teachers choose not only to leave their school but also to join another. The conventional perspective therefore not only tends to neglect features of ‘receiving’ schools that might influence teacher turnover. It also overlooks the possible influences of similarities between ‘sending’ and ‘receiving’ schools, alongside those of cross-school partnerships and previous mobility events, on teacher turnover. If these phenomena did indeed influence cross-school mobility events, then they would challenge the idea that teacher turnover is fundamentally due to teachers ‘fleeing’ certain kinds of schools.
The proposed paper will investigate cross-school mobility events using a unique census dataset of all teachers in state-funded secondary schools in London. England (and therefore London) is a particularly interesting case to investigate the importance of cross-school dynamics and partnerships, as it features both well established and newer, rapidly expanding multi-school partnerships that influence teachers' working conditions. The paper uses the census dataset to construct a network dataset of cross-school movements between these schools in 2010-2016, combining this with data on schools’ pupil composition, high-stakes inspection results, and membership of multi-school partnerships.
Through estimating Exponential Random Graph Models (ERGMs) and their longitudinal extensions, this paper will compare the importance of sender, receiver, homophily, multi-school partnership, and previous mobility effects in explaining cross-school mobility events. These effects represent different tendencies: the first for teachers to ‘flee’ certain school environments; the second for teachers to be ‘attracted’ by certain environments; the third and fourth for teachers to remain within particular kinds of school environments when they move; and the fifth for teachers to move where previous colleagues have moved.
Preliminary, descriptive findings suggest that not only do sender and receiver effects occur as predicted by previous literature, but that homophily effects also pertain to cross-school mobility events. They also suggest that multi-school partnerships powerfully affect teacher mobility. These findings suggest that existing frameworks, which foreground teachers’ tendency to ‘flee’ schools with undesirable working conditions, inadequately explain teacher turnover.