Correlations & Limitations

The claim ladder

1. What the data shows: In the NAEP data, mean mathematics and reading scores for grades 4 and 8 were lower in 2022 and 2024 than they were in 2019.

2. What the data may suggest: The period that included the pandemic is associated with a drop in achievement that has only partly reversed for math and has not reversed for reading.

3. What the data does not prove: That COVID-19 — or school closures, or remote learning — caused the decline. The data document a change over time, not a cause.

4. What would be needed to claim more: Individual-student data following the same students over time, controls for the pre-existing downward trend, and comparison groups not exposed to the same disruption.

Confounding factors

• Pre-existing trends. Reading was already drifting down before COVID (about −2 points from 2015 to 2019), so part of the post-COVID decline continues a trend that started earlier.
• Compositional change. NAEP tests a new sample of students each cycle, so shifts in who is tested can move the average without any student having learned less.
• Broad social and economic disruption. The pandemic coincided with many changes beyond schooling, so "the pandemic" is not one isolated cause.

Limitations

• NAEP is sample-based — the lines connect different groups of students over time, not the same students.
• NAEP is biennial — there are no scores for 2016, 2018, or 2025, leaving gaps in the timeline.
• The 2017 switch from paper to digital testing may affect year-to-year comparisons, though it does not change the broad pattern.
• Texas students are included in the "U.S. national" figure, so the two lines in the comparison are not fully independent.
• This project compares averages, which can hide wide differences within a group.

Because of these limitations, every finding above is phrased as a description of what the data show — not as a causal explanation.