The deadly effect of high-stakes testing on teenagers with reference-dependent preferences

This paper explains why suicidal tendency and test performance of teenagers may not be inversely related when individuals have reference-dependent preferences. Using panel survey data of South Korean secondary school students, I show that the relationship between suicidal ideation and test performance is consistent with reference-dependent preferences. When a student’s rank in the high-stakes College Scholastic Ability Test falls below her average ranks in prior national examinations, she exhibits greater suicidal tendency. The reference dependent effects, however, are absent for low-stakes in-school academic performance. The findings highlight the potential adverse consequences of disappointment in high-stakes testing.