Data Dive

Georgia State's extensive predictive analytics efforts are leading to better grades and student retention -- and more minorities graduating from STEM programs.

July 19, 2017
 
Georgia State
Effective use of data is helping Georgia State graduate thousands more students, many who are poor and minorities.

Some colleges have data systems that notify advisers when students are in danger of failing a course or flunking out, or recommend majors to students based on their interests and academic performance. Instructors also increasingly analyze their students' assignment and exam data to improve pedagogy and improve outcomes.

But Georgia State University is taking predictive analytics to a whole new level.

The institution's systems update student grades and records every night, and they review 800 risk factors for each of the 50,000 students on a continuous basis. So when a first-year student is getting poor grades on exams in an introductory math course, an adviser is alerted and reaches out to the learner, whose current struggles may be a sign that they won't be prepared for more complicated business or science courses during their junior and senior years.

Based on system alerts, dozens of Georgia State advisers have held 200,000 meetings with at-risk students during the past five years, including 52,000 during the last academic year, said Timothy Renick, vice president for enrollment management and student success and vice provost. Those meetings lead to students going to campus tutoring centers and instructors for remedial help and other academic support services.

The grand-scale efforts are achieving big dividends for the downtown Atlanta university whose student body is 60 percent low income and 60 percent minority, and whose learners are predominantly the first in their families to attend college. According to Renick, Georgia State now is graduating 2,000 students more per year than it did five years ago; degrees awarded are up by 30 percent during the same period, outpacing the university's 5 percent enrollment growth.

"The big increase [in the graduation rate] has been in the number of minority student graduating in STEM fields, Renick said. "Only a small part of the increase is due to increases in enrollments of minority students."

The overall efforts also have produced an additional $10 million per academic year in tuition and fees, because fewer students are dropping courses or leaving altogether, he said.

Lots of Potential

Small liberal arts colleges and large public universities alike see potential in predictive analytics, which uses data to support informed decision making about institutional strategy and to help learners progress through courses toward graduation.

“We’re reaching a point where the idea of predictive analytics is really catching on,” Iris Palmer, a senior policy analyst at New America, told Inside Higher Ed earlier this year. “As institutions are pressured to improve completion and graduation rates and provide more support to students, it’s going to become more common.”

Georgia State’s use of predictive analytics isn't typical, said Michael Abbiatti, vice president of educational technologies at WICHE Cooperative for Educational Technologies (WCET). “It’s a complete transition as a university,” Abbiatti said. “The breadth and depth of the deployment of analytics is campuswide.”

Georgia State is part of the University Innovation Alliance, a coalition of 11 public research universities that serve large numbers of first-generation, low-income students. Art Seavey, a senior program officer for postsecondary success at the Bill and Melinda Gates Foundation, said that Georgia State was the first alliance member to scale the use of predictive analytics; now the others are implementing or have implemented predictive analytics, he said.

“They are [seeing] early evidence on how well Georgia State’s efforts are working,” Seavey said.

Prior to 2012, Georgia State advisers conducted about 1,000 meetings per academic year with students, but those sessions typically were with conscientious learners who were going to graduate or with those who were really floundering. “Those meetings were largely ineffective,” Renick said. “When we looked at the data, the advice the advisers were giving [to the struggling students] was coming too late.”

Working with EAB, a research/technology solutions company, the university launched a data system in August 2012 to track its then 30,000 students on 700 initial risk factors.

“Now we know that if you get a C in your first math course … it might impact you in your upper-level accounting course,” Renick said. “If you are waiting for students to self-diagnose and identify the problem, it is too late.”

Georgia State’s efforts not only are helping elevate overall grades and retention rates, but also increasing enrollment of nonwhite students and minority students graduating with STEM degrees. Renick said that the university’s Hispanic student body has increased by 22 percent since 2012, and that degrees conferred to Hispanic students in STEM fields is up 226 percent.

Black male enrollment is up 15 percent in five years, and the number of black men graduating with STEM degrees is up 111 percent. “Black males would come in as pre-med or science majors and would get D’s and F’s and no one was paying attention,” Renick said. “Now by the end of their freshman  year, they can proceed in STEM.”

In 2016, Georgia State merged with Georgia Perimeter College, a two-year institution in the Atlanta suburbs with a similar student body, and gained 20,000 students. Renick said the transition hasn’t slowed or impeded the institution's analytics work or advising efforts. (The university received a $225,000 grant from the Gates Foundation to help transfer and incorporate the Georgia Perimeter students’ data.)

Stepping Up the Data

More recently, Georgia State began providing advisers with information about job prospects and starting salaries for many industries as well as students’ financial aid packages. In the past, advisers did not want to discuss issues related to financial aid, but Renick said that data can be critical to whether a student graduates or drops out. For instance, if a student decides to add a second major, they might not have enough Pell Grant funds or other financial resources to pay for the extra classes.

Likewise, the financial data help staff advise students about critical nonacademic decisions. “It is unrealistic of us to think that a first-generation student who is offered a nice single room should know that it is going to cost more than a room with other students,” Renick said.

“We never provided an integrated look,” he added. “We needed to treat students holistically. The ability to put the data in the hands of staff helps students make wise choices.”

Not Without Challenges

Faculty members did not initially embrace administrators’ ambitious efforts. With predictive analytics, there always is potential for abuse -- namely that data could be used to judge students and limit their opportunities -- and Renick said that instructors feared some students would be steered toward easier majors. But that’s not happening, he said, noting that the university's two fastest-growing majors are computer science and biology.

“Students are getting interventions three or four weeks into the semester,” he said. “They are told … ‘we’re going to work with you.’ Give students the information they need so they have a fighting chance.”

Faculty members also initially thought the predictive data would be have a “big brother” effect, exercising too much control over students. Renick said the university has not received pushback from learners about how their information is collected or is being employed.

Georgia State’s efforts have been effective, but at a cost. The university hired 42 new advisers in the first six months after launching its data initiative in 2012, at an annual cost of $2.5 million per year. However, Renick said, student retention has increased by 4 percent -- which equates to $10 million a year in tuition payments and fees.

“What we need is a mind-set of wise investments for students,” he said. “Too many campuses look at the risk of investing $2 million. Any business that takes that approach may be left in the dust.”

WCET’s Abbiatti said investments in data collection and analysis technologies won't solve institutions' student success and retention problems. “It’s not a hardware issue or a software issue,” Abbiatti said. “It’s a skin-ware issue.” Georgia State, he said, has hired staff who can analyze and employ the data -- and that’s the big takeaway of its wide-scale initiative.

Read more by

What Others Are Reading

  • Viewed
  • Past:
  • Day
  • Week
  • Month
  • Year
Back to Top