Teachers have a lot of demands, and not enough time for as much professional development as they would like. ClassInSight uses sensors to collect data like student attendance, participation, facial expressions, and hand raises. The goal is to use these data to help instructors improve. For now, the work is taking place in college classrooms being taught by student instructors. One of the common issues in these contexts is challenge of getting students to participate in class discussion, especially in technical courses.


graphs of classroom data
Data taken from classroom is turned into a report for teachers to use for reflection.

One way to address this goal is by giving instructors live updates about the state of the classroom in order to help them remember to try different strategies. For example, in one study we showed instructors a tablet that used different colors to represent when they were talking, when the students were talking, and a 3-second sliding animation whenever they went silent. Research shows that when teachers ask a question they generally don’t wait long enough before answering it themselves. Likewise, rather than wait for students to elaborate on their answers or allow another student to speak up, teachers will typically evaluate every student comment immediately. By extending these pauses, however, teachers can dramatically increase their students’ participation, as well as test scores. We showed teachers an orange screen when they were talking. After they stopped, the screen would animate a sliding transition to a different color that represented that they had waited long enough. We found that we could use this peripheral display to get instructors to think more deeply about the amount of time that they wait, and to wait longer in order to try to get their students to talk.


Another way to use classroom data is to support teacher reflection. We use this sensor-driven approach to gather information about how much students are talking, how long the wait time is, what types of questions instructors are asking, and so on. After they are done teaching, we send the instructors data visualizations and ask them questions that support reflection and goal setting, as well as help expose them to new strategies.


This is an online training system with personal data embedded within, kind of like what you would get out of a fitbit. Unlike a step-counter, however, the innovation with this work is that it can guide people through learning and practicing completely new teaching strategies that they may not already know about. By seeing what they did, reflecting on its efficacy, and seeing if they are achieving their interaction goals, instructors can use the app to plan for what they will do in the next session. Over time, we find that those who use these technologies may even increase in their sense of self-efficacy for teaching.

David Gerritsen
Amy Ogen
John Zimmerman
Chris Harrison
Yuvraj Agarwal