Student Success Collaborative | Whitepaper
Predicting Success and Persistence
Of all the sources of real-time student success data available at a university, perhaps none seems to hold as much untapped promise as Learning Management Systems (LMS). For years, technologists have tried to develop risk algorithms based on course engagement, grade performance, and learning behaviors collected by an LMS. Unfortunately, these efforts to date have largely been unsuccessful.
The problem is one of scale. LMS adoption is far from universal, and usage patterns vary considerably from course to course and term to term. As a result, LMS-based risk models are inherently difficult to develop, costly to maintain, and have only limited applicability.
Our EAB data scientists wondered if it would be possible to design around these limitations and design an LMS-based risk algorithm that was optimized for scalability while still returning useful risk signals. Partnering with George Mason University, we found the answer wasn't more data; it was less data. The results offer hope of developing a broadly scalable LMS-based risk algorithm.