Washington Monthly - September/October 2019 - 24

have progressed and spread, and what policymakers are doing
to advance them.

Predictive Analytics
One of the biggest challenges facing higher education is the
need to not just enroll more students, but keep them. Only
around 40 percent of the students who enroll in college each
year graduate in four years. (For students of modest means,
the odds are even worse.) With such a high number of dropouts, institutions are under increasing pressure to help students persist until they earn a diploma. It turns out that the
emergence of Big Data can help.
Georgia State University, in partnership with the consulting firm EAB, was one of the first universities to pioneer predictive analytics-the use of past data to forecast future behavior
and design interventions to shape it-in an academic setting.
The aim was to use troves of digital information to identify predictors of when a student was at risk of failing. The school found,
for example, that if a student earned a C or lower in an introductory course for their chosen major, they were statistically less
likely to graduate. So now, when that happens, an adviser will

Predictive analytics isn't
magic. It only works when
a college uses the data as a
basis for changing the way
it interacts with students.
reach out and help the student identify another course of study
with a better chance of success. In recent years, GSU's graduation rate has risen by more than 30 percent. Even more remarkably, the school has eliminated the racial achievement gap, and
now awards more bachelor's degrees to African Americans than
any other nonprofit college in the country.
Now, more than 5,000 schools are partnering with EAB
and other firms to apply predictive analytics models to improve student outcomes and replicate Georgia State's success.
One of the most innovative companies is Civitas Learning,
based in Austin, Texas, which is now working with more than
a dozen colleges, including the University of South Florida and
Utah State. Civitas's basic thesis is that the best indicator of
whether a student will graduate is not grades or test scores or
demographics-it's their behavior. Habits like logging into
their online class portal reflect engagement, which is what ultimately determines whether a student succeeds.
Predictive analytics isn't magic. It only works when a college uses the data as a basis for changing the way it interacts
with students. The University of South Florida, for instance,
which Civitas began working with in 2014, created a "persistence
committee" representing different parts of the university-
24

September/October 2019

from faculty to student affairs-to meet weekly to review student data and identify red flags. For students who raise alarms,
the committee dispatches an academic advocate to figure out
what's going on and develop a remedy. It seems to be working.
In 2011, USF's six-year graduation rate was 52 percent. Now it's
73 percent.

The Texting Revolution
While predictive analytics can identify students who need help,
providing that help on a mass scale is expensive. It's simply
not possible for most of the schools that serve lower-income
students-which tend to be poorly funded open-access public
universities and community colleges-to hire a lot more trained
guidance counselors. And so the race is on in higher education to
find low-cost digital solutions.
One of those involves something called chatbots. If you've
ever sought help from your bank or Apple using their online chat
function, you've probably experienced this without knowing it:
your initial questions are being answered by an algorithm. It's
only when those responses prove insufficient that a human being is brought in.
Georgia State was the first to apply this automated live
chat service in a college setting, allowing students to ask questions, whether about financial aid or enrollment deadlines, and
receive an answer almost immediately. The technology also addresses the fact that, as university president Mark Becker put it,
"Students are vampires, and they tend to ask questions at one
o'clock in the morning. And we tend not to have our financial aid
staff sitting at their computers at one a.m."
Students who seek information on their own are one
thing. It's those who don't, but should, who are most likely to
drop out. The most cost-effective way to find these students is
not in the dining hall or the dorms or even the classroom. It's
on their phones. That's why Georgia State began using automation to send text messages to students with critical reminders
about deadlines and opportunities, and, when necessary, to let
them know they have an outstanding assignment that needs to
be turned in.
Another pioneer in connecting via text is Trinity Washington University, an urban campus in Washington, D.C.
Trinity uses Starfish Retention Solutions to notify students
when they've missed an assignment or earned a poor grade.
At the same time, the school sends out "Kudos" notices after they've earned a good grade, or after a professor puts in
a notice through the system. "You had people who actually cared that you were getting everything that needed to be
done," said Deanna Stripling, who graduated from Trinity
last year. "If something needed to be done, they would send
you a reminder."
A number of other schools have adopted similar techniques.
The software company AdmitHub created Georgia State's twoway messaging platform, and now, one of its competitors, Signal Vine, is partnering with schools like the University of Texas-
Austin, Texas A&M, and the University of New Mexico.

Washington Monthly - September/October 2019

Table of Contents for the Digital Edition of Washington Monthly - September/October 2019