The use of advanced data tools by US universities to boost graduation rates is taking a controversial turn with steps some see as potentially regressive, such as suggesting that struggling students consider different majors.
Various data tools have become popular among colleges for tracking student progress on numerous measures inside and outside the classroom, and for offering targeted help. Hundreds of US institutions are understood to be trying them on some level, with some seeing encouraging results among traditionally disadvantaged groups.
Leaders include Georgia State University and the University of Maryland, Baltimore County, both of which originated as white-majority institutions and now enjoy growing reputations for producing accomplished minority graduates.
The two universities differ, however, on the wisdom of using data-driven predictions of academic performance to steer some students away from their chosen fields of study – a move that some fear could reinforce some of the same demographic stereotypes the institutions are trying to overcome.
UMBC doesn’t do it, said Robert Carpenter, the university’s associate provost for analytics and institutional assessment. The tools aren’t yet reliable enough to avoid “flagging a student as being at high risk for a poor outcome when they ultimately had a good outcome”, Professor Carpenter said.
Instead, he explained, UMBC stuck to using predictive analytics to identify students at risk of failing a course, and then provided the necessary additional support within the class. “We see this as holding doors open rather than closing doors to them,” Professor Carpenter said.
Georgia State does use predictive data to suggest that certain students consider new fields of study, and sees the objections as overblown. Colleges have long talked with their students about their choices of majors, said Timothy Renick, the senior vice-president for student success at Georgia State.
“What is new”, Professor Renick said, “is providing students with data about their likelihood for success in different majors and allowing them to make informed decisions. What is new is getting students this information before they waste time, money and aid eligibility, before they move to a better fit.”
Experts studying the question say they recognise the potential for gains in the Georgia State approach, but suggest that many institutions aren’t taking seriously enough the possibility of harming students who need encouragement to stay on more challenging pathways.
There are some extreme outliers. Mount St Mary's University, a 1,600-student Catholic liberal arts institution in Maryland, was found to be planning to use predictive analytics to encourage the actual withdrawal of students seen as likely to fail.
But even among more well-intended users of advanced data, said one expert, Iris Palmer of the policy study group New America, colleges appeared to be embracing powerful interventions without having fully studied their effects.
One of the few assessments, funded by the US Education Department, evaluated the results of one year of data-driven “intensive, proactive coaching interventions” at 11 large state institutions.
The study, carried out by the research firm Ithaka S+R and published last year, found significant benefit at only one of those 11 institutions, Georgia State.
The result wasn’t surprising, the study authors said, since the anticipated benefits of predictive analytics are expected to be longer term. But the “fairly weak results” and the general lack of research into the question, said another expert in the use of predictive analytics in education, Kyle Jones of Indiana University, should make institutions wary of taking risks with student career choices.
