This paper uses new administrative records to assess the role of general ability in explaining gender gaps in teacher-assigned grades across ten “university-preferred” STEM and non-STEM subject areas. The evidence comes from England, where A-level students apply to university using teacher predictions rather than exam results. We find that, conditional on exam grades, boys receive less favourable predictions from their teachers. However, this differential grading is substantially reduced when accounting for gender differences in general ability. In STEM, the gap is rather reversed, with a grade penalty identified against girls with similar general ability and achieved grades at A-level. Our findings provide evidence that teachers are not neutral to students’ attributes captured in our measure of general ability, underscoring the serious implications of relying on predicted grades for university applications instead of exam results.

The analysis is based on newly linked administrative data that include all university applicants, providing comprehensive details on their applications, exam results, and key socio-demographic characteristics. Our empirical strategy follows two main steps. First, we investigate whether there are systematic differences between predicted grades and exam results by student gender across fields of study. Second, we examine whether these differences can be explained by variations in boys' and girls' general ability, extending beyond subject-specific proficiency. To gain a deeper understanding of what drives these disparities, we investigate a range of factors potentially linked to general ability, as well as the predicted-achieved grade gap, including individual student characteristics and aspects of the application process.

We find substantial gender gaps in predicted grades, conditional on achieved grades. Consistent with previous research, these gaps favour girls and are evident across all levels of the achieved grade distribution. Similar to Lavy (2023), we observe more pronounced gender differences in non-STEM subjects, with less pronounced gaps in STEM. The results remain consistent across alternative specifications and robust against a range of potential issues, including measurement error in exam scores, statistical discrimination, and sample selection biases. After adjusting for gender differences in general ability, the gender gap in predicted grades against boys is substantially reduced in non-STEM. In STEM, the gap is rather reversed, in favour of boys.

Lavy, V. (2008). Do gender stereotypes reduce girls’ or boys’ human capital outcomes? Evidence from a natural experiment. Journal of Public Economics, 92(10):2083–2105. Lavy, V. and Sand, E. (2018). On the origins of gender human capital gaps: Short and long term consequences of teachers’ stereotypical biases. Journal of Public Economics. Lavy, V. and Megalokonomou, R., 2023. The Short-and the Long-Run Impact of Gender-Biased Teachers. American Economic Journal: Applied Economics Breda, T., Ly, S.T., (2015). “Professors in core science fields are not always biased against women: Evidence from France.” American Economic Journal: Applied Economics 7 (4), 53 75 Burgess, S., & Greaves, E. (2013). Test scores, subjective assessment, and stereotyping of ethnic minorities. Journal of Labor Economics, 31(3), 535–576. Carlana, Michela. (2019). “Implicit Stereotypes: Evidence from Teachers’ Gender Bias.” The Quarterly Journal of Economics 134(3): 1163–1224. Cavaglia, C., Machin, S., McNally, S., & Ruiz-Valenzuela, J. (2020). “Gender, Achievement and Subject Choice in English Education, Paper” Prepared for Oxford Review of Economic Policy issue on Gender