It’s a saying we’ve all heard in one history class or another: Those who do not learn from the past are doomed to repeat it. The maxim can apply to business intelligence as well – especially when it comes to predictive analytics. In fact, learning from past experiences is key to making predictive analytics work.
Data analysts are able to use past results to plot out the most likely outcomes in the future. Many colleges and universities are using predictive analytics to increase retention. Let’s take a look at how they’re doing it and exactly what information they are using to make the most informed decisions.
The challenges of retention
For different types of schools, there are different types of issues that impact student retention. For some it’s simply academic or financial – the classes are too difficult and the student can’t keep his or her grades up, or the school is too expensive and the student has had second thoughts about whether it is worth the expense to stay. But there are other situational factors that come in to play, like whether a student is the first member of a family to attend college. That student might not be getting support from home and the school might have to take that factor into consideration.
Schools know all of this, of course. Any information that could have an impact on a student’s chance of staying at the school they have enrolled in is gathered by the institution. Many colleges and universities will even conduct surveys either during summer orientations or during the first few weeks of school (or both) to get a feel for how the student is doing and to try to get behind just the numbers to gauge the student’s happiness at school.
The data behind predictive analytics
Although the reasons a school might be seeing a decline in retention could be attributable to any number of different factors from school to school, it is data that they all have in common when it comes to trying to keep students on campus. They all have data that can be used to help them figure out how to improve retention.
For example, Bellarmine University in Kentucky turned to predictive analytics in an attempt to improve its retention numbers. Some of the variables that it used in its model were a student’s high school GPA, which was the top variable in the model, providing an overview of what the student’s engagement looks like over four years of high school; whether or not a student was a student-athlete; students with dual credit coming from high school, and whether those dual credits were from a 2-year or 4-year institution; as well as personal problems – how the student was feeling at various checkpoints during the year at Bellarmine. There were also variables that the school considered but that did not end up in the model, including residency status, Pell eligibility, or a parent’s education level.
Bellarmine, like any school working with predictive modeling, knew it could not account for every factor that would contribute to a student leaving. It could, though, use these significant factors that have historically proven to indicate attrition or retention, and use them as tools along with the other resources available to improve communication and outreach in terms of retention.
More than just the data
It doesn’t matter if a school changes its entire technological infrastructure to identify the students it needs to help if it can’t help those students. It needs to have effective support systems that will allow the school to step in and make a difference. Adjustments can certainly be made after the fact, but the structure of support needs to be in place. A school can’t start from scratch with support when it needs to make an impact immediately.
One reason Bellarmine found success with the data it was using was because it also had support systems in place that it knew would be able to help the students it identified as needing support. The goal of its Student Success Center is for students to find success in their first year and persist to graduation.
Some schools take reactive steps to improve. An accrediting agency might be gathering information and the school recognizes that its retention numbers are low, so it scrambles to address the issue. The schools that are the most successful are the ones that are proactive, making the kinds of adjustments that would satisfy an accrediting agency, but doing them because it’s part of the school’s mission, not because it’s going to be checked up on.
Predictive analytics might start with looking at missteps in the past and trying to fix them for the future…but that future might involve looking back at the successes of the past and trying to repeat what worked.
Latest posts by John Sucich (see all)
- The Role of Data Quality in Higher Education Analytics - September 4, 2019
- 3 Ways Analytics Can Make Higher Ed More Efficient - July 24, 2019
- Finding Higher Ed Success with Data through Dashboard Visualization - July 10, 2019