Colleges and universities are doubling down on learning analytics. They’re trying to figure out how to better use the rich data they’re increasingly capturing about their students and how to improve our collective understanding of the impact of analytics on teaching and learning. At the University of Michigan, more than 100 learning analytics scholars from around the world met earlier this summer for the third annual Learning Analytics Summer Institute (LASI).
I sat down with event co-chairs Stephanie Teasley and Tim McKay to talk about how to use data to improve teaching and learning and how the field of learning analytics is advancing. Teasley is a research professor and faculty director of the Learning, Education & Design Lab at the University of Michigan’s School of Information and also the president-elect for the Society for Learning Analytics Research (SoLAR). McKay is the Arthur F. Thurnau Professor of Physics, Astronomy and Education, and faculty director of the university’s Digital Innovation Greenhouse.
“To be able to personalize education at scale is central to [the University of Michigan’s] mission and data is central to being able to do that,” McKay said. Here, he and McKay share how learning analytics is taking shape on their campus and beyond.
What’s changed about the learning analytics movement in the last five years?
TM: The movement has evolved in significant ways, including the ability to access relevant data beyond what’s traditional as well as the emerging understandings of the ways in which that data might be used. Almost every institution is now delivering a large fraction of their education in digitally mediated ways. Not just online education—all of it is happening through digital tools, and that’s true on every campus. Fifteen years ago this type of data was only available to people doing online education and now it is available to everybody in every learning environment. I think you’re seeing traditional campuses look toward learning analytics in a much more aggressive way than was the case five years ago.
ST: Some of the earliest players in learning analytics came from online institutions like Athabasca in Canada and Open University in the UK. The work that we saw then really focused on dashboards and providing information to remote students. As the field is growing and maturing we’re increasingly looking at residential institutions, and adding more learning theory and knowledge from the field of human computer interaction for designing analytic displays. This notion of personalization being more than just displaying feedback to learners is definitely a trajectory that we’re on.
What are some of the questions that are emerging as we gain greater access to data?
TM: Some of the most exciting are ways in which we might use data to understand every student much more deeply. To give a simple example, students do a tremendous amount of writing in college, and that writing is very rich in information about their understanding of topics. In the past, that writing was committed to a piece of paper and handed to another individual who assessed it and then handed back and it would disappear. There was no way to treat it as data. To connect papers the student was writing in their application package to writing they do in their final classes as they graduate would be something really new. It would let us form a portrait of that student and how their understanding of things has grown. That’s the kind of data which will let us understand what's happening in higher education much more richly than we do today.
ST: In the last 20 years we've really come to understand how learning happens in a way that is different than it was conceptualized at the peak of cognitive science. The cognitive revolution thought very mechanistically about how students learn content. We now understand that learning is part of a larger environment, and individuals react and act in those different environments in very different ways. Now we have access to very rich and very intensive data about learning and have far more sophisticated techniques for investigating that data. So it provides a new opportunity for an evidence base of what constitutes learning from a theoretical perspective.
Who benefits from significant investment in learning analytics?
ST: The students and the instructors benefit. Students benefit because we can change the course of their learning throughout their experience starting from the very beginning by helping them make course corrections, understand best practices and to become better students. We can also improve how we deliver instruction and to partially (if not completely) move away from the “sage on the stage” lecture format and rather support other kinds of practices that help students learn more deeply in a way that's accessible to them as individuals.
TM: One of the things that can come out of learning analytics, too, is an opportunity to give students a hands-on training in making evidence-based decisions. We would like to put all the data that we have to work in helping students make their educational decisions while on campus. Later in life, they can go out and be citizens, workers and ultimately individuals who are able to marshal evidence and use it in making decisions. By doing that throughout the course of their education they’re learning the kinds of skills and habits of mind they need to make good choices throughout life.
What would you say to individuals who believe there's a tension between predictive models in learning and serendipitous learning?
ST: People often associate predictive analytics with closing doors and I like to think that it's a way to open doors. It gives you some idea about which paths are open to you and how these lead to the next opportunity. It offers a different view of what could be and not what has to be based on what some data set said about who you are.
TM: Remember predictive analytics are not really predicting things. They're really looking at what has happened to people in the past and giving us the opportunity to learn from that experience. It's all about learning from the experience in the past and making the future what you want it to be. It might be that you discover from learning from experience that a student with your kind of background has very rarely accomplished the goal that you're after. It's best to know that if you want to accomplish that goal. You can still do it, but now you know that you’ll be a pioneer. It’s about learning from experience in order to change the future rather than learning from experience to predict the future.
How can learning analytics tackle inequity in college access and persistence?
ST: We're hoping to democratize that experience in a different way for students who come to Michigan. My hope is that we can make it so that you don't have to know the right person to know whether you should take this class next or what opportunities are for you with this particular major. We can really help more of our learners understand how to find their way through our curriculum without relying on a social network that disadvantages some students.
TM: What we're trying to do with learning analytics is to enable every student to learn from the experience of every student who has come before, instead of from the two or three people that they talk to about their prior experiences. If a student is wondering or thinking about majoring in physics or majoring in math, they want to be able to look at the experiences of students who followed those paths in the past and weigh the information that they find there.