Sex differences in academic achievement are modulated by evaluation type

Studies on sex differences in academic skills have often reported diverging results depending on the type of evaluation used, with girls typically obtaining better school grades and results at national examinations, and boys scoring higher at standardized tests. In this paper, we provide a framework for better understanding and interpreting these differences, integrating previously established factors that affect variations in the gender gap across evaluation types: writing skills, stakes, self-discipline and grading bias. We apply this framework to a dataset containing the results of 23,451 French students in three evaluations characterized by different combinations of these factors: teacher evaluations, national examinations, and standardized tests. We find that, overall, girls show lower performance than boys in mathematics and higher in French. However, this main effect is modulated by evaluation type: relative to boys, girls over-perform in teacher evaluations and under-perform in standardized achievement tests, compared to national examinations. These effects are larger in mathematics than in French. These results offer new insights regarding the extent to which writing skills, stakes, self-discipline and grading bias may influence the observed gap.