How a Little Data Can Solve One of Higher Education's Biggest Problems

Michael Morgenstern for The Chronicle

May 20, 2013

Last fall, Shaun Fowler started his sophomore year at Georgia State University still owing $500 on his tuition bill. The finance major from Atlanta had only a few days of classes before the university would be forced to kick him out for nonpayment. Then Fowler would become yet another stain on the reputation of large urban public colleges with stubbornly low graduation rates. Only about half the students at Georgia State graduate within six years.

Fowler couldn't come up with the money, so he was dropped from the class rolls. But within hours, he received a call from Georgia State's financial-aid office with an offer of a $500 grant. "I jumped at the chance," Fowler told me. Instead of shrinking his course load to lower his tuition bill, as he had planned, Fowler was back as a full-time student. "Without that money, I had no way to pay for school," he said.

A few hundred dollars might not sound like much to college administrators, but at campuses like Georgia State—where a third of the students come from families making less than $30,000—a couple hundred dollars may as well be several thousand.

As a result, Shaun Fowler has plenty of company among students struggling to pay tuition even after classes start. In any given semester, Georgia State typically finds itself with upward of $20-million in unpaid bills just a day before the university is required to cut students.

Most end up scraping together the money they need. But sadly, others don't. Rather than just watch those students drop courses or, worse, leave the university, Timothy M. Renick, Georgia State's vice provost and chief enrollment officer, wanted to find out more about them. By scouring the vast amount of data the university collects on students, he learned that some of the unpaid bills were as small as Fowler's and that the students who were cut from classes often had decent grades or were just a few semesters from graduating.

A few years ago, armed with a donation from Georgia State's president, Renick tried an experiment: He provided mini grants to nearly 200 students after they were dropped for nonpayment. It worked, helping to bring students back to classes and generating more than $660,000 in tuition and fees for the university that would have otherwise been lost.

So the university expanded the program last fall, awarding $600,000 to more than 700 students, making the average grant less than a $1,000. "We move quickly to reach students," Renick said, "because if they don't hear anything, they stop going to classes and fall behind."

The Panther Retention Grant is just one of several innovative approaches to student aid that Georgia State is taking to improve its retention and graduation rates. Like at other public universities where financial aid hasn't kept up with tuition increases, "unmet need"—the gap between a student's tuition bill and what financial aid actually covers—is a nagging problem at Georgia State. The average unmet need of students is about $5,400. As that figure grows, the university has found, students' academic performance falls.

Rather than just lament the sorry state of public higher-education financing or take a scattershot approach to giving financial aid to students, Georgia State has joined a small but growing number of universities using Big Data to guide decisions, much like Amazon and Netflix use algorithms to recommend books and movies to customers.

By mining the data, the university is able to spend its limited funds on students who have the potential to do the most with the extra dollars. Georgia State calls them "structured interventions": Find a problem, comb the numbers to figure out a solution, test the idea on a small group of students, and either tweak it or expand it if it works.

Algorithms also figured into another experiment, in which officials studied the university's biggest courses with the most D's, F's, and withdrawals. They found the students who performed well in those classes and are on financial aid and awarded them work-study dollars to be peer tutors the semester after they completed the course. The average grade in those courses with tutors has risen, as has the overall retention rate of those students.

At a time when many public institutions across the country continue to hand out merit aid to high-performing students—including in Georgia, where the state's Hope Scholarship pays tuition for its high-school graduates with a B or better average—the results of various financial-aid experiments at Georgia State, to focus dollars on students who need it the most just to stay in college, are notable.

With 24,000 undergraduates, Georgia State, like other public regional research institutions, occupies an often overlooked middle ground in American higher education. I came across the university's use of Big Data to better inform student-aid decisions through a project I'm working on with the New America Foundation. The project seeks to recognize a handful of public regional research institutions that have developed new models to serve the growing socioeconomic diversity of college students, a group the country's elite universities seem to have given up on.

What Georgia State demonstrates is that by better using data, universities can successfully weigh pressures to enroll a student body that reflects their state, and keep up retention and graduation rates at the same time.

The question is, why aren't more colleges following Georgia State's lead? I know, data mining unsettles many academics worried about privacy issues, but colleges already collect this information.

And the experiments at Georgia State confirm that data can actually be used to better serve students. The small bets the university is placing on students also shows that you don't need millions of dollars to solve some of the biggest problems in higher education. That pitch plays well with potential donors, too. "I have lunch with people I don't know to get $1,000," says Paul A. Alberto, interim dean of Georgia State's College of Education. "I tell them, 'If you won't name the College of Education, would you at least give me a $1,000?"

Finally, the Georgia State efforts exhibit that as states begin to finally reward colleges for graduating students, and not just enrolling them, institutions must reconsider the significant amount of money and effort they spend on recruiting students and finessing their aid offers to attract them to campus.

"Historically, the focus of colleges has been on the front end, getting students in the door," says Donald E. Heller, dean of the College of Education at Michigan State University and an expert on financial aid. Universities, he says, "looked at retention as an academic issue, not a financial one" for students.

That just doesn't work anymore. Shaun Fowler's experience illustrates that as tuition prices continue to climb, financial issues become a greater obstacle for a student who wants to stay in college—but not necessarily an insurmountable one.

Jeffrey Selingo is editor at large of The Chronicle and author of College (Un)Bound: The Future of Higher Education and What It Means for Students (New Harvest/Houghton Mifflin Harcourt, 2013).