Tracing the Origins of Gender Bias in Teacher Grades

Abstract

This paper exploits a unique institutional feature of the university admissions process in England, where applications are based on teacher-predicted grades, to compare differences between predictions and actual exam results by student gender. Using newly linked administrative data, we find that boys receive lower grade predictions relative to equally performing girls. A substantial portion of this gap can be explained by girls’ advantage in overall scholastic competence. Once we account for this skill differential, the gender gap in non-STEM fields diminishes significantly, whereas in STEM, it shifts in favour of boys. We interpret the remaining gaps as indicative of potential gender bias in schools.