What Explains the Sustained Decline in U.S. K–12 Academic Achievement That Began Long Before the Pandemic?

This article asks why U.S. K–12 academic achievement has been declining since before the COVID-19 pandemic. Using national and state NAEP data from 2009–2019, along with supporting evidence from TIMSS and SEDA, the study finds that achievement losses began around 2013 and were especially large among low-performing students. Scores at the 10th percentile in eighth-grade math fell by about 0.17 standard deviations—roughly equivalent to 4.5 months of learning. The analysis shows that nearly 40–50% of post-pandemic learning losses were already underway beforehand, suggesting that long-term factors, not just the pandemic, explain much of the nationwide decline.

The Policy Scientist’s Perspective

This article addresses one of the most consequential questions in American education policy: why national achievement gains that persisted for three decades reversed well before the pandemic. The issue matters broadly because it speaks to the long-term capacity of U.S. schools to sustain learning improvements central to workforce quality and social mobility. Wyckoff, a respected education policy scholar, publishes here in the Journal of Policy Analysis and Management, a leading, methodologically rigorous outlet. The data—national and state NAEP scores—are credible and representative, permitting cautious generalization across states. Although the study is primarily descriptive, its analytic precision and synthesis of multiple datasets make it an important and timely contribution. By quantifying that roughly 40% of pandemic-era losses were preexisting, Wyckoff reframes the debate from short-term recovery toward structural causes of academic decline. The clarity of trend analysis and cross-state comparison provide a compelling foundation for subsequent causal research into policy determinants of learning outcomes.

Full Citation and Link to Article

Wyckoff, J. (2025). Puzzling over declining academic achievement. Journal of Policy Analysis and Management, 1–14. https://doi.org/10.1002/pam.70052

Extended Summary

Central Research Question

Jim Wyckoff’s article, “Puzzling Over Declining Academic Achievement,” investigates the origins and nature of the sustained decline in U.S. K–12 student achievement that began well before the COVID-19 pandemic. The study’s central question is straightforward but critical: what explains the nationwide reversal of student performance trends after decades of steady improvement, and to what extent do these pre-pandemic patterns help account for the large post-2020 learning losses? While policymakers and the public have largely attributed the recent drops in achievement to pandemic-related disruptions, Wyckoff’s analysis suggests that a significant share of the decline—approximately 40 percent—can be traced to preexisting trends. His work aims to document the timing, magnitude, and variation of these declines and to assess competing hypotheses about their underlying causes, including weakened accountability systems, reductions in school funding following the Great Recession, implementation of the Common Core, state-level policy divergence, and the broader social and technological environment affecting student learning.

Previous Literature

Wyckoff situates his work in relation to decades of scholarship examining the determinants of educational achievement. Early studies on achievement growth in the 1990s and early 2000s emphasized the effects of standards-based reform, accountability, and increased school funding on improving outcomes. Dee and Jacob (2011) found that the No Child Left Behind (NCLB) Act improved math performance, especially among disadvantaged students, demonstrating the potential for consequential accountability to narrow achievement gaps. Similarly, Jackson, Wigger, and Xiong (2021) linked increases in school spending to measurable gains in NAEP scores. Yet, by the mid-2010s, research began to note stagnating or reversing progress, with analysts attributing the change to various social and institutional developments—among them the waning influence of NCLB, the fiscal constraints of the Great Recession, and the disruptive effects of technology on attention and engagement.

Wyckoff also draws on comparative studies such as the Trends in International Mathematics and Science Study (TIMSS) and the Stanford Education Data Archive (SEDA). Both confirm the pattern of pre-pandemic decline, lending credibility to the interpretation that the downturn is not an artifact of measurement. A 2022 study by Kuhfeld, Soland, and Lewis demonstrated large pandemic-related losses, but Wyckoff builds on this by extending the timeline backward, showing that declining trends were already entrenched. By synthesizing these literatures, he provides a broader context for understanding the reversal of achievement growth as a long-term structural issue rather than an acute shock.

Data

The analysis relies primarily on data from the National Assessment of Educational Progress (NAEP), both national and state-level, covering grades four and eight in reading and mathematics from 1990 through 2024. Wyckoff uses disaggregated percentile data (10th, 50th, and 90th percentiles) to capture the distributional nature of achievement change—allowing him to distinguish between losses among low-, middle-, and high-performing students. He supplements NAEP data with two other major datasets: TIMSS, which provides internationally comparable math assessments, and the SEDA database, which harmonizes state test results to a common scale from 2009 to 2019. These additional sources serve as external validity checks on the NAEP findings.

Wyckoff also integrates fiscal data from the NCES Common Core of Data School Finance (F33) to examine post-recession spending trends, as well as descriptive indicators of student demographics—such as the share of students eligible for special education, English learners, and poverty rates from the Supplemental Poverty Measure. Together, these data sources enable both temporal and cross-state analysis, highlighting variations in the onset, depth, and potential correlates of achievement decline.

