An analytics approach to investigate teacher turnover

Abstract

Some educational problems embed spatial and temporal complexities, and the aggregation of these data may cause contextual information to be lost. One such example regards teacher turnover, which impacts directly the students' learning processes. In this work, we adopted an observational cross-sectional methodology, using visual analytics techniques to identify complex patterns in the mobility data of teachers in public schools from the city of São Paulo between 2016 and 2017. For this, we used education open data from the Brazilian government, which maps which teachers teach in which schools through a yearly school census. In addition, we sought to understand which are the main factors that, along with institutional rules, influence this sort of decision. To contextualize the main factors, we used synthetic indicators developed by the Brazilian government to identify different motivation clusters that may influence teachers' decisions to move to another school. As result, we identified different patterns varying according to their contract type and their respective geographical patterns. The clusters also identified as main factors: school performance, school climate, and management complexity.