Gender discrimination in exam grading?

Girls, on average, obtain higher test scores in school than boys, and recent research suggests that part of this difference may be due to discrimination against boys in grading. This bias is consequential if admission to subsequent education programs is based on exam scores. This study assesses the causal effect of blind grading, exploiting two separate identification strategies. The first derives from a unique full cohort natural experiment with a grading reform, providing exogenous variation in blind grading. The other strategy derives from a field experiment where the exact same exam papers are scored twice (blind and non-blind). Both strategies use difference-in-differences methods. Although imprecisely estimated, the point estimates indicate a blind grading advantage for boys in essay writing of approximately 5-8% SD, corresponding to 9-15% of the gender gap in essay exam grades. The effect appears to be more pronounced among low performers. Moreover, evaluators tend to give higher grades to boys’ essays when they are led to believe these essays were written by girls. Additional analyses for math suggest a (poorly determined) blind grading effect in favor of girls of 1-3% SD. The overall tendencies are in accordance with statistical discrimination as a mechanism for grading bias in essay writing and with gender-stereotyped beliefs of math being a male domain.