How do learning environments help people learn? To find out, Ryan Baker, Associate Professor of Cognitive Studies, is using advanced computer technologies to sift data on learners’ behavior generated by intelligent tutor¬ing systems and other online learning environments. Such data mining can reveal why a student fails to subtract properly, pinpoint the killer homework problem that stumped the class and identify the best science curriculum for a district.

In graduate school, Baker sought to learn when students went off task and why. He developed software to analyze their clicks and keystrokes in online math lessons. He is currently co-directing an eight-year study to see how choices stu¬dents make using an intelligent tutoring system affect long-term outcomes, such as whether they go to college.

Now Baker is focusing on MOOCs (mas¬sive online open courses), introduced in 2012 by Harvard and other leading uni¬versities. Last fall his MOOC on educational data mining enrolled 44,000 students. Baker has mined data from the course to address questions that preoc¬cupy MOOC instructors everywhere: What is the goal of MOOC education? What draws so many potential students to a MOOC? How effective are MOOC feed¬back systems? And how does MOOC instruction compare with in-classroom learning? MOOC lectures are “mostly pretty good,” Baker says, but instructors need to assign better homework.

Anyone up for grading 44,000 papers?

— Barbara Finkelstein

Published Monday, Jun. 2, 2014