We use the random allocation of graders to different exam questions at Stockholm University to evaluate the existence of same-sex bias in exam correction. We find evidence of same-sex bias before anonymous exams were introduced. Notably, once anonymous grading was in place, the effect disappears.
When separating the effects by grader´s sex, both groups of graders favor male students, although male graders favor male students to a larger extent than female graders. Again, after anonymous grading was introduced, the effect disappears. There is no evidence of compositional changes across the pre-and post-anonymous grading regimes. In sum, our finding is consistent with theories of stereotyping, e.g., the genius being male.
