Students Under Surveillance?

James Dragna had his work cut out for him when he became “graduation czar” at California State University at Sacramento, in 2016. The university’s four-year graduation rate sat at 9 percent. It hadn’t moved in about 30 years, he says.

Like many student-success experts at public colleges these days, Dragna combed through academic data about students that the university had on hand — grades, attendance, advising information — to track how they were doing as each semester wore on. He fed those data into predictive-analytics software to look for potential problems or hurdles that might lead to failing grades or dropping out, and to identify students who might benefit from a little extra support.

Three years later, the university’s four-year graduation rate is up to 20 percent. Its six-year rate has risen to 54 percent from 47 percent.

Stories like that dot the higher-education landscape as more colleges take advantage of burgeoning Big Data technology to keep tabs on their students and find more places where they can successfully intervene. But recently, the practice of tracking students has taken a more literal turn.

Sacramento State plans to gather data on where some of its students spend time on the campus and for how long, joining 14 other institutions using software from a company called Degree Analytics. When a tracked student — a freshman who has opted in — enters the student union, her smartphone or laptop will connect to the local Wi-Fi router, and the software will make note of it. When the student leaves and her phone connects to the router in the chemistry building, or the library, or the dorm, it will capture that, too, 24 hours a day, seven days a week.

It isn’t hard to imagine the wealth of observational data such location tracking might produce, and the student-success insights that might arise from it. For example, knowing that A students spend a certain number of hours in the library every week — and eventually communicating that to students — might motivate them to study there more often.

Neither is it hard to imagine that a university that tracks its students, wherever they go on campus, might make a lot of people nervous — even if, for the most part, the students being tracked don’t care.

The more closely students are tracked, and the more personal the data collected, the more potential there is to violate their privacy, whether they know it or not.

Wi-Fi-tracking technology has been used in retail environments for years, and the University of Alabama at Tuscaloosa has used the technology to track students at football games through an app. Students who stay for all four quarters earn extra “loyalty points” and better access to tickets for future games. But academe is supposed to stand separate from the marketplace, and be driven by loftier ideals — student autonomy among them. The good intentions of university administrators focused on student success are running up against concerns about privacy and ethics. How much insight is too much? And when does tracking become surveillance?

Using student data to examine specific theory-based questions about their performance is one thing, says D. Christopher Brooks, director of research at the Center for Analysis and Research at Educause, a nonprofit organization that advocates for technology in higher education. Collecting data on students “for the sake of doing so, or for purposes that may not be clear, I think that’s where we get into this fuzzy ethics space,” he says.

The roots of the tracking trend go back more than a decade, when colleges first realized how much data they had about students, and how little they were doing with it.

A lot of the information “was not integrated in a way that allowed for the efficient or effective use of it to support things like student services or advising and counseling,” Brooks says. “The realization that we had the computing power to begin to match that data up and begin to mine it really sort of perked up a lot of ears.”

In 2009, Purdue University introduced one of the earliest student-tracking initiatives, a homegrown platform called Signals. The university’s computers crunched the numbers on grades, test scores, dining-hall use, and other data points to try to determine which students were on time to graduate and which weren’t. Critics raised questions about whether the program actually increased retention, as claimed. Purdue no longer uses Signals, though it is exploring next-generation analytic tools, according to a spokesman.

Predictive analytics still work more or less as they did a decade ago: Software sorts the enormous amount of information colleges already gather about their students. But student data are still often siloed in different offices, different departments, and different databases throughout an institution. Private companies have largely taken over from university-produced platforms like Signals by perfecting algorithms to aggregate the data into more-useful information, or processing the information to sift for spots where individual students might veer off course.

For example, it would be helpful for an academic adviser to know whether a student was registered for the next semester. If he isn’t, it could be a sign that he is getting discouraged or could use some help. “It seems so simple,” says Edward Venit, managing director of EAB, a technology firm that works with more than 1,500 colleges. “It’s a single data element, yes or no. And the registrar, they have it.” But many college advising offices do not.

