When Students Stop Classes: Measuring the Academic Effects of Student-Led Strikes through Causal Inference in Engineering Chilean Higher Education

• In Event: Politics, Identity, and the Educational Institution

Proposal

Strikes in educational settings have been subject of debate because of their political, economic, and educational implications. Literature has paid particular attention to the causes and effects of K-12 teacher strikes. Examples of research exist worldwide (e.g., USA (Greenglass et al. 2002); Canada (Baker 2013); Colombia (Abadía-Alvarado et al. 2021); Belgium (Belot and Webbink 2010)). However, research is more limited in higher education (e.g., USA (Jacquemin et al. 2020); Nigeria (Ezinne et al. 2024); UK (Braakmann and Eberth 2025)). Analyzing these kinds of strikes is particularly relevant because interrupting the regular learning process during college has direct consequences for the country’s economy and the labor force (Addo et al. 2025). Moreover, even scarcer is the analysis of student-led strikes, which differ significantly from teacher strikes because students autonomously decide to interrupt their regular learning process as a form of activism and civic engagement.
The extremely scarce literature on the consequences of student-led strikes relies primarily on perceptions of students, faculty, and staff (Addo et al. 2025; Awe et al. 2022; Olaniyi and Onajite 2019). The few studies that consider measurable outcomes typically use only descriptive or correlational analyses (e.g., (Jacquemin et al. 2020); for an exception, see (Braakmann and Eberth 2025)). This proposal addresses these gaps by analyzing the impact of a five-week student strike on the academic performance of first-semester students in a highly selective public Chilean engineering institution. While findings are local, the study’s timing and its rigorous design—combining matching with time-series analysis—allow for robust causal conclusions with clear relevance to international debates on the academic consequences of student activism.
The leading research questions are: 1a) To what extent does experiencing a student-led strike affect students’ academic performance? 1b) Does the academic performance before the strike moderate the effect? 2a and b) Are the results equivalent when the effect on the likelihood of passing the courses is considered?
Theoretical framework
This proposal draws from the Self-Regulation Theory (SRT; Panadero 2017; Winne 2018; Winne and Hadwin 1998). This is a pertinent lens for two reasons. First, strikes are disruptions to the regular instructional rhythm required for students’ learning (Rivkin and Schiman 2015). Evidence suggests that forced interruptions negatively affect students’ performance (see (Banerjee and Bharati 2025; Engzell et al. 2021) for COVID examples, or (Sacerdote 2012) for a natural disaster example). Second, student-led strikes are distinctive in that students decide to interrupt their learning process. Thus, differences in how they autonomously organize their time—balancing protest participation, rest, and study—may generate heterogeneous impacts on academic performance.
Methods
The data consist of the academic transcripts of first-semester engineering students at a highly selective public Chilean institution from 2012 to 2018, excluding 2016—a total of 4,850 students. Each record contains grades from six written exams, a pass/fail status in two mathematics courses, and controlling variables known to affect college GPA in STEM fields (e.g., gender, socio-economic status, admission pathway (Celis et al. 2019)).
To achieve high internal validity, a causal inference design (Cunningham 2021) was used. The treatment group consists of students who experienced a strike lasting at least five weeks in 2013, 2015, and 2018 (n = 2,427; 50%). The counterfactual for each strike year was its adjacent year (2012–2013, 2014–2015, 2017–2018). Grades were divided into pre- and post-strike periods. Significant differences across covariates between treatment and control groups were eliminated using 1-to-1 matching without replacement, with a caliper of 0.1 standard deviations (Cunningham 2021).
For RQ1, the outcome measure is the normalized average grade across the two courses in each exam. For RQ2, the outcome is the number of courses students passed (0, 1, or 2). Regarding the analytical design, for RQ1a) a comparative interrupted time-series design with a mean baseline trend model and control variables was implemented (Cunningham 2021). For RQ2a), ordinal logistic regressions with control variables were applied. For subquestions b), additional interaction terms were added. Standard errors and confidence intervals were computed using cluster-robust covariance estimators based on matched pairs.
The main limitation of this design is its external validity. However, the representativeness of the covariate distribution after matching allows generalization to students from the same institution. Furthermore, having several pre-treatment measures makes it possible to assume that additional unobserved confounders (e.g., students’ motivation or study habits) are still controlled for, which strengthens internal validity.
Preliminary results and discussion
The results suggest that in the exam immediately after the strike, there is no significant effect on students’ academic performance. However, in the second exam after classes resumed, there is a negative effect (β = –0.243, p < 0.001). Models with additional interaction terms show the same pattern: no significant differences in the first exam, but consistent negative effects in the second (β_worse_pre_strike_performance = –0.235, p < 0.001; β_mid_pre_strike_performance = –0.256, p < 0.001; β_best_pre_strike_performance = –0.328, p < 0.001).
Regarding the effect on the number of approved courses, results indicate a significant negative impact (β = –0.216, p < 0.001). With interaction terms, the results remain significant and negative for the two groups that were passing (β_worse_pre_strike_performance = –0.057, p = 0.65; β_mid_pre_strike_performance = –0.419, p < 0.001; β_best_pre_strike_performance = –1.264, p < 0.001).
These findings indicate that, in the short term, students’ academic performance—regardless of their prior achievement—is not significantly affected by the strike. However, in the medium term, all groups are negatively impacted. A possible interpretation is that during the strike, students review material taught before the interruption, which helps them in the first exam. Yet, after classes resume, they must learn the new content within a compressed schedule and with a distinct study rhythm than what they had before the strike. This suggests that, although students may exercise some degree of self-regulation during the strike, it is insufficient to restore pre-strike learning patterns once classes resume.
Significance
This research provides causal evidence on student-led strikes, revealing their delayed yet broad academic consequences. It informs global higher education debates on how prolonged, self-imposed learning interruptions challenge the students’ self-regulation they need to sustain performance.

Author

• Juan Pedro Ross Olivares, UNIVERSITY OF CALIFORNIA