Improving student retention

Research by academics in The Open University’s Knowledge Media Institute (KMi) has produced a software tool which can reduce student drop-out rates.

What this research does

When students fail courses or drop out, it is not only a major setback for the individual involved, but also costs universities money in lost fees and funding.

This has prompted Open University computer scientists to develop a software tool which can identify students at risk of failure at an early enough stage to intervene and put them back on track.

The OUAnalyse tool was initially developed for use with The Open University’s distance learning students, but researchers have now successfully created a customised version called StudentAnalyse, for use in traditional classroom-based universities.

For example, in a two-year trial at the Czech Technical University, student failure rates fell dramatically from 33.2 per cent to 16.71 per cent, and the number of failing students decreased by more than 49 per cent.

The university calculated that the number of students retained by the use of StudentAnalyse brought them an additional income of around £464,000 over the two-year trial.

The technology will be piloted in other Czech institutions, and further trials are planned in UK, US and Japanese universities.

We are now working to apply our analytical framework more widely in other university settings. One thing we hope to look at is how can we deploy the tool directly to students and piloting the personalised study recommender. This could change the way tutors communicate with students and also help already good students to become even better learners.

Dr Martin Hlosta
The Open University

The tool analyses demographic and individual student data, identifying critical points in student courses, known as ‘milestones’. At these milestones performance of successful and failing students begin to diverge.

StudentAnalyse prompts tutors to contact at-risk students earlier so that appropriate support can be provided before the students fall too far behind.

The system spots potential failure before human tutors can, because it has access to more data, which it analyses using a predictive algorithm.

OU scientists are also supporting further research by sharing their research data about student learning with the world – the only publicly available learning analytics dataset free to download and use for experiments.

Research publications

• OU Analyse: analysing at-risk students at The Open University
• Improving retention: predicting at-risk students by analysing clicking behaviour in a virtual learning environment
• Ouroboros: early identification of at-risk students without models based on legacy data
• Are we meeting a deadline? classification goal achievement in time in the presence of imbalanced data
• Implementing predictive learning analytics on a large scale: the teacher's perspective
• Open University Learning Analytics dataset

People

• Dr Martin Hlosta

• Follow:

  Martin Hlosta