Description:
We investigate inherent gender bias in frequently used exam designs in higher education. Specifically, we study whether grade penalties for wrong answers in multiple-choice tests hurt female performance. We conducted a field experiment in a real-life final-exam setting, where students were randomly allocated to different exam permutations, which differed in the questions that carried penalties for wrong answers. Our results indicate that penalties do not lead to female underperformance but, on the contrary, women tend to do relatively better when there is negative marking for wrong answers. This performance difference is explained by differences in students abilities rather than gender-related attributes (e.g., differences in risk aversion), as women in our sample are of higher ability. Overall, we find that ability is the main determinant of differential performance under alternative grading schemes