THIS WEEK, thousands of students will submit their Change of Mind form to the CAO. This time next year, one in six of those students will have dropped out of the course they chose.
Educations experts and business leaders have asked: are students prepared for college? Do they have enough information about their courses? Can they handle the freedom of third level?
What makes one student successful when another fails? So far we have approached the question using educational theory and psychology. However, we are not using a valuable resource: data.
Deepening our understanding or how students engage with education
Data analysis is widely used in business and increasingly in health and public service delivery. Learning is one domain where it has not (yet) been usefully leveraged. Learning analytics (the use of data to understand what factors influence learning) holds a lot of promise for deepening our understanding or how students engage with education.
College courses now feature many online components – from podcasts of lectures to online seminars. Every engagement by a student online creates valuable information for staff, if they are equipped to interpret it. That’s where data science comes in.
In the Open University, data on student usage of the system and student performance are used to alert lecturers when students need help.
A US company called Civitas Learning uses data analytics to recommend college courses to students based on the data collected about them during their school career, and data about similar kinds of students and the successful choices they make.
Learning analytics has huge potential to address some of the problems that Irish students encounter. It can help them to make better decisions and support them in this environment of independent learning.
Picking up on issues that might otherwise have been missed
The Insight Centre for Data Analytics applied advanced learning analytics to help individual students in their progression through first year undergraduate degrees. Last academic year 1,200 students in DCU signed up to PredictEd a weekly email alert service that predicts what their exam result will be, based on their level of engagement with course material online.
Students who opted to receive the alerts performed better in their exams, with a net increase of 3% in exam marks, based only on the weekly alerts.
The human interaction between student and tutor is irreplaceable, but with up to 500 students in some key first year modules, these individualised alerts supported the tutors’ work and picked up issues that might otherwise have been missed.
These weekly alerts are the equivalent of loyalty bonuses from a supermarket or offers from a mobile network provider. They are personalised, based on data from a large population, and ultimately benefit the user and the organisation.
There are crucial differences, however, between commercial data analytics and learning analytics in higher education. Unlike industry, data research in higher education institutions is conducted in a highly regulated environment. It is managed by ethics committees, who have a long tradition of defending the subject in research. Participants give informed consent and there is no consequence if someone wants to withdraw.
The value of learning analytics, from a student outcomes perspective
HEIs are using learning analytics already, but they are doing so largely at an institutional level, determining which feeder schools and media outlets to target for student recruitment advertising. However the value of learning analytics, from a student outcomes perspective, is starting to get attention. The National Teaching Forum has just announced a call for a large-scale collaboration among HEIs to deliver learning analytics programmes. It is important that this call results in activity that benefits not just the institutions or the individual lecturers, but also the students themselves.
If we can learn from the online behaviour of first year students, and use that information to help them over early speedbumps, it will spare at least some people the stress and expense of dropping out. In turn we must use non-completion data smartly to help the next crop of Leaving Certs to make better choices when the Change of Mind process rolls around again. If we harness learning analytics to benefit students, the story of first year drop-out rates could be different in another five years.
Professor Alan Smeaton is Director of the Insight Centre for Data Analytics at DCU.
To embed this post, copy the code below on your site
have your say