Methods

Wyckoff’s approach is primarily descriptive and trend-based rather than causal. He employs linear regression extrapolations of pre-2013 NAEP data to estimate counterfactual trajectories of achievement had the prior gains continued, enabling comparison between predicted and observed performance. This technique provides an approximate measure of the “gap” between expected and actual achievement, expressed in standard deviation (SD) units and converted into months of learning using established benchmarks (Hill et al., 2008).

He then disaggregates results by percentile to reveal distributional effects, showing that low-performing students (10th percentile) experienced the steepest declines. Cross-state correlations are computed to assess the degree to which performance changes at different achievement levels or in different subjects move together. The analysis also matches state-level changes in NAEP scores to changes in school spending, accountability regimes, and other potential explanatory factors.

While the study does not employ experimental or quasi-experimental identification strategies such as difference-in-differences or instrumental variables beyond citing relevant causal studies, its descriptive rigor provides a valuable empirical foundation. The method is designed to characterize the timing and scale of decline, rather than to assign definitive causal weights to any single factor.

Findings/Size Effects

Wyckoff documents a pronounced and widespread decline in academic achievement beginning roughly in 2013, following more than two decades of steady improvement. Between 1990 and about 2013, national NAEP scores rose by approximately 8–10 percent, equivalent to 0.40 SD, representing one of the largest and longest sustained improvements in modern U.S. education. However, from 2013 to 2019, scores fell substantially, particularly among the lowest-performing students. Eighth-grade math at the 10th percentile declined by 0.17 SD—about 4.5 months of learning—while projected counterfactuals suggest total potential losses closer to 0.38 SD, or roughly a full year of learning.

These patterns were not uniform. Fourteen states recorded declines exceeding 0.30 SD in eighth-grade math for students at the 10th percentile, while six states achieved modest gains. Declines occurred across subjects, grades, and demographic groups but were most severe for students already struggling academically. In contrast, students at the 90th percentile generally maintained or modestly increased performance, widening achievement gaps.

When projected forward, the pre-pandemic trendlines explain nearly half of the total learning loss observed through 2024. Specifically, 47 percent of the pandemic-era eighth-grade math decline, and 44–92 percent of the loss in other subjects, were predicted by preexisting trends. This finding suggests that pandemic disruptions intensified but did not initiate the downward trajectory.

Wyckoff evaluates several leading explanations for these declines. The evidence indicates that weakened school accountability following the transition from NCLB to the Every Student Succeeds Act (ESSA) likely contributed modestly—perhaps 0.02–0.04 SD per year in math—but cannot explain reading declines. Reductions in per-pupil funding after the Great Recession, averaging $1,000 per student in real terms, may have reduced achievement by roughly 0.04 SD on average, accounting for about one-quarter of the total loss. Evidence on the Common Core is mixed: Bleiberg (2021) found small positive effects in math among advantaged students, while Song et al. (2022) found negative reading effects of up to 0.10 SD, particularly among vulnerable groups.

Other factors—such as differential state adoption of evidence-based literacy instruction (e.g., Mississippi’s “science of reading” reforms), increases in students classified with learning disabilities or English learner status, and the rapid rise in smartphone and social media use—likely interacted in complex ways. Yet Wyckoff emphasizes that no single cause fits all states or subjects. The heterogeneity across states underscores that the decline reflects multiple converging influences, both policy-driven and societal.

Conclusion

Wyckoff’s analysis reframes the understanding of recent learning loss by demonstrating that academic decline is a long-term structural phenomenon rather than an isolated pandemic shock. The data reveal that achievement growth began to slow around 2007–2009, with sustained declines emerging by 2013. Low-performing students bore the brunt of these losses, and inter-state variation was substantial, suggesting that policy context matters. Importantly, the magnitude of the pre-pandemic decline explains nearly 40 percent of total losses observed after 2020, indicating that remedial efforts focused solely on pandemic recovery are insufficient.

The study’s central implication is that U.S. education systems face a systemic challenge in maintaining achievement gains amid shifting political, fiscal, and social conditions. Although the analysis is correlational, the findings provide a credible empirical baseline for future causal inquiry into how accountability structures, resource levels, and technological environments influence learning. The quality and scope of the data—especially the inclusion of state-level NAEP and validated external measures—make the conclusions broadly generalizable within the U.S. context, though less so internationally due to institutional differences.

While Wyckoff’s descriptive approach cannot isolate specific mechanisms, his quantitative synthesis advances the field by identifying the persistence, scale, and distribution of achievement losses. It builds logically on earlier causal studies demonstrating the importance of accountability and funding but extends their relevance to a new phase of educational decline. In policy terms, the article signals that addressing learning loss requires more than pandemic remediation; it calls for sustained structural reforms—stronger accountability, targeted spending, and modernized instructional policies. The work stands as one of the most analytically clear and policy-relevant contributions to emerge in recent months, setting the stage for the next generation of causal research into the deep roots of America’s academic slowdown.