Gender differences in multiple-choice questions and the risk of losing points

I study the gender differences in performance in multiple-choice questions in a setting where wrong answers are penalized and the objective is to score as high points as possible. I exploit data from an undergraduate level microeconomics course at a Finnish university across a six-year period of 2010 and 2012-2016. The course consists of two equally weighted exams that include both multiple-choice and open-ended questions. The results show that, when controlling for the performance in the first exam, women omit more multiple-choice question items (MCQ) in the second exam than men which, in turn, translates to fewer points. Women do not do worse in open-ended questions that are similar to the MCQ, neither is the probability of them answering incorrectly to the MCQ higher. Hence, gender differences in test results might reflect differences in behavior in a very particular test setting rather than genuine differences in skills.