Big data may lead to big gains in student success

USF is using big data to change university decision making based on student risk.  ORACLE PHOTO/ADAM MATHIEU

Big data is coming to the university system to implement statistical methods for determining how students will fare in college and change the way campus decisions are made. 

USF is one of the universities currently working on using data to help students maintain higher grades, graduate on time and find well-paying jobs after graduation. To achieve its goal, the university is using predictive analytics, a type of statistical analysis that is meant to predict the future, rather than analyze the past.

”It’s a way to direct and change our resources to students at a time when they might most need it. That’s really, for me, the heart of the concept,” senior director of academic tracking and advising for the Office of Academic Advising Initiatives Travis Thompson said. 

The university is working with two programs — Inspire for Advisors and Illume — that fit with a statistical approach. 

According to USF Vice President of Student Affairs Thomas Miller, Inspire for Advisors looks at individual students while Illume looks at large groups of students as a whole. Thompson said the university has been working to incorporate the former into advising for eight months.

According to Miller, Inspire for Advisors will help advisers identify students most at risk of failure and try to give them a path to success. Miller also designed a model for analyzing reasons for student attrition, or dropout rates, which the university has been using since 2006.

 “Predictive analytics would allow advisers to help students develop plans before they go too far down the road. Early decision-making about an academic path is helpful in the advising process,” he said. “The tool for advisers is out there, but it’s not yet perfect, because some of the data is incomplete.”

According to Thompson, big data is also helping USF achieve other goals, such as attaining a 92 percent first-year student retention rate, as compared to the current first-year student retention rate, which Miller said hovers between 88 and 89 percent.  Analytics also help the university meet its goals for the state’s performance-based funding, which contributed over $22.2 million to the university in 2014. 

Illume monitors the entire student body. According to Miller, it looks at general trends from the student body, such as the graduation rate of male students in science, technology, engineering and math fields. The program can also look at other factors, such as living on campus, to see if it makes a difference to student success overall.

Illume and Inspire for Advisors both come from a relatively new company called Civitas Learning, which has only existed for about two or three years, according to Thompson. The company also offers a variety of other products, such as Inspire for Faculty and Degree Map. 

USF’s project with Civitas uses over 200 characteristics to predict a student’s success or failure, such as high school GPA, credits earned in the previous semester and cumulative GPA, according to USF Vice Provost of Student Success Paul Dosal.

“(Inspire for Advisors) provides advisers with a fairly easy-to-read and -understand snapshot of where a student stands and provides some indicators of the likelihood of the student persisting,” he said. 

The university draws its data from the Office of the Registrar and the admissions database, Miller said. Dosal and Miller said that they hope to incorporate data from Canvas into the database, such as participation rates in online discussions. 

Concern over predictive analytics comes, in part, with USF’s membership in the Florida Consortium for Research Universities, an education partnership established last year. As one of the Consortium’s “four pillars,” predictive analytics show a shift toward statistics in higher education.

According to Thompson, the University of Central Florida, St. Pete College, Georgia State University and other institutions have begun experimenting with predictive analytics for monitoring students.  

Miller said the university has switched to big data because it provides firmer answers to important questions, as compared to case-by-case examples that are harder to apply to other students.

Hard numbers also help the university make decisions that affect the whole student body.

“I think there’s a challenge with using anecdotal evidence to make policy. We have an Office of Decision Support, and that basically is our institutional research office, and their responsibility is to gather data and analyze it and try to help inform decisions that need to be made about policy or practice,” he said. 

“If you don’t have data-driven decision-making, it’s all guess work.” 

Dosal said USF needs to use these analytics with sensitivity because of the potential for error in statistical analysis. 

“We’re also aware that even the best predictive models can’t predict with 100 percent accuracy whether a student will succeed or fail, so the tool has to be used with caution and great care,” he said. 

Dosal said he expects Inspire for Advisors to be ready to use within a month, though it could take longer.