Gender gaps under different evaluation methods

Abstract Gender gaps in measured academic performance often influence access to selective university programs, and, consequently, educational sorting and subsequent career trajectories. Differences in the design of high-stakes evaluation methods may therefore affect these gaps by changing how students’ performance is assessed. We examine whether differences in evaluation methods contribute to gender gaps in measured performance using population-level data from a non-self-selected setting in which continuous teacher-assigned grades and a blind, standardized university entrance examination jointly determine university access. Within-student comparisons show that gender gaps in measured performance are approximately 0.46 standard deviations larger under non-blind continuous assessment than under blind, standardized testing. Our findings provide little support for explanations based on competitive pressure or systematic advantages enjoyed by female students under continuous assessment arising from teacher grading practices or differential effort. The results are instead consistent with gender differences in non-cognitive traits that are more fully captured under continuous assessment, as well as with differences in the extent to which students identify with the content being evaluated.