Gender Bias in Closed-Ended Questions with Negative Points

Assessment of student performance through closed-ended questions as the so-called multiple-choice questions (MCQ) has become popular in higher education over the last 30 years in a context of rapid growth of the student population. If the rapidity of correction of MCQs is universally admitted, debates exist about their objectivity because of gender bias they may induce, due to different guessing and risk-taking strategies of women students with respect to men. This paper tests whether first-year university male and female students show different abstention behaviors when assessed through negatively marked MCQs leading to a gender gap across exam sessions. Data are collected from mathematics exams and analyzed by a nested logit statistical model using utility maximization principles. It is observed that for comparable students’ proficiency levels, female students abstain more than male students and that this gender gap regularly decreases from the first exam session to the following ones. The relative percentage of mean loss of points for female students on a first attempt exam is around 3% to 4.5% depending on the type of question and the difficulty. This confirms that in contexts where assessment makes extensive use of MCQs, using negatively marked MCQs can lead to a gender bias which questions the goal of fairness within education.