💻 computer science

Early Academic Capital as the Causal Origin of Dropout in Constrained Educational Systems -- Evidence from Longitudinal Data and Structural Causal Models

Using causal computational social science methods on longitudinal data, this study demonstrates that early academic progress is a more significant causal driver of student dropout than later academic failures, suggesting that interventions should focus on early trajectory formation rather than downstream events.

Original authors: Hugo Roger Paz

Published 2026-04-28

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Hugo Roger Paz

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The "Broken Treadmill" Theory: Why Students Drop Out

Imagine you’ve just joined a high-tech, high-speed gym. The goal is to run a marathon on a specialized treadmill. To keep up with the group, you have to hit certain milestones: every few minutes, the treadmill speeds up, and you have to "unlock" new levels of intensity by completing specific tasks.

Most people look at students who quit the gym and say, "They tripped over a hurdle" or "They failed a specific speed test." They focus on the moment the person falls.

But this research paper argues that we are looking at the wrong thing. The person didn't just trip; they were already falling behind the rhythm of the machine long before they ever hit a hurdle.

1. The Core Idea: The "Time vs. Progress" Gap

The researcher, Hugo Roger Paz, studied over 16,000 engineering students. He found that dropout isn't usually caused by one bad grade or one failed exam. Instead, it’s caused by a "misalignment" between two things:

Clock Time: The months and semesters passing by.
Academic Capital: The actual "points" or subjects a student earns.

The Analogy: The Leaky Bucket
Think of a student as someone trying to fill a bucket with water (their degree) while walking down a moving walkway (the university timeline).

The "Healthy" Student: They are walking at the same speed the walkway is moving, and their bucket is filling up steadily.
The "At-Risk" Student: They are walking, but their bucket has a leak. They are moving forward in time, but they aren't actually accumulating "water" (credits).

By the time they reach the second semester, if their bucket is nearly empty, they aren't just "behind"—they are in a "Capital Trap." Because engineering subjects are like a chain (you can't do Step 5 without Step 1), being empty at the start means they are now locked out of almost everything else.

2. The "Symptom vs. Cause" Problem

The paper makes a brilliant distinction between Symptoms and Causes.

In engineering, there is a famous "bottleneck" subject (like Algebra) that many students fail. Most universities see a student failing Algebra and say, "Quick! Give them a tutor! This failure is why they are dropping out!"

The researcher says: "Wait a minute."

Using advanced math (Causal Inference), he proved that failing Algebra is often just a symptom. It’s like seeing smoke and blaming it for the fire. The "fire" was actually the student's lack of progress in the very first few months.

The Math Proof:

The Real Cause: Having very low progress in the first two terms increases the chance of dropping out by about 26%.
The Symptom: Failing that one big Algebra class only increases the risk by about 13%.

The "real cause" is twice as powerful as the "symptom."

3. The "Early Capital Trap"

Why is it so hard to recover? Because engineering degrees are built like a Jenga tower.

If you don't secure the bottom blocks (the first few subjects) early on, the whole structure becomes unstable. As time moves forward, the "rules" of the university make it harder to catch up. You run out of options, you run out of time, and eventually, the weight of the system becomes too much, and you drop out.

4. The Solution: Change the Timing, Not Just the Topic

If we want to stop students from dropping out, the paper suggests we need to stop waiting for them to "fail" a class before we help them.

The Old Way (Reactive): Wait for a student to fail a big exam, then offer a tutor. (This is like trying to fix a car after it has already crashed into a wall.)

The New Way (Proactive): Watch the "rhythm." If a student is moving through time but not accumulating "capital" (subjects) by the end of their second term, intervene immediately. (This is like noticing the car is running low on oil and fixing it before the engine explodes.)

Summary in a Nutshell

Dropout isn't a series of unlucky accidents; it is a structural mismatch. Students don't usually quit because they hit a single "wall"; they quit because they lost their momentum early on, and the system's rigid rules made it impossible to find it again.

Technical Summary: Early Academic Capital as the Causal Origin of Dropout

1. Problem Statement

The paper addresses a critical limitation in higher education research: the tendency to treat student dropout as a series of isolated, observable academic "events" (e.g., failing a specific course or repeating a subject). The author argues that this event-based perspective conflates downstream symptoms (the failures) with upstream causes (the underlying trajectory deterioration).

In constrained engineering curricula—specifically the Argentine Ciclo Básico Común (CBC)—students must navigate strict prerequisite chains. The study hypothesizes that dropout is driven by a "time–capital misalignment": a decoupling where chronological time (active terms) advances faster than academic capital (successfully completed subjects). This misalignment creates a "structural trap" where early deficits become irreversible due to institutional temporal constraints.

2. Methodology

The study employs a causal computational social science approach, moving beyond predictive modeling (which identifies who might drop out) to causal inference (which identifies why they drop out and where to intervene).

Data: Longitudinal administrative data from 16,868 engineering students at the National University of Tucumán (UNT), Argentina.
Design: A leakage-free panel design focusing on students who survived to their second active term.
Treatment Definition: "Low early academic capital," defined as passing $\le 1$ subject by the end of the second active term.
Outcome Definition: Institutional dropout within a fixed three-year horizon.
Causal Identification Strategy: To account for time-varying confounding (where variables like "current term load" act as both causes and effects), the author uses two distinct, triangulated estimators:
1. G-estimation of Structural Nested Mean Models (SNMM): Targets the additive causal effect ( $\psi$ ) by finding the value where the residualized outcome is independent of treatment assignment.
2. Marginal Structural Models (MSM) with Inverse Probability of Treatment Weighting (IPTW): Creates a pseudo-population using stabilized weights to remove the influence of measured confounders.
Confounders: The models adjust for cumulative subjects enrolled and current term load at $T=2$ .

3. Key Results

The study finds a robust and large causal effect of early capital accumulation on dropout, confirmed by both methodological approaches:

Primary Causal Effect: Low early academic capital increases the probability of three-year dropout by 25.3 percentage points (pp) via G-estimation and 27.4 pp via MSM-IPTW.
Comparison with Downstream Events: The study compared this to the direct causal effect of a common "bottleneck" event: the first-time repetition of the gateway Algebra course. The Algebra repetition had a direct causal effect of only 12.7 pp.
Conclusion of Comparison: The effect of early trajectory misalignment is approximately twice as large as the effect of the most significant academic bottleneck event.

4. Key Contributions

Theoretical: Introduces the Time–Capital Misalignment Hypothesis, conceptualizing educational progression as a dynamic trajectory where early divergence from the expected rate of capital accumulation generates irreversible risk.
Methodological: Demonstrates the utility of advanced causal inference tools (G-estimation and MSM) in educational research to distinguish between predictive signals and causal leverage points.
Empirical: Provides direct causal evidence that the "origin" of dropout is structural (trajectory-based) rather than event-based.

5. Significance and Policy Implications

The paper shifts the paradigm for institutional intervention:

From Symptoms to Causes: Current interventions often trigger after a failure occurs (e.g., tutoring after failing Algebra). The study suggests these are "downstream" and potentially too late.
Early Intervention: The "structural hinge" is the end of the second active term. Interventions should be mobilized at the end of Term 1 for students showing low capital accumulation.
Curricular Design: Because the effect size is so large (~25 pp), the author suggests that the problem may not be individual student capability but rather a mismatch between the curriculum's temporal architecture and student preparation, necessitating changes in prerequisite sequencing and enrollment structures.