Big Data: Government's Next Frontier
Everywhere we turn these days, we bump up against "big data." It's in the news, sometimes in ominous-sounding contexts. It's on our computer screens every time a website suggests items we might like because of something we've purchased before. It's on our smartphones when we get a tweet from our bank notifying us of "possible fraudulent activity." Big data, data mining, predictive analytics, data analytics—all of these terms refer to the same idea: that people or companies with the right access, analytical tools, and training can comb through large repositories of information to find patterns that can help them predict outcomes and make informed decisions.
In the context of shrinking government budgets, the public sector is also starting to turn to big data to help get the job done. Carolyn Bourdeaux, associate professor of public management and policy at the Andrew Young School of Policy Studies at Georgia State University (GSU) in Atlanta, has a particular interest in the subject. "Big data is one of the first breakthroughs I've seen in a long time, with government able to achieve some real efficiencies," she said. "I really believe it's the new frontier in new government services."
Bourdeaux pointed to a couple of big data initiatives in Georgia, including at GSU. In the mid-2000s, the school piloted a handful of big data projects based on student information and saw its graduation rate increase by about 35 percentage points in the past decade.
Big data goes to school
In many ways, GSU is a typical urban public research institution. For one, it serves a diverse student population. Of a total of 32,000 undergraduates, 40 percent are first-generation students, 51 percent are on Pell grants, 60 percent are nonwhite, and 33 percent come from families with incomes of under $30,000. But in one very important way, the school is not typical. At a time when many public colleges and universities are being criticized for their dismal graduation rates—only about 30 percent—over the past decade, GSU has brought its rate up significantly, from a little more than 32 percent in 2005 to around 67 percent in 2013. "We are in the top five for the most rapid increase in our graduation rate," Renick said. "Our goal is to bring it up another 3 percentage points this fall." He credits the increase in large part an early alert program.
"We have become one of only three schools in the country to have a large academic and advisement tracking system that very much relies on big data," he said. That system is designed to intervene before a problem occurs. Every night, the school's computer system mines student financial data, looking for students who are at risk for dropping out for financial reasons.
The Panther Retention Grant program was piloted in the fall of 2011. Armed with a single donation from the university president and information from the data, Renick and others began calling some of the students who were dropped from the roster for financial reasons. "That first fall, we were able to help more than 40 students," said Renick. "Some students actually hung up on us because they thought it was a joke." The program has grown exponentially in the short time since the pilot. Last year, the program brought more than 1,700 students like Maria back to school, awarding them grants averaging less than $1,000 each.
"We've had large numbers of donations because people find the story compelling," said Renick. In addition to uncovering financial challenges, the computer system combs the data to flag students who are on track to drop out for academic reasons. "We took seven years of Georgia State student data—including over 2 million grades—and used those to develop analytics to indicate what students did that put them at risk for not graduating," Renick explained.
The school found that students who receive Ds or Fs in particular courses are likelier to drop out. The program automatically notifies school advisers about those students, and the advisers contact the students to set up a plan of action—either a tutoring arrangement with a student who has performed well in the class or advice that the student might want to consider another major. That advice is also based on predictive analytics, giving the adviser more concrete information on what course of study the student is likely to perform well in than a gut feeling on the adviser's part.
Renick noted that large public universities like GSU are receiving much criticism these days—about wasted dollars and about failing the very students they are designed to serve. He credits this criticism for GSU's success in part because it has "lit a fire" to tackle some of these issues. "We believe that it's not acceptable to take student tuition dollars and not provide them a clear path to success."
Welcome good news
"There are dozens and dozens of government services that could benefit from big data," Bourdeaux said. And thanks to big data already in action, Georgia residents like Maria are better off.
This article is a condensed version of a story written by Nancy Condon, managing editor for Extra Credit, for the third- quarter 2013 issue of the Atlanta Fed's quarterly magazine, EconSouth. You can find the full article on the Atlanta Fed's website.
January 21, 2014