The 'Murky Middle'

Colleges and universities trying to improve retention and graduation rates may be directing academic support services to the wrong students, emerging research suggests.

Researchers at the Education Advisory Board, a Washington-based company, think predictive analytics can inform the student-success strategies institutions employ. The company is collecting student data from roughly 110 four-year colleges and universities as part of its Student Success Collaborative. So far, about 60 institutions have provided 10 years’ worth of transcripts as well as demographic details such as students’ home towns, said Ed Venit, the lead researcher on the company’s student-success project. The participating institutions are public and private, of varying sizes and selectivity rates.

As Venit and others look at the data, they’ve noticed some patterns that challenge the common understanding of when and why college students drop out.

The ‘Murky Middle’

Colleges lose the biggest share of students in their first year, so institutions direct most of their student-success resources to freshmen, Venit said. (Such resources include tutoring, mentoring programs or other interventions.) “That’s where everyone’s devoting a lot of attention,” Venit said. “But we’re not moving the dial.”

Colleges fixate too much on freshmen, he said. Many students complete their first year but leave without a degree. Forty-five percent of total dropouts nationwide finish a year of college and with a grade-point average between 2.0 and 3.0, the company's research has found.

“We have a mystery here,” Venit said. “Because students between a 2.0 and 3.0 are dropping out. But they’re not dropping out after their first year.”

Nine out of 10 students who finish their first year with a G.P.A. of 2.0 or higher return for a second year, he said.

Some students will graduate no matter what, Venit said. Others will almost certainly drop out. And at most colleges and universities, academic support goes disproportionately to the students who are thriving, because they seek it out, and to the students on the cusp of failing, because the college sees they’re at risk. But what is the dividing mark between the students who will typically succeed and the students who will almost surely fail?

First-year G.P.A., the researchers contend, offers a powerful indication of a student’s chances of graduation. Students who end their first year with a G.P.A. of 2.0 or lower are unlikely to graduate despite the best efforts of their colleges, Venit said. Students with a G.P.A. of 3.0 or above, on the other hand, graduate in high numbers.

But there’s also a third group: the students who end their first year with a G.P.A. between 2.0 and 3.0, who make up nearly half of total dropouts. Their fates are more uncertain. These students belong to what Venit and his colleagues call the “murky middle.”

“It’s called the murky middle because in that group of students [students with a first-year G.P.A. between 2.0 and 3.0], a certain number are going to leave and a certain number are going to stay, and they look about identical to each other,” Venit said. “They are making progress against their major. They aren’t tripping any alarms. They aren’t showing up on anyone’s radar as being particularly at-risk. But some of them aren’t going to come back.”

Two students, each with a 2.7 G.P.A., might look the same superficially. But one student might receive mostly Bs, while the other earns a mixture of As and Fs. The student with Fs is at much higher risk, because she may not be completing enough course credits.

Within this group, Venit said, small academic improvements correlate with greatly heightened chances of graduation. Thus the “murky middle” offers colleges a powerful “return on investment,” he said. Just a small nudge – one-on-one tutoring, time management counseling – could keep a student on track to graduate.

The typical “murky middle” dropout spends 4.5 to 5.7 semesters at college before dropping out. “They may actually be easier to inflect than the ones who are really, really in trouble,” Venit said. Sophomore-year interventions could reduce “murky middle” dropouts, he said.

A dip in grades typically starts several semesters before a student in the “murky middle” drops out, the company has found.

Doing ‘Triage’

Given the limits on how much money and time colleges can spend on academic support, an implication inevitably follows from the company’s findings. To maximize retention and graduation rates, colleges ought to shift some of their student-support resources from first-year students and from students unlikely to graduate and allocate them toward at-risk students in the “murky middle.”

Redirecting student-support resources relies on the “same hard unfortunate logic of triage on the battlefield,” said Nate Johnson, a Tallahassee-based consultant and the founder of Postsecondary Analytics. “If doctors have 20 wounded soldiers to treat, they’ve got to focus on the ones who are going to die without their help, not on the ones whose fates are already sealed.”

Failing to make such decisions means “letting some students who could succeed in college fail because we haven’t allocated our limited resources appropriately,” he said.

But colleges should approach with caution any plan that calls to redirect resources from high-risk students to less high-risk students, Johnson said.

“I suspect that we do probably put a lot of resources into students who are just not going to ever make it no matter what,” he said. “On the other hand, I’m very cautious about withholding resources from those students because they’re the ones with the biggest potential to change the educational and socioeconomic configuration of the country. So every win for the high-risk student is worth a lot for what it does for the country or for a state or an institution.”

Alan Seidman, executive director of the Center for the Study of College Student Retention, was less impressed by the emerging research. He said he thinks the predictive-analytics approach misses the point.

“Institutions have the data now, without a big data set, to know who’s going to be a successful student,” Seidman said. “Having a big gigantic database that says we looked at all these schools and it says x, y and z – it might not hold true for my particular institution ... So why do we need this big national study?”

There’s a much simpler solution, he said. The real problem afflicting student retention, he argues, has to do with curriculum. If textbooks in introductory courses are at a reading level that a student isn’t prepared for, or if a foundation course requires writing and mathematics skills that a student doesn’t have, then the student will not pass, he said.

“We can look at students all we want,” Seidman said. “We’ve looked at student background. We’ve looked at academics. Yet retention rates have not increased. If the textbook is at a level they can’t comprehend and can’t read, then they’re not going to be successful no matter what they do.”