Abstract
Multiple-choice exams are widely used in educational settings, whether during course exams or university admissions, carrying important implications on how students are evaluated. A critical evaluators’ dilemma is whether or not to impose penalties for wrong answers. On the one hand, penalties can discourage students from guessing; on the other hand, they tend to disproportionately disadvantage female test-takers, who are found to be more loss-averse than males. The latter finding has been conducive to policy proposals aimed to address and reduce the gender gap in questions skipped. However, little is known about students’ preferences regarding exam penalties, specifically related to both competitive and non-competitive environments, and about how these preferences depend on relevant personal characteristics such as ability, confidence, and gender. To address these questions, we propose a novel experimental design in which we elicit students’ preferences over imposing penalties for wrong answers. We will assess these preferences in both competitive and non-competitive environments.