When outliers become evidence: Rethinking 'LEAP gains' in LEPT performance
Recent data briefs from the Teacher Education Council (TEC), particularly AnalyTECs Data Brief No. 6 on “LEAP gains” in the Licensure Examination for Professional Teachers (LEPT), make an important contribution to evidence-informed dialogue in Philippine teacher education.
By presenting licensure data in an accessible and visually engaging format, the brief helps draw attention to post-pandemic changes in TEI performance and invites stakeholders to reflect on recovery, resilience and institutional improvement.
This initiative is valuable, especially in a policy environment where timely and understandable data are needed to support decision-making. At the same time, while the narrative of post-pandemic recovery is compelling, a closer reading of the TEC data reveals significant methodological and interpretive concerns.
At the core of the issue is representativeness. The TEC brief derives its claims from only three out of 1,651 TEIs at the elementary level and five out of 2,492 TEIs at the secondary level. This is less than one percent of all institutions.
More critically, the selection criteria for these institutions are not made explicit, raising concerns about potential selection bias in how “gains” are identified and reported. Yet these cases are used to suggest broader patterns in teacher education performance.
This approach effectively elevates outliers into proxies for system-wide evidence. While it is reasonable for TEC to highlight high-performing institutions, presenting them as indicative of broader trends risks misrepresenting the state of teacher education. These are not trends; they are exceptional cases within a much larger and more varied system.
The problem is compounded by the instability of the metrics used in the TEC analysis. Passing rates are frequently calculated from very small and highly variable numbers of examinees.
In several instances, denominators range from as low as one to fewer than twenty examinees. For example, a change from 0 percent (0 out of 1) to over 70 percent in subsequent years appears dramatic but is statistically fragile.
When cohort sizes fluctuate significantly, percentages become unstable and potentially misleading indicators of performance.
Without reporting confidence intervals or minimum cohort thresholds, these figures risk overstating the magnitude and reliability of improvement. Without appropriate weighting or normalization, the apparent “gains” in the TEC data may simply reflect small-sample volatility rather than genuine institutional improvement.
The temporal framing in the TEC brief also raises concerns. The analysis compares 2019 with post-pandemic years (2021–2024), but 2020 was a non-testing year because the LET was not administered during the pandemic. This creates a discontinuous timeline that complicates direct pre- and post-pandemic comparisons.
Moreover, the post-2021 period reflects not only recovery but also changes in examinee participation. Deferred test-taking, disrupted cohort flows, reduced cohort sizes and selective endorsement practices may have altered the composition of examinees. As a result, improvements in passing rates may partly reflect changes in who took the exam rather than improvements in institutional quality.
Equally important is the absence of a comparative baseline in the TEC analysis. The brief does not benchmark the selected TEIs against national averages, regional distributions, or comparable institutions. Nor does it assess whether the observed improvements exceed pre-pandemic trends.
Without such comparisons, it is difficult to determine whether these “LEAP gains” are exceptional or simply part of a broader recovery pattern. The attribution of these gains to “policy adjustments” is therefore not empirically established by the TEC data. While policy changes may have played a role, the analysis does not establish a causal link. At best, the relationship is correlational; at worst, it is speculative.
The TEC brief also risks committing an ecological fallacy, where institutional-level passing rates are interpreted as evidence of improved teaching or learning. This conflation collapses institutional performance into assumptions about instructional quality without sufficient evidence.
Passing rates are influenced by multiple factors, including student selection, review practices and institutional decisions about who is allowed to take the exam. In contexts where cohort sizes are small, institutions can significantly influence outcomes through selective endorsement. Without disaggregated data on examinee characteristics or institutional practices, the improvements reported by TEC cannot be confidently attributed to enhanced teacher preparation.
From a visualization standpoint, the line graphs used in the TEC brief further amplify these issues. Line graphs imply continuity and stability, yet the underlying data are uneven and based on inconsistent sample sizes.
The visual smoothness of the trends masks the underlying volatility and exaggerates the appearance of the data, giving the impression of sustained improvement. A more appropriate representation would foreground raw counts, variability, and uncertainty, allowing for a more cautious interpretation of the findings.
Perhaps the most concerning aspect of the TEC brief is the normative leap to policy recommendations. The call for broad stakeholder engagement and even amendments to Republic Act No. 7836 is not supported by the limited and selectively constructed evidence presented.
This risks turning data briefs into instruments of advocacy rather than tools for critical policy analysis. While such reforms may be necessary, they require a stronger empirical foundation. Using a small number of non-representative cases to justify system-level policy changes risks undermining the credibility of evidence-based policymaking.
The framing of these findings as “LEAP gains” further reinforces this overreach. The term implies systemic transformation, yet the TEC data document only a handful of cases with rapid improvements under highly specific conditions.
Such labeling risks prematurely stabilizing a narrative that the underlying data cannot yet sustain. These cases may be valuable for exploratory analysis or as starting points for further research. However, they cannot be taken as indicators of system-wide progress.
A more constructive approach is to treat the TEC findings as hypothesis-generating rather than confirmatory. Why did these particular TEIs improve? What institutional practices or contextual factors contributed to their performance? How do these cases compare with the broader population of TEIs?
Addressing these questions would require more robust analytical designs, including larger samples, longitudinal comparisons and appropriate controls. Only then can meaningful conclusions be drawn about the effectiveness of policy interventions in teacher education.
In the broader context of education reform in the Philippines, the implications are significant. Data briefs produced by institutions like TEC play an important role in shaping policy discourse. However, with this influence comes the responsibility to ensure that data are interpreted with methodological rigor and presented with appropriate caution. When outliers are presented as trends and recovery is framed as reform, the resulting narrative risks obscuring more than it reveals.
In sum, the TEC “LEAP gains” narrative illustrates how easily selective evidence can be overinterpreted. If teacher education reform is to be genuinely evidence-informed, it must resist the temptation to elevate exceptional cases into system-wide narratives and instead commit to analyses that are representative, transparent, and methodologically sound.
Angelo Mark Walag ([email protected]) is a professor at the University of Science and Technology of Southern Philippines Ronilo Antonio ([email protected]) is an associate professor at the Bulacan State University. Nikolee Marie Serafico-Reyes ([email protected]) is an associate professor at the Philippine Normal University (PNU). Arlyne Marasigan ([email protected]) is a professor at PNU. The views expressed herein are solely those of the authors and do not necessarily represent the official positions of the institutions with which they are affiliated.
- Latest
