How Data Science Can Help Boost College Graduation Rates
A data philanthropy project is helping one university identify students at risk of dropping out – and intervene before it’s too late.
- By using analytics to identify and intervene with college students at risk of not completing their degrees, academic counselors hope to increase graduation rates.
- If students complete their degrees, they have a much better chance of achieving a higher income in their careers and succeeding in a rapidly changing job market.
In today’s economy, earning a college degree is viewed as essential for employability and higher earning potential. According to Georgetown University’s Center on Education and the Workforce “College Haves and Have-Nots” report of 2016, “The economy has added 11.6 million jobs since the  recession bottomed out – 11.5 million, or 99 percent of them, have gone to workers with at least some college education.”
But, low-income students have lower rates of college completion, which puts them at a disadvantage when competing for the high-quality, good-paying jobs being created in a rapidly changing job market. The National Center for Education Statistics (NCES) reported in 2015 that only 14 percent of low-socioeconomic status (SES) students received a bachelor’s or higher degree within eight years. Compare that to the high-SES students, 60% of which received a bachelor’s or higher degree within eight years.
John Jay College of Criminal Justice at the City University of New York (CUNY), like many other colleges across the U.S., is working to improve its graduation rates. Its four-year graduation rate is 27 percent and its six-year graduation rate is 47.4 percent. So, John Jay turned to data philanthropy to boost the college’s graduation rates.
With the support of the Center for Inclusive Growth and the Robin Hood Foundation, the college commissioned nonprofit DataKind to develop more than 20 machine-learning models using the college’s existing data for students with 90-plus credits, to predict the risk of dropout or delayed graduation. This will enable John Jay to identify students at risk and intervene with advisory services in an effort to increase their chances of graduating. John Jay advisors use the tool as an aide to prioritize interventions, rather than use it to dictate mandatory interventions; this human oversight is critical to ensure ethical outputs from the algorithm.
Melinda Rolfs, the Center’s director for data and analytics, spoke with Dara Byrne, associate provost for undergraduate retention and dean of undergraduate studies, and Steven Lee, managing director, income security, at the Robin Hood Foundation, to discuss the impact of the program and why they believe this initiative could be a game-changer in boosting graduation rates.
Melinda Rolfs: What’s behind the student attrition rate at CUNY John Jay College of Criminal Justice?
Dara Byrne: Typically, schools are able to make a dent in the attrition problems by investing in students from freshman year – particularly lower-income students like we have at CUNY. But there continues to be a lag, particularly around junior year, because that’s when students start reducing the number of courses they can carry due to the financial burden of going to college.
As affordable as CUNY is, our students often self-support. It takes a huge toll on them. A student at CUNY starts at around age 18 and has a nontraditional experience, working and commuting more than two hours to college. Some support children, whether their own or a family member’s. It’s a lot on their shoulders.
Rolfs: How does the problem of delayed graduation or dropping out disproportionately affect those students who come from a low-income background?
Byrne: The lower the income, the more likely you are to be impacted by that burden of how to afford all aspects of college, including time. Many of our low-income students are eligible for New York State’s Tuition Assistance Program (TAP), which they receive for up to eight semesters. If the student doesn’t finish on time, their financial aid package changes. Around junior year, they discover they’re running out of financial aid. Without access to the resources that smaller colleges are able to provide, that student often has to figure it out on their own.
Rolfs: What’s the potential for this use of data analytics as a way to improve graduation rates?
Steven Lee: We can intervene sooner with the [students] that we think have a likelihood of dropping out. There’s efficiency, but more importantly, targeted efforts that really get to the crux of what’s impacting low-income students.
Byrne: Absolutely. This also allows us to echo the issue of efficiency – to test what is possible before we overhaul an entire system. What this experience allowed us to do was come to a middle ground – test a small area, look deeply through the data and learn from what other industries are doing in order to scale a targeted approach. It has been a fantastic experience.
Rolfs: How will you use the predictive model and the findings to improve graduation rates at John Jay?
Byrne: Most people focus on freshmen because the research shows that if you do meaningful work with students at the beginning, they’re more likely to stay to the end. No one has looked at the students who are at late junior or senior standing. In one academic year, we lost close to 1,700 students at the 90-plus credit mark. That was disturbing because I didn’t understand why, I didn’t know what to do and we don’t have resources at the senior level.
This opportunity allowed us to dig into a mysterious population. Why would a student who has almost finished walk away? And do they come back? For the first time in undergraduate studies, we will develop an intervention to target students before the problem occurs. That is truly revolutionary.
Lee: I don’t think this level of analytics is being done much across the country to help low-income students. When you have all this data, not only can you intervene earlier, but you can also nudge students to do things that are to their advantage.
Rolfs: What implications does a project like this have in addressing barriers to opportunity and the future of work?
Lee: We want kids to stay in school and graduate so they can get work that’s going to have a meaningful impact in their life. That’s what, in many ways, college is for. The use of data [to help universities identify and intervene with at-risk students] can make it more possible for students to get that job.
Byrne: The tool is allowing us to see trends and patterns and help us to rethink our approaches earlier. It’s a huge advantage to have one small tool that costs us next to nothing. The more we engage with the opportunities and data skills that were provided by this project, the more it allows us to do more with our students around academic skills and job preparedness.
I am very hopeful about what this will mean in terms of preparing young people for the future, but also returning to that idea of a university being a place of opportunity. College degrees matter because they allow people to move beyond the first rung on the ladder. There was a study last year that [found John Jay] was third in the nation in terms of black economic and social mobility. [Editor’s note: 21 percent of John Jay’s students are African-American.] That is incredible because that is what we believe in.