Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Talent is Everywhere, Mobility is Not: Mapping the Topological Anchors of Educational Pathways

Francisco Ríos, Fernanda Muñoz, Valeria Bravo, Gonzalo Castillo, Inti Núñez, Jorge Maluenda-Albornoz, Carlos Navarrete

TL;DR

The study analyzes administrative records for 2.7 million Chilean students (2021–2024) to map post-secondary trajectories using unsupervised clustering, revealing seven distinct student archetypes within an Educational Space defined by academic performance and family background. PCA and k-means clustering uncover structured regions in this space, linking cluster position to enrollment patterns and cross-regional migration. A logistic regression with cluster fixed effects shows that higher academic performance increases migration propensity, but socioeconomic status strongly moderates this relationship, with high-SES students more likely to relocate to capital institutions. The approach provides a scalable, data-driven framework for diagnosing structural constraints on educational mobility and guiding policy to reduce spatial and social inequality, with implications transferable to other countries using administrative data.

Abstract

The relationship between socioeconomic background, academic performance, and post-secondary educational outcomes remains a significant concern for policymakers and researchers globally. While the literature often relies on self-reported or aggregate data, its ability to trace individual pathways limits these studies. Here, we analyze administrative records from over 2.7 million Chilean students (2021-2024) to map post-secondary trajectories across the entire education system. Using machine learning, we identify seven distinct student archetypes and introduce the Educational Space, a two-dimensional representation of students based on academic performance and family background. We show that, despite comparable academic abilities, students follow markedly different enrollment patterns, career choices, and cross-regional migration behaviors depending on their socioeconomic origins and position in the educational space. For instance, high-achieving, low-income students tend to remain in regional institutions, while their affluent peers are more geographically mobile. Our approach provides a scalable framework applicable worldwide for using administrative data to uncover structural constraints on educational mobility and inform policies aimed at reducing spatial and social inequality.

Talent is Everywhere, Mobility is Not: Mapping the Topological Anchors of Educational Pathways

TL;DR

The study analyzes administrative records for 2.7 million Chilean students (2021–2024) to map post-secondary trajectories using unsupervised clustering, revealing seven distinct student archetypes within an Educational Space defined by academic performance and family background. PCA and k-means clustering uncover structured regions in this space, linking cluster position to enrollment patterns and cross-regional migration. A logistic regression with cluster fixed effects shows that higher academic performance increases migration propensity, but socioeconomic status strongly moderates this relationship, with high-SES students more likely to relocate to capital institutions. The approach provides a scalable, data-driven framework for diagnosing structural constraints on educational mobility and guiding policy to reduce spatial and social inequality, with implications transferable to other countries using administrative data.

Abstract

The relationship between socioeconomic background, academic performance, and post-secondary educational outcomes remains a significant concern for policymakers and researchers globally. While the literature often relies on self-reported or aggregate data, its ability to trace individual pathways limits these studies. Here, we analyze administrative records from over 2.7 million Chilean students (2021-2024) to map post-secondary trajectories across the entire education system. Using machine learning, we identify seven distinct student archetypes and introduce the Educational Space, a two-dimensional representation of students based on academic performance and family background. We show that, despite comparable academic abilities, students follow markedly different enrollment patterns, career choices, and cross-regional migration behaviors depending on their socioeconomic origins and position in the educational space. For instance, high-achieving, low-income students tend to remain in regional institutions, while their affluent peers are more geographically mobile. Our approach provides a scalable framework applicable worldwide for using administrative data to uncover structural constraints on educational mobility and inform policies aimed at reducing spatial and social inequality.

Paper Structure

This paper contains 10 sections, 4 equations, 4 figures, 1 table.

Table of Contents

  1. The All-Around Achievers (Achievers)
  2. The Strivers (Low-Income Talent)
  3. The Atypical Academic
  4. The Privileged Average
  5. The Challenged Environment Academic
  6. The Resilient Underperformer
  7. The Disadvantaged
  8. Individual Performance Assessment
  9. Quality of Education Received
  10. Cross-Regional Migration

Figures (4)

  • Figure 1: A. Centroid-representation of each cluster B. Total cluster composition C. The Educational Space of Chile's Post-Secondary Students.
  • Figure 2: Density maps showing the distribution of careers in the educational space, highlighting the concentration of specific groups in particular disciplines. A) Medicine, B) Education, C) Engineering, D) Law, E) Psychology, F) Business Administration, G) Technical Careers, H) Architecture, I) Anthropology. Visualizations were created using Seaborn.
  • Figure 3: Distribution of student clusters across eight undergraduate programs.
  • Figure 4: Cross-regional migration between 2021 and 2024. A. Full cohort. B. Medicine. C. Law. D. Engineering. E. Technical careers. F. Anthropology. Regions on the Y-axis are ordered by overall migration rate (across all fields).