Is Assessment Gendered?

Link al seminario:

Abstract: Multiple-choice tests are popular to verify how well students meet learning objectives and because grading is unbiased relative to open-ended assessment. We test whether a psychological effect due to scoring barely at or above a nationally used cutoff in the Spanish education system varies, first, depending on whether assessment is multiple-choice or open-ended and, second, by gender. We estimate the effect by using rich data from the compulsory first-year Mathematics course at the University of Alicante by way of a regression discontinuity design and exploit heterogeneity over the type of midterm exam administered in two academic years: an open-ended format in 2018/19 and a multiple-choice one in 2019/20. We find that barely passing a first midterm (a low stake exam) has a positive effect on the final exam score, thus providing evidence on behavioral channels determining educational outcomes. While this psychological effect does not seem to depend on the exam type, we find that this hides gender heterogeneity. Our results point that boys and girls react in a different way to the “more objective” signal of the multiple-choice. These findings can be explained by the combination of several mechanisms related to the way in which the signal is interpreted. On the one hand, multiple-choice tests do not leave space for scores’ manipulation that could allow gender-biased instructors to favor boys. On the other hand, boys, who are known to be overconfident and less risk-averse, tend to answer more questions than girls, and in the multiple-choice format cannot take advantage of points for partially right answers.

Keywords: Continuous assessment, gender, Mathematics, Multiple-choice tests, Open-ended assessment