Timothy McKay has been teaching large, introductory physics courses at the University of Michigan for about 20 years. For a long time, he had a nagging feeling that students could be doing better in his class. To help them, he developed a system called EÂ²Coach, which sends students study tips and advice based on their individual backgrounds in physics, their goals, and their grades.
Last November, McKay and other University of Michigan faculty and staff received internal funding to form the Digital Innovation Greenhouse, an accelerator, of sorts, for learning analytics tools developed on campus. Participants will test and finesse their creations, and maybe expand them university-wide.
Across the country, colleges are using data to create new tools and processes that help students learn. At Michigan, a selective state flagship, faculty aren't focused on preventing students from dropping out. They're focused on helping students excel. "I think there is no student on this campus that can't have a better educational experience," McKay says.
Here are three innovations the university aims to develop.
The EÂ²Coach System
In a lecture hall that seats hundreds, students can seem interchangeable. But when McKay analyzed data the university collects about his students—things such as their overall grade point average, major, and other science and math classes they're taking—he found that his students came from a range of backgrounds.
He wanted to send every student personally relevant information, and it turned out that his colleagues at the School of Public Health had created a tool for doing just that. The Michigan Tailoring System is free software that users can program to send specific messages to selected groups of people.
"The EÂ²Coach system is 'e squared coach,'" says McKay. "That stands for 'expert electronic coach.'" Since fall 2012, his students have been invited to sign up for the system at the beginning of the semester. They take a short survey that explores their goals, their experience in previous physics classes, and whether they believe they're good at physics.
"We're trying to work on helping them understand what they have to do as a student," McKay says. Throughout the course, students can log on to a personalized website that offers encouragement and study tips based on their survey responses and how their performance in class compares with their peers and the experience of past students. Information might be delivered as a graphic, or as a testimonial.
Early studies suggest that students who use the EÂ²Coach system extensively earn higher grades than they otherwise would. The tool is now being used by introductory chemistry, statistics, and microbiology courses, and reached about 5,200 students per semester last year.
McKay says the system has wider implications. "We also want to start to use computer-tailored communication across the spectrum of interactions between the university and the student," he says, everything from recruiting to academic advising to career services. "It seems obvious that all those kinds of communications ought to be personalized."
Not too long ago, the engineering department asked Steve Lonn's lab for help with a challenge. "By the time the student actually shows up and says, 'I have a problem,' it's often too late to make a difference," Lonn says. Advisers in the department wanted earlier information about a student's progress through a course.
His team looked at student data such as course grades and dug into the university's learning management system, the online portal where students can do everything from participate in discussion sections to download practice problems, and where professors enter assignment grades. Lonn's team then tested to see what predicted a student's success in a course.
The result was Student Explorer. The current version pulls grade data automatically from the learning management system, compares each student's performance with the course average, and factors in how often students log into the course website—a variable that stands in for a student's effort.
It's easy for advisers to use because it classifies students in one of three ways: Encourage (which means the student's doing great), engage (which means the student's falling behind), and explore (which means the student is somewhere in the middle). "We always want the adviser to take some action," Lonn says, whether that's expressing praise or reaching out to see what's wrong.
Moving forward, his team wants to learn from EÂ²Coach's success with crafting messages and figuring out students' goals. One of the toughest things about learning analytics, Lonn says, is figuring out why a student is doing what she's doing. "That's also where an adviser comes in," he says.
Members of the digital-innovation greenhouse would also like to find a way to use information about past students' course selection to advise current students. "The university has a tremendous amount of data about how students pass through this campus," McKay says. "We know who has taken what classes when, and how they've done, and what their background was, and what they've went on and graduated in."
The next iteration of the university's Academic Reporting Toolkit, or ART, may give students access to that kind of information when it's time to pick courses. "We're actually working with student groups to do 'design jams' on this," McKay says. With students' input, faculty and staff can make sure they build something useful.
Learning analytics tools can now draw from an enormous range of data. "Every system out there—whether it be Twitter, Facebook, or one of these homegrown, smaller tools—everything lets off what the director of our lab terms 'data exhaust,' or 'digital exhaust,' " Lonn says. He estimates that more thanr 90 percent of courses at the University of Michigan now have an online component. The difficult part is figuring out, from all that chatter, what really signals a student's likelihood of success.
Next America's Education coverage is made possible in part by a grant from the New Venture Fund.