Colleges Are Banding Together Digitally to Help Students Succeed. Here’s How.

Imagine data from millions of students, at dozens of universities, that educators could analyze for sweeping trends but also sift down to the level of each participant in each lesson of every class. Now imagine cross-referencing that information with other stats on those students’ backgrounds and demographics. The resulting insights could elevate teaching and student success to a whole new level.

A forward-looking consortium is building just such a database. And as artificial intelligence gets more complex, more promising, and more expensive, such joint ventures could well be the new shape of digital higher education.

The consortium, founded in 2014 with four members, is called Unizin and has grown to include 11 institutions with 25 campuses, collectively enrolling more than 900,000 students. This month the Austin, Tex.-based nonprofit group is set to add two more member institutions as well as two new major partner vendors. Its board is discussing adding as many as three to five new member institutions annually. Unizin plans to serve as an enormous “learning laboratory” to enable colleges, departments, and individual faculty members to shape instruction in ways that best meet students’ needs.

Resources and Leverage

Unizin is the brainchild of Brad Wheeler, vice president for information technology and chief information officer at Indiana University, and James Hilton, now dean of libraries and vice provost for academic innovation at the University of Michigan. They started thinking about it during the height of MOOC mania, around 2012. At the time, many colleges were offering online courses and gathering data on student performance, but they didn’t have the resources to make necessary upgrades. Nor did they have, on their own, bargaining leverage with vendors. And with limited number-crunching power, they couldn’t get the statistical insights regarding student success that they wanted.

Unizin offers members up-to-date shared learning platforms and digital resources at significant savings. For instance, in 2018, Wheeler reports, Indiana’s web offerings with the consortium saved students roughly $9 million in comparison with full-cost texts. And member institutions say they get system upgrades and tools that would be beyond their resources.

But as important or more, says Hilton, is what can be learned from Unizin’s capacity for data collection and analysis. Members have access to the online-learning platform Canvas and its associated data — not only from each member institution but also collective, anonymized data. That will increasingly enable discovery of broad trends from the overall membership, as well as micro-information for every class. Combine that with data on students’ demographics and background, Unizin says, and you have a powerful tool to make predictions, spot problems, and improve pedagogy.

At one level, says Wheeler, the data will help colleges discover, for their campuses and in general, which courses are bottlenecks where students get stuck, and what sequences of courses work best. But more and more, he says, the statistics will also enable departments and instructors to make each lecture hall, classroom, and online extension a minute-by-minute, hour-by-hour learning laboratory.

Take students’ clickstreams and pageviews on the learning-management system, their writing habits, their participatory clicks during classroom discussions, their grades. Then combine that with information on their educational and socioeconomic backgrounds, their status as transfer students, and so on. You end up with “a unique asset,” says Wheeler, in learning what teaching methods work.

Digital publishers have learned a lot about students who use the publishers’ texts and other resources, he says, but the demographic puzzle pieces are key to discovering impediments to learning.

With good data, however, come great responsibility and thorny dilemmas. Machine learning has already made educational predictions eerily accurate, says Jennifer Sparrow, senior director of teaching and learning with technology at Pennsylvania State University, another Unizin member. Given sufficient data, a system can predict within a half-letter grade, 88 percent of the time, how a student will do in a particular course, she says.

Will that information inspire students to buckle down and try harder, seek tutoring, and prove the know-it-all machine wrong? Will it become a self-fulfilling prophecy accepted with a sad shrug? Will it scare students away from a course, or a subject, or a major? By avoiding potential failures, will students never learn about failure, or the resilience that can grow from it?

“By not thinking about this, we are sticking our heads in the sand,” Sparrow says. And so a learning-analytics subcommittee has been, for three years, considering this ethical realm.

Other potential uses of the data are less controversial, she says. For example, if analytics show broad equality in the outcomes of courses from the universities in the consortium, that may topple some needless barriers to credit transfers.

