Starting from Scratch – Building Your Learner Analytic Capacity
The promise of Big Data to increase effectiveness and success in higher education continues to generate “buzz” at many institutions where leaders are looking for ways to gain ground quickly. Building your know-how and capacity for learner analytics can be characterized as more of an organizational change process than the deployment of a new technology. Learner analytics are distinct from other Big Data sets, but require coordination and collaboration from the same players. Vernon Smith, from MyCollege Foundation, will share approaches to implementing enterprise-wide learner analytics efforts including potential barriers and strategies to take a leadership role at your institution. In this session, participants will 1) Explore business and technical processes involved in building learner analytics, 2) Identify learner analytic potential at their organization, and 3) Review strategies to create learner analytic capacity.
- Slides are available here.
- Goal is to explore business and technical processes involved in building LA
- Identify LA potential at your college
- Review strategies to create LA capacity
- Technical process: charter, capture, report, predict, act, refine. (Campbell & Oblinger, 2007)
- LA effects business process rules and breaks down silos.
- Key questions for stakeholders focus around: “At Risk Behaviors”, “Student Outcomes”, and “Early Interventions”
- Predictive Model #1 – On the 8th day of class includes 30 factors selected from the use of the LMS.
- Predictive Model #2 – Integrated warning within LMS. (e.g. login frequency, site engagement, points earned, points submitted, etc.)
- Student and Faculty Views of Early Alert System is important.
- Lessons Learned: Interventions
- Course welcome emails encourage students to engage early!
- Gen-ed students who login on the 1st day of class succeed 21% more often than students who do not.
- In Fall’09 at Rio Salado, there was a 40% decrease in the drop rate.
- 8th Day At-Risk Interventions for students who did receive direct contact succeeded more often than those who were unreachable.
Predictive modeling and executing it requires staffing and dedicated support!