Using Big Data to Match Students and Colleges

SMITH BRAIN TRUST — It's college decision time for many high school seniors. But the process that students and their families use to choose where they'll enroll is far from optimal. "Students and their parents look at prices, and they look at rankings, but they might not put emphasis on other details that might be more important," says Anamaria Berea, a postdoctoral research associate at the Robert H. Smith School of Business, at the University of Maryland.

In a partnership with the startup Vibeffect, Berea has been exploring how big data can help improve matchmaking between students and universities. That big data might come from surveys or from information "scraped" from Twitter, or both.

Researchers have long sought to determine what causes high school graduates to thrive in college. Is it mostly brainpower or the rigor of courses taken in college? Or is it more about a trait like grit? Could it be the ability to say goodbye to one world (home) and embrace a new one (college)? In a study published in May 2015 in Decision Analytics, Berea — affiliated with the Smith School's Center for Complexity in Business — and four co-authors established that the causes of thriving were more complex than most scholars had thought. Certain constellations of personality traits, tastes and talents of students led to thriving on campuses with certain characteristics. But students could belong to multiple, overlapping constellations, and so could colleges, making the matching process complicated on both ends.

The researchers established this by surveying some 2,857 young people, 18 to 24, most of them sophomores, juniors, and seniors in college (but some dropouts, too) about the institutions they had attended. They asked about such things as how far a college was from a student's home, size of enrollment, whether it was public or private course of study, and what a student's perceptions were of other students. Students were also asked about their own demographic characteristics, personality traits, and likes and dislikes.

Finally, they were asked 18 questions that probed various aspects of whether the students had thrived at college, inside and outside of class.

Certain clusters of traits led to success at certain institutions. The researchers showed that, in theory, they could both identify those matches and also trade off the tightness of the fit with the accuracy of the estimate. Perfect and near-perfect fits are hard to predict with accuracy, but the model works well for students seeking an 80 percent match. Who would benefit from such a tool? "This is a kind of win-win situation," Berea says. "Both students and universities want students to get the most out of their four-year experience."

A weakness of the study is the lack of follow-up with the interview subjects, something that could be the subject of a future study. (One follow-up was conducted, but mainly to validate the algorithm.) Vibeffect is now making use of the algorithm in a commercial digital product.

Another recent paper by Berea and a different set of co-authors — including Smith School professor William Rand and Smith undergraduate student Kevin Wittmer — analyzed the attitudes that high school students and others express on Twitter about the college application process and the college experience. The study is more about demonstrating the ability to analyze tweets than about delivering a firm conclusion. Still, it shows that the comments on Twitter about college selection, applications and attendance were notably negative. That suggests there is lots of room for improvement in college outreach to students.

READ MORE
"A New Multi-Dimensional Conceptualization of Individual Achievement in College," by Anamaria Berea, Robert H. Smith School of Business; Maksim Tsvetovat, George Mason University; Nathan Daun–Barnett, University of Buffalo; Mathew Greenwald, Greenwald & Associates; and Elena Cox, Vibeffect, appeared in Decision Analytics in May 2015.

"Social Media Analysis for Higher Education," by Anamaria Berea; Kevin Wittmer, Smith School; and Gerard Wall, Vibeffect, appeared in the Fall 2015 issue of the Journal of the Washington Academy of Science