Page 29 - Campus Technology, January/February 2020
P. 29

“Every institution should know the top 10 indicators that a student is at risk for
dropping out. If we know those indicators, we can begin to redesign our processes and intervene before it’s too late.”
Frazee: To provide some context on industry partnerships and illustrate their value relative to promoting student success, take this example around predictive analytics.
Recognizing that grades ultimately predict grades, SDSU is now exploring additional vari- ables which are giving way to a more holistic view of students’ course-based experiences. The digi- tal footprint from course engagement allows for what are proving to be early and accurate predic- tions of students’ ultimate course performance trajectories. In other words, SDSU is moving be- yond leveraging grades and points earned to in- form student success interventions, and looking instead at students’ behavior as a proxy for effort and motivation to learn at the course level.
In 2018, SDSU partnered with Pearson, and
began to look specifically at students’ engage- ment with MyStatLab homework assignments in a high-challenge introductory statistics class. Nearly 25 percent of the students in this under- graduate course receive a “repeatable” grade — a struggle that has persisted for years. Leveraging Pearson data generated by students’ engagement with their statistics MyStatLab homework, San Diego State employed a new algorithm focusing specifically on the number of days between stu- dents’ homework initiation and their due date.
Predictions of students’ placement in one of four clusters — Early Compliant, Compliant, Late Compliant or Non-Compliant — were more than 50 percent accurate within the first two weeks of the course, using only the homework engagement data variable. Predictions from week nine reflected more than 85 percent accuracy. Further analysis

   27   28   29   30   31