Mitigating implicit bias in student evaluations: A randomized intervention

We conduct a randomized control trial to assess the efficacy of utilizing modified introductory language in student evaluations of instruction to mitigate implicit bias. Students are randomly assigned within courses to three treatment arms and shown so‐called “cheap talk” scripts referencing implicit bias, the high stakes associated with student evaluations, and the combination of the two. We analyze both the impact assignment of the treatment has on completion rates as well as the effect on average instructor rating. Our analysis indicates assignment has statistically significant effects on the likelihood of response for those assigned the combined treatment, though the effects are heterogeneous with respect to both instructor and student race/ethnicity and gender. We further find the high‐stakes treatment leads to higher average scores for racial/ethnic minority instructors with no significant effects from the implicit bias and combined scripts.