Feedback information and exam misplacement: Evidence from a natural experiment

Knowing how one's characteristics compare to those of others is important in every setting of economic decision-making. This paper examines the effect of information provision about relative ability on students’ attainments and repetition of university admission exams. Our identification strategy exploits a natural experiment where high school students are provided with feedback through externally graded standardized national exams one year prior to the high-stakes exams. We use administrative data from Greece that span 8 cohorts, and employing Difference-in-Difference techniques we measure the causal effect of feedback information on exam ranking placement. Next, we measure the impact of exam misplacement on the repetition decision of university admission exams using binary response models. We provide unconditional and conditional evidence that feedback information from sitting similar exams prior to university admission high stake exams improves the predictability of national exam ranking by within-school ranking. In particular, feedback is found to improve the performance and the relative national rank of the better students by 0.2 standard deviations and 4-6% respectively. Moreover, the provision of feedback information on own prior performance in similar exams induces lower repetition rates among the better students one year after graduation from high school. The results are consistently more pronounced for females indicating greater sensitivity to feedback. Our findings are consistent with novel literature on gender differences in grade discouragement.  We also find that low achievers male students are little discouraged by feedback in comparison to the females. Our findings imply that individuals can be uncertain of their relative characteristics when optimizing behavior in a multi-person environment and therefore, feedback matters.