Vendors, while they feel market pressure to offer discounts to consortia, also get good ideas from such large groups of customers, says Jared Stein, vice president for higher-education product strategy at Canvas, the learning-management provider that industry observers say controls about 28 percent of the market. Unizin and other consortia, he says, give Canvas a continuous sense of what tools and data colleges find most helpful, and that influences priorities in updating the system.

As artificial intelligence makes students’ digital learning more sophisticated and expensive, data ownership will become more important, says Nancy Rubin. Rubin is dean of continuing education and distance learning at Northwestern Health Sciences University, in Bloomington, Minn., and a consultant who previously led the University of Northern Colorado Extended Campus and Columbia University’s online-learning program. And as colleges, especially small ones, cut course offerings in some areas, they will increasingly have to join forces to keep up — whether through nonprofit ventures, for-profit companies, or both.

Missing a Course? Borrow Ours

College Consortium is a different kind of organization from Unizin, although the for-profit company, which started less than three years ago, is also in growth mode and also based in Austin. Its soon-to-be 200 member institutions share thousands of online courses. Say a particular class isn’t offered in a given term at a college, or there’s a schedule conflict for a student with sports commitments or a semester abroad. That student can take an equivalent course — once it is approved by the professor and the registrar — through the consortium.

The collective includes members of existing groups like the Council of Independent Colleges and the Kansas and Georgia independent-college associations, and members can filter the offerings. So, for instance, institutions in the 180-member Council for Christian Colleges & Universities can search one another’s courses.

“This is a really nice way to fill in those gaps for the students,” says Brenda Buckley-Hughes, chair of communication arts at Judson University, a Baptist institution in Elgin, Ill. Sierra Chavez, a senior who had switched her major into that department, was three credits short and still trying to reach her goal of earning her bachelor’s in three years. She did so by taking three courses through College Consortium, offered by William Peace University, Lamar University, and the New England College of Business.

Unizin and College Consortium have contrasting objectives and business models (see box at left). Both, though, represent colleges’ effort to customize digital offerings to their individual needs and resources, adding homegrown alternatives to offerings from Coursera, EdX, and other MOOC providers, and to general-education courses through companies like StraighterLine.

Lenny Favara is president of Central Christian College, in McPherson, Kan., which has 325 residential and 400 online students. He calls College Consortium “a tremendous benefit” that allows other Christian colleges to act as “an adjunct teacher” for some courses. Because the courses have been vetted, he says, grades can transfer. It’s “mutually beneficial,” he says, “and — that’s the key — mutually rigorous.”

Conversely, if one of Central Christian’s online courses has only a few people enrolled, the consortium allows it to fill those extra digital “seats.” Favara and Judson’s Buckley-Hughes predict that an increasing number of colleges will need access to such arrangements as some departments shrink or even disappear.

But, says Buckley-Hughes, professors will have to overcome fear of and resistance to the online presence of faculty members from other colleges on their disciplinary turf. Consortia growth, she says, is “coming, but there’ll be a number of professors kicking and screaming.”

With growth also come questions of governance, says Wheeler, Unizin’s co-founder. With two board members from each member institution — one voting and one nonvoting — Unizin’s board will soon exceed two dozen and could start to feel a bit unruly. But Unizin, he says, has been anticipating this next phase, and its members have discussed switching in the not-too-distant future to a manageably small elected board with an increasing network of committees and subcommittees to handle the growing lists of particulars.

There are good models for such transitions, he says: the transformation of Google’s digitization program to HathiTrust, or the management of the open-software programs of Sakai and Kuali.

How big will Unizin get? There’s discussion among board members about that, says Penn State’s Sparrow. But in her view, as long as the growth is accompanied by quality control and careful management, the more members it has, the better its data, its options, and its bargaining power.

“The more, the merrier,” she says.

Alexander C. Kafka is a Chronicle senior editor. Follow him on Twitter @AlexanderKafka, or email him at alexander.kafka@chronicle.com.