Retention and other measures of student success can be positively influenced by the use of learning analytics to identify students at risk, combined with an effective intervention programme. These are indeed trying times for higher education (HE) providers.
A burning question still remains– how can universities make engagement more meaningful and improve student outcomes? The key to getting on the front foot of the student HE experience is data.
Ian Marshman, University of Melbourne suggests that data could set universities on the right path today to better plan for the future:
“There is now this new generation of systems, learning analytics and business analytics which certainly compel universities to move on and shape their strategy by informed decision making, based on information and data rather than leading solely by intuition.”
Inside the field of learning analytics
The field of learning analytics is undoubtedly one that is compelling for higher education, as institutions around the globe are increasingly challenged to meet rising expectations in a rapidly changing environment. Shifts to outcome-based assessment and reporting models are placing an increased responsibility on institutions to understand the patterns and trends that impact student engagement and ultimately, their success.
Presently, universities identify students who are at risk of dropping out of their course, or who are not making the desired academic progress based on a limited set of downstream indicators. While these indicators, such as academic results remain very important, universities can unlock powerful insights and improve student journeys further by looking at the data that reside within university services.
" Intervention strategies with struggling students could be extremely important for institutions: if you are identified as at risk but left alone you are not only considerably more likely to fail but your result is likely to be much worse too."
- Jon Baldwin, Managing Director Higher Education, Tribal Group
By tracking student interactions with university services, educators are able to predict student performance and identify potential “at-risk” students, so interventions can be made at an early stage to fully enable them to meet their potential.
Staff can access information about levels of engagement, to quickly identify students who are at risk of not completing their course successfully. It can also provide information that helps them to understand the student cohort and what factors could be changed to bring about improvements to the student experience and increase retention.
The University of Wolverhampton in the UK partnered with Tribal to develop the learning analytics software Student Insight, which was built using the data warehouse within the university.
The software uses a predictive model to gauge student academic success based on demographics, social background, summative assessments and activity data from the library, VLE (Virtual Learning Environment) and campus PC usage. The project determined that the model can predict student success with an accuracy of 70 percent.
Predictive modelling for intervention programmes
Tribal works alongside 50 institutions and multiple suppliers with JISC, the higher further education and skills sectors’ not-for-profit organisation for digital services and solutions. The effective learning analytics project is one of the largest consortium projects of its kind currently running.
In January 2017, JISC highlighted in its report studies in the US and Australia using control groups which show that retention and other measures of student success can be positively influenced by the use of learning analytics to identify students at risk, combined with an effective intervention programme.
"By tracking student interactions with university services, educators are able to predict student performance and identify potential “at-risk” students"
At New York Institute of Technology, recall of their predictive model is 74 percent; in other words, approximately three out of every four students who do not return to their studies the following year are predicted as at risk by the model. This high recall factor is reported to be due to the choice of model, careful testing of alternatives and the inclusion of a wider range of data than other similar models: financial and student survey data were included in the model as well as pre-enrollment data.
At the University of South Australia, 730 students across a range of courses were identified as at risk.
Of the 549 who were contacted, 66 percent passed with an average Grade Point Average (GPA) of 4.29. 52 percent of at-risk students who were not contacted passed with an average GPA of 3.14.
This appears to be a significant finding, implying that intervention strategies with struggling students could be extremely important for institutions: if you are identified as at risk but left alone you are not only considerably more likely to fail but your result is likely to be much worse too.
"Retention and other measures of student success can be positively influenced by the use of learning analytics to identify students at risk"
At the University of New England, New South Wales, the dropout rate was cut from 18 percent to 12 percent during early trials of their “Automated Wellness Engine”, which analyses data from seven corporate systems every night and runs a model based upon 34 triggers identified as suggesting at-risk behaviour.
While many institutional initiatives are still at an early stage, the studies to date and the work we have undertaken with some of these institutions certainly suggest huge opportunities from data to make student engagement more meaningful and successful. Analytics enable staff to better understand student needs and where they may end up.
All across the world, the challenge of widening successful participation for higher education remains. As local educators brace themselves for the start of many hopeful student journeys this year, the potential of data is worth considering in underpinning the success of those journeys.