🔬 physics

Active Learning vs Traditional Lecturing in Introductory Mechanics: A Pooled Pass-Rate Benchmark Under Common Departmental Assessments from a Latin American Institutional Change Initiative

This study reports that at a large public university in Mexico, sections of introductory mechanics utilizing active learning achieved statistically higher pooled pass rates on common departmental assessments compared to those using traditional lecturing, providing a transparent benchmark for institutional decision-making despite the lack of isolated causal evidence for specific instructional techniques.

Original authors: Isaac Pérez Castillo, Lidia Jiménez Lara, Orlando Guzman, Juan Ernesto Chavero Amador, Juan Miguel Carrillo Gil, Luis Alberto González Flores, Miguel Angel Morales Olvera, Oscar Enrique Bonfil Urbalej

Published 2026-02-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Isaac Pérez Castillo, Lidia Jiménez Lara, Orlando Guzman, Juan Ernesto Chavero Amador, Juan Miguel Carrillo Gil, Luis Alberto González Flores, Miguel Angel Morales Olvera, Oscar Enrique Bonfil Urbalejo, Eric Burkholder

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

Imagine a massive university in Mexico City as a giant kitchen. For years, the "Introductory Physics" class has been like a tough cooking challenge where many students burn their dishes (fail) or drop out before the meal is served. The kitchen staff (teachers) have been using the same old recipe: standing at the front, talking non-stop, and hoping the students absorb the instructions by just listening. This is what the paper calls Traditional Lecturing (TL).

Recently, the kitchen managers decided to try a new approach called Active Learning (AL). Instead of just listening, students in these classes were encouraged to "cook" along with the teacher. They worked in groups, solved problems together, and used clickers to answer questions on the spot. The teachers acted more like sous-chefs guiding the process rather than just reading from a script.

Here is what the researchers found, explained simply:

The Big Experiment

The university didn't just guess; they ran a transparent test during one specific term (Autumn 2025).

The Setup: They had eight sections of the same physics class. Two sections were the "Active Learning" test groups, and six were the "Traditional Lecturing" control groups.
The Rules: To make it fair, everyone took the same tests. A committee of teachers (who didn't know which class was which) wrote the midterm and final exams. This is like giving every cooking class the exact same list of ingredients and the same final taste test, regardless of who taught them.
The Catch: The researchers were honest about the limitations. They didn't randomly assign students to classes (students picked their own schedules), and the teachers in the "Active Learning" group were actually less experienced with this specific course than the traditional teachers. It was a "real-world" test, not a perfectly controlled lab experiment.

The Results: Who Passed?

The researchers looked at the pass rates (getting a grade above 6.0 out of 10) for everyone who stayed in the class.

The "Listening" vs. "Doing" Gap:
- In the Traditional classes, about 32% to 45% of students passed the exams.
- In the Active Learning classes, about 42% to 67% of students passed.
- The Difference: The Active Learning groups consistently had more students passing. The biggest gap appeared on the final comprehensive exam and the final course grade, where the Active Learning groups had roughly 23 more passing students per 100 compared to the Traditional groups.
The Confidence Check:
The researchers used statistical tools (like a safety net) to make sure this wasn't just luck.
- For the final exam and the final grade, the "safety net" did not touch zero. This means they are very confident that the Active Learning groups genuinely performed better.
- For the earlier midterms, the results were positive but the "safety net" was a bit wider, meaning there was more uncertainty, though the trend still favored the Active Learning groups.

What the "Kitchen" Looked Like

To prove the Active Learning classes were actually different, the researchers sent in "spies" (educational psychologists) to watch the classes using a checklist called COPUS.

Traditional Classes: The teacher did 90% of the talking and writing on the board. Students mostly just listened.
Active Learning Classes: The teacher spent much less time lecturing and more time walking around, guiding small groups. Students were seen working together in teams much more often.
The Metaphor: In the Traditional class, students were like an audience watching a magic show. In the Active Learning class, students were the magicians practicing their tricks with a coach nearby.

The Bottom Line

The paper concludes that in this specific Mexican university, switching to a "hands-on" teaching style helped more students pass the course, even when the teachers were new to the method and the students weren't randomly assigned.

Important Caveats (The "Fine Print"):

It's a Benchmark, Not a Magic Bullet: The study says this proves that "Active Learning as implemented here" worked better than "Traditional Lecturing as implemented here." It doesn't prove that every active learning technique works everywhere.
No Causal Guarantee: Because students chose their own classes, we can't say for 100% sure that the teaching style caused the success. Maybe the students who picked the Active Learning class were just more motivated. However, since the Active Learning teachers were actually less experienced with the course, the fact that they still did better is a strong sign that the method helped.
Real-World Success: The success happened despite the teachers having to learn a new way of teaching from scratch. This suggests that with the right support (training and extra helpers), this method is a viable way to help more students succeed in tough science classes.

In short: When the kitchen stopped just talking about recipes and started letting the students cook, more of them successfully served the meal.

Technical Summary: Active Learning vs. Traditional Lecturing in Introductory Mechanics

Problem and Context
Improving student success in introductory physics remains a persistent challenge, particularly in Latin American contexts where resources are constrained and entering student preparation is heterogeneous. While Physics Education Research (PER) has established the efficacy of Active Learning (AL) over Traditional Lecturing (TL) in North American and European research-intensive universities, there is a significant lack of literature describing the implementation and outcomes of these practices in large, public Latin American institutions. This study addresses that gap by reporting on a pilot institutional change initiative at the Universidad Autónoma Metropolitana-Iztapalapa (UAM-I) in Mexico. The specific problem is to provide a transparent benchmark of student passing outcomes in "Elementary Mechanics I" (EM1) under AL versus TL conditions, using common departmental assessments to ensure comparability across sections.

