Fiona Firstyear is the first person in her family to go to college. She’s from Small Town, Ky., and her parents aren’t able to contribute much toward her education—in fact, she qualified for a federal Pell Grant, meaning she’s among the neediest college students in the country.
At most universities, student advisors would be all over Fiona: First-generation students and those with low incomes are traditionally at higher risk of dropping out, and therefore receive the bulk of colleges’ retention efforts.
But at Bellarmine, the staff of the Student Success Center wouldn’t be too concerned. Are they crazy? Heartless? Not at all. Fiona also took all of the AP classes her high school offered, maintained a 4.0 grade-point average, and plans to join Bellarmine’s Pioneer Scholars program for first-gen students. Using a new technological tool that was developed at Bellarmine and has gained national attention, they can predict she will handle college just fine.
That tool, called the First-Year Predictive Model, “allows us to move beyond this very simplistic, ‘If X, then Y,’” said Dr. James Breslin, dean of student success at Bellarmine. The model analyzes more than 80 variables that can affect whether students return for their second year, as well as how constellations of variables relate to each other. “For example, for the last several years here at Bellarmine, simply being a first-generation student does not make a student at risk,” Breslin said. “If they are first-generation and a few other things, that could mean they are.”
Dr. Breslin developed the model along with Dr. Kristen Wallitsch, associate dean of student success for academic support, Drew Thiemann, director of institutional research and effectiveness, and Elizabeth Cassady, assistant dean and senior director of academic services, with support from the university’s Information Technology office. In October, they received the National College Learning Center Association’s Innovative Use of Technology Award. The model also received the American College Personnel Association’s 2018 Innovative Academic Support Initiative Award in February.
The team drew from five years of Bellarmine student-outcome data to select the variables for the model, which include behaviors such as attending orientation and accessing course websites in addition to demographic information.
“Single categorical factors that higher educators used to think translated to at-risk status (e.g. underrepresented minority, or first-generation students, or male, or from a weak high school) are no longer a risk factor by themselves,” Thiemann said. “There is much more of what we call ‘intersectionality’ in these data, and as a result, what we’ve found through the First-Year Predictive Model is that Bellarmine students with a single factor (such as one of the above) are just as likely to succeed as they are to fail. [When retention issues arise] is when you compound these factors.”
The First-Year Predictive Model assigns a risk factor to every student in the incoming class. The class is then broken into five equal groups, from those at extremely low risk of dropping out at one end of the spectrum to those at extremely high risk at the other.
This allows the Student Success Center and other support staff to target the students who really need help and to pinpoint which resources will be the most effective. Imagine using a watering can, rather than a firehose. “Instead of giving every student a barrage of ‘Here are 50 things you could do to make your experience better,’” Dr. Breslin said, “we can say, ‘Here are some things that could be highly impactful for you.’”
Best of all, he added, “it doesn’t replace or add anything we are doing” in terms of student support. “It just makes everything we are doing more strategic.”
By running the model again after the end of the drop/add period, Bellarmine can also see how the risk of attrition changes over time, something that Dr. Breslin said most universities are not in a position to do. A student who was not a retention concern in August might move into a higher-risk group later in the semester if he or she isn’t attending class or accessing services like tutoring, for example.
The team is continuing to add variables to the Predictive Model. In fall 2017, students took a new survey that helps the Success Center assess non-cognitive skills such as motivation and self-management. “We are trying to gather as much holistic information as we can. We recognize that students aren’t just students—they come from a family, and they come from a community, and they choose to do things or not do things here,” Dr. Breslin said.
“What it comes down to is, what are we as humans doing for them as humans? It might be interesting to know what is going on with a cohort, but what are we going to do about it?”
A student’s high school grade-point average is the single strongest predictor of retention, says Drew Thiemann, Bellarmine’s director of institutional research and effectiveness and one of the architects of the First-Year Predictive Model.
Four other highly influential predictors:
- Students who attended an independent, Roman Catholic high school are 2.4 times more likely to retain.
- Students in a learning community are 1.9 times more likely to retain.
- Students who earned advanced standing credit (AP, IB, etc.) in high school are 1.6 times more likely to retain.
- Recruited student-athletes are 1.6 times more likely to retain.
By Carla Carlton | firstname.lastname@example.org