Using Data Science to Transform K-12 STEM Education
As a boy, Paulo Blikstein attended a progressive school run by Madalena Freire, daughter of the Brazilian philosopher Paulo Freire. In line with his philosophical theories, the school disdained grades, textbooks, and learning by rote. Instead, the students read books and wrote poetry and did hands-on projects on whatever topics caught their curiosity. They debated freely with teachers and even helped create the class curriculum.
Blikstein is now an internationally-known expert in educational technology and learning analytics, an associate professor of communications, media and learning technology at Teachers College, and an affiliate associate professor of computer science at Columbia University. He says the school was a “formative experience” that shaped his views on education and influenced his career. “It was a place where it was a pleasure to learn. In my work, I try to come up with ways that reproduce that joyous experience of learning for all students.”
He pioneered the practice of bringing makerspaces to public schools, where students work together on projects in free and open environments—like in his elementary school. Following that success, he created FabLearn in 2010 and built digital fabrication labs in middle and high schools in Russia, Mexico, Spain, Australia, Finland, Brazil, Denmark, Thailand, and the U.S. Overall, 22 countries have used his ideas to create fab labs and makerspaces in schools. His overall objective is to help students learn by building and creating—to learn by “systematic exploration”.
At Teachers College, Blikstein, who received his doctorate in learning sciences from Northwestern University, directs the Transformative Learning Technologies Lab and FabLearn. His team conducts research on how new technologies can improve K-12 learning in science, technology, engineering, and math. One of the team’s major initiatives is to expand the makerspace movement into public schools, particularly less affluent districts that serve underprivileged students. But to do that, Blikstein must first prove to policymakers that makerspaces spur creativity and enhance learning. And to make that case, he needs one essential thing: data.
“Public school systems will only adopt makerspaces if we collect data that measure their success,” Blikstein says. “It’s difficult to measure creativity, but my teams have advanced the research methods to help us do that.”
Blikstein, who joined the Data Science Institute (DSI) this semester, uses a novel data collection method called multimodal learning analytics. In pilot studies, he installs cameras and sensors that detect how students move and work in makerspaces. He also uses biosensors that determine students’ stress levels and eye trackers to track where they focus their attention. Afterward, he applies machine learning techniques and algorithms to mine that data for interesting patterns—patterns about how students move their hands, their eye movements, and their interest levels. His data helps teachers and researchers understand children’s actions: if they’re learning, if they are engaged, or if they are too stressed out to concentrate.
“We analyzed students as they worked on hands-on activities like building robots and found that children who are more active—the ones in whom we detected more hand movements and gestures—tend to learn more. But we were surprised to find that a better predictor of learning was not how active students were, but rather how many times they alternated between being active and passive while working on a project. Our hypothesis is that those moments of physical inactivity are actually for reflection and systematization of knowledge, so one important takeaway from this research is that we need to build moments of reflection into hands-on activities. Not only hands-on, but heads-in.”
Blikstein says he joined DSI because he was drawn to the “data for good” mission and its insistence on the responsible use of data. For example, people are not always aware of how their data is being used, nor how profitable it is to commercial entities. He thinks it is important to teach K-12 students about “surveillance capitalism”, or how some companies harvest personal data to predict behavior and persuade consumers. “We have a generation of middle and high school students who are generating tons of data, but they don’t really understand it.”
Students typically learn tools from the 19th century, but Blikstein says they should be learning modern data science by working on hands-on projects that help society, including studying the spread of coronavirus. “So much of the discussion today about the virus revolves around data science. And the tools scientists use to study the virus—the tracking maps, the curves, and forecasts, or the graphs that illustrate how the virus spreads—are all data science tools,” he notes. “Why not teach students to use data science to understand the science behind the virus, so they can’t be hoodwinked by those who discount the use of science to understand problems like the virus?”
— Robert Florida, Data Science Institute