Learning analytics: It’s one of those things when as soon as your attention is called to it, you start to see it everywhere. Taking a break during the composition of this piece, I sat to watch the Master’s tournament—and up popped an IBM commercial promising to revolutionize the classroom with the power of next-generation data collection and analytics.
Late last week, it appeared in my email in-box: Columbia Teacher’s College is offering the first ever (on the planet) program in Learning Analytics. Listen to the premise: “The fields of learning analytics and educational data mining have emerged with the aim of transforming this data into new insights that can benefit students, teachers and administrators.” It’s becoming inescapable.
Smoke and mirrors, or something legitimately transformative? Hard to say. The term and concept cover such a wide array of ideas and practices that prediction becomes nearly impossible. But there is certainly something here that is capturing a lot of interest, something that may well soon become a much larger part of our work as leaders, administrators, and teachers.
I’m conducting conversations with online educators across the US for a forthcoming Blackbaud E-Book (Summer 2015), on the topic of grading and assessing online student work. In one, Brad Rathgeber, Executive Director of the Online School for Girls, suggested that perhaps my timing for this project is all-wrong. He thinks we are on the cusp of something big: so many significant changes are so coming so fast that my project might very soon be rendered passé. His list of the forces coming soon to “disrupt” grading and assessment include adaptive learning platforms, standard-based grading, gamification, badges, and data analytics.
One of the most notorious examples of analytics in the corporate sector comes from Target stores; many readers probably remember the remarkably sticky New York Times story 0f what happened when a young woman, still living at her parents home, not yet “showing,” and still entirely circumspect about her condition, began receiving in the mail from Target coupons for diapers! Target’s analytics had determined how to identify women early in their pregnancy from their shopping patterns, and to use it to try to capture these valuable consumers for their market.
What are we talking about when we talk about learning analytics? There may be no universal consensus definition, but certainly it includes the idea that we can now and increasingly collect far, far more evidence of what students are doing when they’re in school or participating in learning activities, and we can use this treasure trove to determine far better how much and how well they’re learning, and what’s most effective in helping them learn.
Important to note that this practice is about breaking beyond the boundaries of what we have traditionally collected: the data collection is about so much more than the answers students provide to the questions we ask them. It’s about how much time they spent answering each question; how about how many texts students consulted to generate a response; how many online searches they conducted and how effective those searches were; how effectively they collaborated with others to solve a problem; and how they managed various distractors presented to them during their study time? When students work in or on technologically powered platforms, there is so much more we can learn about how they’re learning.
Both of my go-to annual K-12 schooling forecasts, NMC Horizon and CORE, have been regularly featuring learning analytics prominently. In 2014, NMC placed it in their two-to-three year horizon category, but explained that “for learners, educators, and researchers, learning analytics is already starting to provide crucial insights into student progress and interaction with online texts, courseware, and learning environments used to deliver instruction.” The report cites examples of analytics powering adaptive learning software such as Khan Academy, which is really just the beginning. Another example of analytics in use comes from school districts identifying—and intervening upon– potential drop-outs by determining what combination of student data elements best predict that behavior.
Similarly, the CORE-ED Top ten trends 2014 includes learning analytics on its list, calling it “a really powerful model for personalizing learning for every student… [allowing us] to track their progress, get early intervention information as soon as possible, and to make informed decisions about strategies that are most likely to make a difference for that student. The other crucial impact of learning analytics is the opportunity it gives us to strengthen partnerships between school, the student, and parents.”
Not that any of this is simple—either simple to use or free from ethical complexity. We all have a lot of work to do to determine what evidence is appropriate to collect and preserve, and as much or more work ahead of us to determine how to translate data into action. The CORE-Ed report delves into both of these quandaries; about the second they write, “There’s a great saying about data: it needs to be useful and used. It must be relevant, reliable and meaningful, but it’s pointless to gather data if we’re going to use it. What are your teaching as inquiry processes like in your school? How well is data used when making decisions about what needs to be learnt next and how students might best learn it? Are you drawing on the rich data you have about your students?”
In my own travels of late, I’ve encountered fascinating people putting data to new and exciting purposes. Last spring I met in San Francisco with Vivienne and Norma Ming, both recent Ph.D.s from Carnegie-Mellon, have a startup named Socos doing this kind of work, and have done research at one online university studying student interactions in chat rooms, and are able to predict with high confidence which students won’t complete the course based on their language patterns in the first two weeks of class. Now they are working on wiring classrooms, collecting enormous vats of data about student speech, movement, and interaction patterns, and then dive into this “unstructured” pool of information. “We scan, analyze, and identify: what predicts successful outcomes? What combination of individual words, of vocabulary and syntax, and of strings of words during the term is significantly correlated with higher grades at the end?”
AltSchool, which I observed in action last year, is a school every forward-looking educator should be watching. In its highly personalized learning environments, in which every student is working on his or her own individualized iPad “playlist,” the company engineers recruited from Google are figuring out what data to collect and how they might use what they collect to power accelerated and deeper learning for every student. As was explained in Fast Company magazine, “AltSchool engineers, for example, have developed a specialized video and audio system that records the school day so teachers can bookmark moments to review and analyze later. They’re also thinking about using facial analysis software to provide analytics on the video feeds and designing a smart lighting system that dims automatically when the noise level in a classroom gets too loud.”
And last week I spoke with Melissa DeRosier, Ph.D. , a clinical psychologist who runs a North Carolina research institute and has spent ten years developing games which assess student social skills, including emotional regulation, impulse control, communication, empathy, cooperation, and initiation.
These games (Zoo U for upper elementary, and coming soon, Hall of Heroes for MS students) have no traditional “tests” for any of these skills; the game-play flows naturally, uninterrupted, as players interact with others and seek to complete certain tasks, but every click, every pause, every motion is captured and collected to gather and establish these evaluations, which have proven to be both reliable and valid. Hall of Heroes, by the way, is seeking pilot sites for its research next school year; contact me if you are interested.
This is learning analytics at the classroom level: teachers are using the game to identify students with special needs; to determine which social-emotional learning lessons to prioritize with their particular classroom; and even to determine how to pair students in collaborative project-based learning.
There’s a great deal of promise in learning analytics. Readers, how are you using this practice now and how do you anticipate using it tomorrow? Is it the real thing, or just another fad?