Study co-author Thomas Dee, a professor at Stanford University's Graduate School of Education and director of the Stanford CEPA, said there were two motivations behind the study: the rapid growth of these online classrooms and the lack of discussion around online classrooms as social spaces.
The study was able to isolate and observe instructor biases by placing randomized comments of fictional students in the discussion forums of 124 different MOOCs.
Dee said the classes were taught out of many different schools and by professors at some of the leading and most selective universities in the U.S.
The fictional students were randomly assigned names that connoted a specific race and gender.
“Simply attaching a name that connotes a specific race and gender to a discussion forum post changes the likelihood that an instructor will respond to that post,” the study said.
Dee said he was surprised by the magnitude of the effect of the instructor’s biases.
“Admittedly, in very high enrollment classes, instructors can’t engage with every comment. But still, seeing a doubling of the instructor’s engagement rate for the white male students was empirically kind of stunning to me,” he said.
The study failed to find general evidence of bias in responses from other students. They did, however, find posts by white females were more likely to receive a response from white female peers.
Dee said discussion boards in online classes are an important component for classes, where students can interact with their instructors and other classmates.
“There’s quite a bit of literature out there that indicates that student engagement in classrooms is a really potent force,” he said. “The concern here is that if instructors are biased in their engagement with students, it’s resulting in the full learning potential of students not being unlocked.”
The study brought up several potential answers to their findings, including anonymity of students. But Dee said this approach may have unintended consequences for students in removing their social identities in classrooms.
Another solution is the potential to de-bias instructors, through methods like professional development.
“We end up arguing that our results are most consistent with the notion that folks exhibit implicit bias, this is a pretty well-substantiated phenomenon, and there’s a growing literature that suggests there are promising strategies for eliminating implicit bias,” Dee said.
The third option is exploring the design possibilities that are unique to online classrooms. Dee suggested the potential to provide dynamic feedback to instructors that could improve interaction with students in equitable ways or altering the construction of online learning environments themselves.
On this point of design, Dee stressed a note of optimism.
“I think our capacity to control the design of online learning spaces is much greater than in brick and mortar classrooms,” he said.