University dropout remains a persistent challenge in higher education systems, yet causal evidence on the mechanisms triggering early disengagement is limited. This study estimates the causal effect of first-time academic failure on subsequent university attrition. Exploiting a sharp institutional grading threshold on a 0-10 scale, we implement a regression discontinuity design (RDD) comparing students who narrowly fail to those who narrowly pass their first attempt. Using longitudinal administrative data spanning multiple cohorts and degree programmes, we estimate local average treatment effects (LATE) for students at the margin of success and examine dropout outcomes within 12 and 24 months following the initial evaluation. Contrary to conventional assumptions, the results indicate that marginal first-time failure is associated with a lower probability of subsequent dropout relative to marginal passing at both horizons. A comprehensive battery of robustness checks - including donut RDD specifications, placebo cutoffs, and formal density tests - supports the validity of the identification strategy. These findings suggest that early academic failure may function as a salient signal that prompts behavioural adjustment or reorientation, while marginal passing may sustain a state of "fragile persistence". The study provides causal evidence on the non-linear effects of early academic performance and highlights the importance of carefully designed institutional responses at critical evaluation thresholds.

翻译：大学辍学仍然是高等教育系统中一个持续存在的挑战，然而关于触发早期脱离学习机制的因果证据仍然有限。本研究估计了首次学业失败对后续大学流失的因果效应。利用一个基于0-10分制的严格机构评分阈值，我们实施了断点回归设计（RDD），比较了首次尝试中勉强不及格与勉强及格的学生。通过使用涵盖多个队列和学位项目的纵向行政数据，我们估计了处于成功边缘学生的局部平均处理效应（LATE），并考察了首次评估后12个月和24个月内的辍学结果。与传统假设相反，研究结果表明，在两个时间范围内，边际性首次失败相较于边际性通过，与后续辍学概率的降低相关。一系列全面的稳健性检验——包括环形RDD设定、安慰剂断点检验和正式密度检验——支持了识别策略的有效性。这些发现表明，早期学业失败可能作为一种显著信号，促进行为调整或方向重新定位，而边际性通过则可能维持一种“脆弱坚持”的状态。本研究为早期学业表现的非线性效应提供了因果证据，并强调了在关键评估阈值处精心设计机构应对措施的重要性。