Some Unpleasant Consequences of Testing at Length*
研究发现，当数学测试成绩取决于学生的耐力时，更长的测试不一定能更准确衡量学生能力，且性别差异在小学高年级才出现。
Abstract We show that, when performance on a math test depends on students’ endurance, longer tests do not necessarily provide a more accurate measure of students’ competencies or a higher correlation between the ranking of classes (or schools) based on test performance and the ranking based on the underlying competencies. We revisit the finding that female teenagers are better able to sustain performance during a math test by considering younger pupils, who were tested in Italian primary schools during their second and fifth grade. We find evidence of a gender gap in the ability to sustain performance only in the higher grade, suggesting that this gap develops during primary school.