Up to 17-18% of Estonian youth do not reach secondary education and are dropping out of studies (Eurostat, EHIS). High rate of people with low educational background will increase the risk of poverty and social exclusion, threatens competitiveness of workforce and thus lead to higher inequalities in societies. While school dropouts have been thoroughly studied in vocational education, its reasons are more uncharted at high school level (i.e. grades 10-12). However, at this level there are almost 3000 students each year, who either drop their studies or who are forced into year repetition. This is a challenge that can be addressed effectively and more systematically on school level through the use of big data.
To predict and prevent dropouts in a maintainable way, Praxis offers a two-part solution that combines big data analysis with latest knowledge of behavioural economics.
1. Big data analysis and creation of an early warning system that uses live data from study information systems (SIS). As SIS is are upgraded daily with grades, truancies and various comments from teachers, this thus offers an opportunity to create an IT solution working in real time (i.e. early warning system) that offers more insights and up to date information than regular questionnaires. As study analytics enables to find factors and patterns that are associated with dropout, early warning system uses that knowledge to help parents, teacher and support agents to notice students while these patterns occur. Data from SIS will be connected with registry data from EHIS.
2. Interventions based on behavioural sciences. To work out a targeted solution, we will apply behavioural economics, which integrate insights from psychological research into economic science, especially what regards to human judgment. This means we will use risk patterns emerging from data analysis to create interventions aimed at individual students and applied either by teachers or support agents with little or no cost. In other words – we will use nudging to influence behaviour. There are various examples in education, where nudging has already been applied to influence dropouts (see for example Dinkelman & Martínez 2014).
While both early warning systems and nudging have been used to influence dropouts, they have not yet been combined on a similar scale proposed here.
This will happen in nudging stage, where potential effect will be seen with randomized controlled trial (RCT). RCT is essentially the gold standard of scientific experiment, that aims to reduce sources of bias when testing the effectiveness of intervention.
We will use big data analysis combined with behavioural sciences to address dropouts at high school level. First year results will include extensive big data analysis, creation of early warning system and beginning of piloting the nudges. Second year will show the effectiveness of nudges, which will be continuously monitored and analysed. Based on first results on drop-out rates, we will either quickly expand intervention schools – thus helping more and more students – or adapt our intervention. While the sustainability of analytics will be ensured by study information systems, nudging will rely mostly on professional staff at schools, such as teachers, social pedagogue or psychologist. For future, we see the clear potential of adding registry data from Tax and Customs Board or Health Insurance Fund.