You are here

2.482X10: Lessons from a Decade of use of Leaner Analytics

Mr Ed Campbell

University of New England

Background: UNE is a regional university with a large cohort (80%) of online learners of the total student cohort (n=22000). Attrition sits midway in the regional comparator group (22.6% in 2014) and the student cohort comprises a significant proportion of highly at-risk students (source: Higher Education Standards Panel Discussion Paper, June 2017). In 2007, the Student Engagement Team began a journey in big data in two key spaces: (1) to identify real-time student disengagement (Automated Wellness Engine); and (2) identify dissatisfied students (eMotion). Subsequent work was undertaken to leverage student sentiment (the Vibe) and discontinuation reasons (Unit Discontinuation Reporting).

Methods: To monitor and react to student satisfaction, a system was developed to allow students to indicate their level of ‘happiness’ with individual Units of study via the student port (myUNE). To measure student disengagement, data from each student’s interactions with IT systems (Callista, Portal, LMS etc) is mapped against 28 weighted triggers and the top at-risk students are automatically identified. Sentiment is mapped through an interactive word cloud and discontinuation through polling at the point of withdrawal from Unit.

Findings: In 2017, support interventions for identified at-risk students rose to approximately 10,000 incidents for the calendar year, with more than 7000 referrals to specific support providers to both internal (Counselling, Student Administration and Academics) and external (Centrelink, a local health provider for o -campus students). Through a series of Regression Analysis and Peer Review, the efficacy of the system has been validated, and engagement has grown, yet the overall impact on institutional retention is difficult to assess.

Discussion: With the bene t of hindsight, several key learnings emerge. Firstly, the ability to evolve the analytics engine, update data sources and conduct regression analysis as business as usual, would have been beneficial. Secondly, transparency around the collection of data is highly valued and institutional adoption could have improved through a user-facing dashboard and the ability to query the data at the desktop level. And lastly, staff churn (particularly in Senior Management and Sponsorship) is a key challenge to implementation of a model and ensuring long-term viability.