In the movie Moneyball, the Oakland A’s reinvent themselves by creatively analyzing data to predict player success. That idea is hitting education in a big way, as The Chronicle reported this week, with colleges adopting new tools that mine data to suggest majors and predict students at risk of dropping out. But what if you could perform that kind of predictive analysis based on student information and course records from many colleges, not just one?
That’s the idea behind an ambitious project that is pooling more than 640,000 student records from six different institutions. Led by WCET, a group that promotes technology in higher education, the project aims to answer three main questions. What factors make some students drop out? What keeps others in college? What demographics make a difference?
The project, which got a $1-million Gates Foundation grant in May, could have a range of implications. To pick one example: Among students at risk of failing, one contributing factor seems to be a course load heavier than six credits, says Ellen Wagner, executive director of WCET.
“The implication for students at risk is just simply that we may need to rethink the way we have created academic programs—the declaration of a full-time status—when in fact perhaps that particular status seems to be implicated in students dropping out,” she says.
Colleges participating in the project include the American Public University system, the Colorado Community College system, Rio Salado College, the University of Hawaii system, the University of Illinois at Springfield, and the University of Phoenix.