Methodology
The study analyzes data from the Autumn 2025 trimester (coded 25-O) at UAM-I. The implementation involved eight sections of EM1: two sections taught by instructors trained in AL methods (AL-labeled) and six sections taught by instructors using traditional methods (TL-labeled).

Operational Definitions: The labels AL and TL are operational, reflecting the specific implementation by faculty rather than a rigid, standardized protocol. AL sections involved instructors and teaching assistants (TAs) who underwent a sequence of workshops, material preparation, and ongoing support, utilizing techniques such as group work and facilitation. TL sections were taught by experienced instructors who did not participate in the AL training sequence.
Data Source: The analysis relies exclusively on aggregated, de-identified section-level counts from official end-of-term reports. Student-level data, demographics, and item-level scores were not used.
Assessments: Four outcomes were analyzed: the first midterm, the second midterm, a common global exam (written by an independent committee), and the final course mark. The first three exams constitute 50% of the final grade; the remaining 50% is determined by instructors.
Statistical Analysis:
- Frequentist Approach: Pooled pass probabilities ( $p_{\beta,a}$ ) were estimated for each modality ( $\beta \in \{AL, TL\}$ ) and assessment ( $a$ ). Uncertainty was quantified using Wilson binomial confidence intervals. The primary effect size metric was the Risk Difference ( $RD_a = p_{AL,a} - p_{TL,a}$ ), with 95% confidence intervals calculated using the Newcombe–Wilson method to avoid the unreliability of Wald intervals for proportions near 0 or 1.
- Bayesian Approach: A reference analysis using Jeffreys priors ($Beta(1/2, 1/2)$) was employed to generate posterior distributions for pass rates and risk differences.
- Sensitivity Analysis: To address potential clustering effects (correlation of outcomes within sections) and the lack of randomization, a hierarchical Bayesian model with a random intercept for sections was fitted. This model mapped section-level predictions back to student-weighted pooled estimands to ensure comparability with the simple binomial model.
- Qualitative Support: Classroom practices were monitored using the Classroom Observation Protocol for Undergraduate STEM (COPUS) to validate that AL and TL sections exhibited distinct interaction patterns (e.g., group work vs. lecturing).

Key Results
Across all assessments, the pooled pass rates were higher in AL sections than in TL sections.

Pass Rates: The AL sections consistently outperformed TL sections. For the global exam, the pooled pass rate was 55.6% (AL) versus 32.6% (TL). For the final mark, the rates were 67.7% (AL) versus 45.1% (TL).
Risk Differences: The risk difference ($RD$) was positive for all assessments. The most robust separation was observed in the global exam ($RD = 22.9$ percentage points, 95% CI [5.1, 39.5]) and the final mark ($RD = 22.6$ percentage points, 95% CI [6.6, 37.2]). The 95% confidence intervals for these two assessments did not include zero.
Midterm Exams: The first and second midterms showed positive risk differences (12.3 and 6.2 percentage points, respectively), but their confidence intervals included zero, indicating that the data is compatible with both a positive effect and no effect at the 95% level for these specific assessments.
Robustness: The hierarchical Bayesian sensitivity analysis, which accounted for section-level clustering, yielded results highly consistent with the simple pooled binomial model. The qualitative conclusion that AL sections exhibited higher pass rates remained unchanged.
COPUS Findings: Observations confirmed that AL sections featured significantly higher prevalence of student group work and instructor facilitation, while TL sections were dominated by lecturing and board writing, validating the operational distinction between the modalities.

Claims and Significance
The paper explicitly frames its contribution as a transparent benchmark rather than a causal identification of a specific instructional technique.

Modest Claims: The authors emphasize that the study does not isolate the causal effect of any single instructional technique (e.g., clickers or peer instruction) nor does it claim a universal effect size. The results reflect "AL as implemented" versus "TL as implemented" in a specific, resource-constrained institutional setting. The design is not randomized, and student self-selection into sections prevents adjustment for compositional differences.
Significance:
1. Contextual Evidence: It provides the first documented results of an institutional-scale AL implementation in a large, public Mexican university, addressing the dearth of PER literature in Latin America.
2. Feasibility: It demonstrates that even with a steep learning curve for instructors and limited resources, a supported AL implementation can be associated with higher passing outcomes in gateway courses.
3. Methodological Template: The study offers a practical template for departments seeking to evaluate instructional change using aggregated administrative data. It highlights the value of common departmental assessments for cross-section comparison and outlines a path for future evaluation involving student-level covariates and more robust qualitative analysis.
4. Implementation Factors: The results suggest that implementation supports (training, material preparation, TA staffing) are integral to the "AL" intervention and should be considered part of the change model rather than incidental details.

In summary, the paper concludes that under the constraints of aggregated institutional reporting and a single trimester, the AL implementation was associated with higher pooled passing probabilities, particularly on the common global exam and final mark, providing an encouraging signal for departmental decision-making in similar contexts.

The Big Experiment

The Results: Who Passed?

What the "Kitchen" Looked Like

The Bottom Line

Technical Summary: Active Learning vs. Traditional Lecturing in Introductory Mechanics

More like this