dorsal/arxiv
View SchemaThe Causal Effect of First-Time Academic Failure on University Dropout: Evidence from a Regression Discontinuity Design
| Authors | H. R. Paz |
|---|---|
| Categories | |
| ArXiv ID | 2601.05987vv1 |
| URL | https://arxiv.org/abs/2601.05987 |
| License | http://creativecommons.org/licenses/by/4.0/ |
Abstract
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.
{
"annotation_id": "729b4bef-beaf-4484-b4ff-7342ac2749a8",
"date_created": "2026-02-17T05:53:05.063000Z",
"date_modified": "2026-02-17T05:53:05.063000Z",
"file_hash": "6f5867b9d821d0140e69964e3d789f0f554b5f85b6ead71faa956cf5d7ce24ab",
"private": false,
"record": {
"abstract": "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.",
"arxiv_id": "2601.05987",
"authors": [
"H. R. Paz"
],
"categories": [
"cs.CY"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"title": "The Causal Effect of First-Time Academic Failure on University Dropout: Evidence from a Regression Discontinuity Design",
"url": "https://arxiv.org/abs/2601.05987",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "c41578c3-4b26-4947-91a0-55237b1c756c",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}