Whether we like it or not, the choices we make on a daily basis are data points for advertisers and politicians. Who we are in terms of our race, ethnicity, gender, age and socio-economic status as well as where we live, decide to shop for groceries, and the websites we visit all become information to be used by politicians seeking votes or companies wanting our business. Sophisticated data analytics try to predict what we will buy, where we will go on vacation, who we will vote for, and which college we will attend. And no, higher education has not been immune from this predictive analytical pathogen. However, in my opinion, as long as we are aware of potential pitfalls and take appropriate antibiotics, predictive analytics can provide many benefits to institutions who want to maximize the limited human resources they have to support and retain students. Data is neutral. How we use data could have either positive or negative consequences.
The recent controversy at Mount St. Mary’s University highlights the potential dangers of predictive analytics. Rather than using data to support students to achieve their higher education goals, faculty were asked to administer a survey during freshman orientation that would identify potential drop-outs. The university also monitored student attendance at campus events, dining activity, and payment of bills. Freshman students who were identified as at-risk of dropping out because of the data collected would then be encouraged to withdraw from the university before the cut-off date for establishing the retention cohort, which is tracked by the federal government. When challenged about this plan, the former president reputedly (and now famously) said, “This is hard for you because you think of the students as cuddly bunnies, but you can’t. You just have to drown the bunnies … put a Glock to their heads.”[1] The uproar that followed, resulting in the resignation of this president, clearly demonstrates that this is not an example of “best practice” to be followed by other institutions. The vast majority of faculty and staff at colleges and universities would rightly condemn this particular use of data. However, there are positive uses of student information data. I would like to describe how we are using data to help students achieve their goals.
Like many institutions that seek to improve retention and graduation rates, Virginia Commonwealth University (VCU) has invested funds to expand advising services for first-year students, incoming transfers and, most recently, upper-level students. At first, VCU focused on establishing an “intrusive-proactive” model of advising for first-year students. Advisors sought to meet all first-year students at least twice for structured advising sessions each semester. Early alert reports were run each semester to identify those students with failing mid-term grades. Advisors then contacted those students to recommend strategies for improving their grades. However, with limited numbers of advisors for students beyond the first year, this model was not practical to implement for the large population of continuing sophomores, juniors and seniors. Therefore, in 2013-14, VCU partnered with the Student Success Collaborative of the Education Advisory Board (EAB) to provide predictive analytics of VCU’s graduation data and implement a “strategic-proactive” model of advising.
After working with students for just a couple of semesters, advisors quickly learn about the academic barriers that delay students from graduating in four years, whether it be withdrawing from too many classes, failing degree requirements, or being placed on academic probation. Predictive analytics draw attention to less obvious barriers. For example, although the minimum grade in a certain required course may be “C,” the predictive analytics may tell us that students are more likely to graduate if they earn a “B” grade or higher. Similarly, although the minimum GPA required for graduation may be 2.0, the predictive analytics may inform us that, for certain majors, students are more likely to graduate if their first-year GPA was 2.6 or higher. In VCU’s strategic-proactive advising model, advisors use insights from the predictive analytics of VCU’s student information data to design targeted, proactive advising campaigns.
The advising campaigns advisors design are “targeted” because they identify a group of students who need a particular type of advising intervention. They are “proactive” because advisors reach out to the students rather than wait for the students to come to them. In many cases, the students targeted may not be aware that they need advising assistance. The campaigns vary in size, timing, format and objectives. Depending on the staffing and resources available, advisors could target small or large groups of students and implement single or multiple campaigns over the course of the semester. In some cases, the campaign could be as simple as sending multiple email messages informing targeted students of available resources. For example, chemistry students who earned a “C” in CHEM 101 the previous semester (which has been identified a potential barrier to graduation in chemistry) and are now attempting CHEM 102 receive multiple messages encouraging them to attend tutoring sessions. In other cases, the campaign may target certain students for more intensive advising that includes email reminders and individual or group advising sessions. For example, an advisor may contact students who repeatedly withdraws from multiple classes over the course of the year and schedule a series of advising sessions during the semester. What all the campaigns share in common is that advisors go beyond their normal course of business to help students succeed. Advisors are not drowning bunnies, they are saving lives.
– – – –
References
[1] Katherine Mangan, “A president’s plan to steer-out at-risk freshmen incites a campus backlash,” The Chronicle of Higher Education, January 20, 2016.
You Might Also Like
