A new and direct test of the ‘gender bias’ in multiple-choice questions

Local and international research has identified a bias in favour of male students with MCQs. If correctly identified, this bias holds implications for reasonable assessment strategies in economics courses. A standard method used in the literature is to relate student performance to various features of the learning environment (such as the type of question) and to student-specific characteristics (such as past performance and lecture attendance). A more direct approach is possible: we set comparable questions (in three categories – graphs, quantitative and theory) in the written and MCQ sections of three tests in the introductory microeconomics course at the University of Stellenbosch. This allows a direct comparison between the performance of male and female students (overall and per question category), without the need to model overall student performance. The number of students in this course, almost 2000, offers a suitably large sample for studying this question. Our evidence does not confirm the strong claims about gender bias in the literature; indeed we find the opposite: a strong positive female gender effect, but for written questions only. We also find no evidence of higher risk-aversion by female students towards MCQ questions with negative marking.