Three years ago, at Kennesaw State University, in Georgia, Chris Hutt, assistant vice president for academic advising, used EAB software to create a list of every student enrolled at the university with at least a 3.0 grade-point average who wasn’t graduating and yet hadn’t registered for the next semester. He sent a 34-word email to about 4,000 such students asking if he could do anything to help. He got about 1,300 responses, many from students facing problems as minor as an unpaid parking ticket.

Public universities, which often enroll many lower-income and first-generation students, have been some of the most eager adopters of data tracking, in hopes of better helping students succeed, and improving their retention and graduation rates. They can often point to promising results.

The California State University system has done away with remedial education for its underprepared students, believing that the noncredit courses bog down students and derail their progress toward a degree, but those underprepared students still have to pass sometimes challenging classes. California State Polytechnic University at Pomona ran a pilot program this past summer that tracked student data in four special sections of intensive mathematics for underprepared students.

Relying on early academic markers, such as quizzes, and reports from instructors, the university swooped in with extra instruction for struggling students. Only 57 percent of similar students who took college algebra last fall passed it. In the summer cohort of the class, 70 percent did. In statistics, the pass rate rose to 91 percent from 55 percent. The pilot expanded to 50 sections this fall.

Colleges have been using data from card swipes to track students’ locations on a piecemeal basis, as they enter or leave dining halls or libraries, for some time, but Sacramento State — at the request of system leaders — is taking a giant step in data gathering by using Wi-Fi connections.

This fall’s incoming freshmen received an email in early October from the Student Success Initiatives office explaining briefly what the university wanted to do, and why. The email said that the university would store the data on an encrypted server, and that anonymous data from the project would be kept for five years.

If, and only if, students opt into the program by clicking on a link in the email, a code will be created that corresponds with their IP address and that is connected to their academic record. That code will then be de-identified as far as the university and Degree Analytics are concerned. “We would have no idea who the student is,” says Aaron Benz, founder and chief executive of Degree Analytics. To the company, “it’s just a bunch of random IDs.”

Around late October, the platform will begin tracking those IDs as the students’ devices connect to different routers in different campus locations. The information will not be gathered in real time — “I’m not at a screen looking at where students are at any given moment,” Dragna says. The data will be compiled 24 hours later and sent to the company, which will put its analytics to work and return an analysis of the aggregate data to the university once a semester.

The analysis could help Sacramento State in its drive to create a campus that’s more welcoming to students, in order to encourage engagement and, ultimately, success.

“This will show us students who are going in the library, the times of day that they’re there, the specific classrooms or areas that they are congregating in at particular times,” Dragna says. That may offer a better understanding of which services are most useful, say, or if students with better grades spend more time in a particular building. The university is trying to correlate behavior with performance, not profile individual students, or groups of students, Dragna adds.

Students have been tracked for years at colleges, largely without their knowledge or consent, and not just through their grades or attendance. Data from learning-management systems like Blackboard have long been tapped for insights into student behavior.

Benz stresses that student privacy is a key concern at Degree Analytics, which is also working with 14 other campuses. The company strongly recommends that all institutions they work with establish an opt-in or opt-out policy for students they might want to track.

Dragna met with administrators, faculty members, and students last fall to discuss the possibility of piloting the tracking program. During a conversation with the executive committee of the Faculty Senate, the notion of tracking students’ locations through their cellphones and laptops “raised hackles” among some professors, says Ántonia Peigahi, education librarian and former chair of the senate. “There are some big implications,” she says. “It sounded like a pervasively different way than we’ve thought about technology.”

Everyone involved agrees on one thing: Students didn’t really care.

Students in a focus group at Sacramento State signed off on the program “almost unanimously,” Dragna says. “‘Great. Sounds fine. No big problem.’”

“The student voice was, Yeah, OK,” says Peigahi. “I think we as faculty were expressing a little bit of shock. Doesn’t that bother you?”

Students generally don’t have any problems with being tracked, says Christian M. Landaverde, a senior and the president of Associated Students Inc., the Sacramento State students’ association. That’s because the technology doesn’t pinpoint individual students and the tracking isn’t done in real time, he says. If location tracking can help struggling students persist and graduate, a little data is a fair trade.

Students appear to be uncertain of what, exactly, they’re getting, however. A forthcoming Educause survey found that more than 70 percent of American college students believed that their institution was appropriately safeguarding their personal information and data, but only 44 percent understood how their institution used such information, and only 45 percent believed that it benefited them.

Still, the students now entering college have never known a world where they didn’t swap access to their data for increased convenience or efficiency. “It’s their Instagram, Twitter, Facebook, Google Maps, their email,” Benz says. “So why in the world would my university that I’m paying thousands, or tens of thousands, of dollars to not use my data to help personalize my experience?”

The colleges and administrators that track students want to help them succeed and to raise the retention and graduation rates on which their success, in turn, is judged. But by tracking students and bringing in companies to collect and crunch their data, critics say, colleges violate privacy and data security, whether or not students care.

Student-data tracking is “intensely invasive, extractive, and problematic,” says Chris Gilliard, a professor of English at Macomb Community College, in Michigan, who studies digital privacy, especially as it relates to students. Tracking can expose student data to potential commercial abuse and does so without proven results, Gilliard says. “I think most of it is snake oil. There’s very little independent research that suggests these things do what the companies say they will do.”

And along the way, universities may be training their students to be more compliant with other forms of monitoring.

“We are grooming the younger generation to become used to being tracked and being surveilled and not being bothered by it,” says Sophia Cope, a senior staff attorney at the Electronic Frontier Foundation, a nonprofit organization focused on civil liberties and technology.

It may seem like no big deal if your alma mater wants to keep tabs on you, she says, but so do many private companies and government agencies, “and there are real consequences in those contexts.”

There is potential for good outcomes through data tracking, says Iris Palmer, a senior adviser for higher education and work force at New America, a nonpartisan policy group. But she agrees that collecting students’ location data may go too far, simply because of how invasive it is. Even when students give their consent, such tracking “is pretty invisible and hard for people to sort of get their minds around.”

There’s also the question of whether looking at students in aggregate could obscure the fact that each one has a different situation and different needs. The University of North Carolina at Greensboro has been tracking students’ academic data for five years, and has found it best not to categorize or respond to students in large groups based on data.

“No matter how many variables we try to use to put them in a category, they’re going to defy that category in some way or another,” says Samantha L. Raynor, assistant vice provost for strategic student-success initiatives.

“As much as we might like to be maximally efficient by quantifying everything, it does take a degree of high-touch personal contact” to respond to each student effectively. Recent research has indicated that “nudges” and other interventions meant to guide students to good academic habits may not be particularly effective.

Colleges must also take care that any interventions based on data do more good than harm. If a student is struggling, being identified as such may encourage a self-fulfilling prophecy, or create a stereotype threat, Palmer says. First-generation students, especially, may interpret an attempt at offering help as a sign that they’re not college material.

“That neutral point of contact may itself not be as neutral as we think,” says Brooks, of Educause. “There might be consequences about how the student perceives themselves, and how they adjust their behavior.”

Palmer says she spoke recently to an institutional researcher at a large public-university system who hesitated to provide some student data to administrators because the researcher didn’t “trust that they know how to communicate to students effectively and not punitively.”

Mental-health and disciplinary records are not typically part of student-data tracking — the former are protected by the federal Health Insurance Portability and Accountability Act. But that doesn’t mean some administrators wouldn’t like to see them. Palmer says she’s spoken with college leaders “who are like, We just want to be able to predict who’s going to be the next mass shooter on campus.”

Benz says that Degree Analytics’s technology deployed on other campuses can alert colleges to “erratic behavior.” Say a student who has been identified as at-risk for dropping out is doing well, with perfect attendance, but suddenly stops going to class. “A lot of times it’s things like depression, like mental-health issues,” Benz says.

There are “almost certainly going to be false positives,” he adds. “But what it allows for is basically a conversation. The worst that can happen is you had a conversation with a student.”

The more intimate student tracking becomes, the more questions about privacy, consent, and the use and fate of student data will arise. Brooks sees that as a good thing.

“The early stages of it is like, Can we do this?” he says. “And then realizing that you can, then it becomes, Well, what can we do with this? And I think we’re now getting into the space of, Should we be doing this?”

Lee Gardner writes about the management of colleges and universities, higher-education marketing, and other topics. Follow him on Twitter @_lee_g, or email him at lee.gardner@chronicle.com.