Harvard academic: NAEP shows widespread declines; Alaska included but limited for local diagnosis

Marty West, academic dean at the Harvard Graduate School of Education and vice chair of the National Assessment Governing Board, told the Alaska House Education Committee on Jan. 30 that the National Assessment of Educational Progress (NAEP) is “really the only common yardstick” for comparing student achievement across the 50 states, but that recent NAEP cycles have shown troubling declines. West said the main NAEP tests (fourth and eighth grade reading and math) showed substantial declines on average between 2019 and 2022 and in many cases through 2024, with declines concentrated among lower‑achieving students.

West explained NAEP’s sampling approach—NAEP selects schools first and then randomly draws students within those schools—and confirmed that the 2026 NAEP administration was in the field at the time of his presentation. He told lawmakers that, because NAEP purposely samples a subset of students and splits the assessment across booklets, the program is legally prohibited from reporting at the individual student or individual school level: “we aren’t able, nor are we permitted by law to report results for an individual student or even for an individual school.” West also noted that Alaska did not participate in NAEP’s optional contextual questionnaires, limiting some disaggregation of results for the state.

Committee members pressed West about Alaska’s many very small and remote schools. West acknowledged that a school with a single eligible fourth‑ or eighth‑grade student could end up in the NAEP sample but said the sampling probabilities reflect schools’ share of the state population: small rural sites can be included, but more students live in larger population centers like Anchorage, and statewide averages will reflect that distribution. West added that NAEP items undergo review for accessibility across cultural and geographic differences and that the governing board looks for differential item functioning during piloting to reduce bias.

On the question of what NAEP can and cannot do for state policy, West said NAEP is best used to describe broad patterns and to generate hypotheses for further local inquiry: the data tell "what is happening to student achievement on average over time and from one jurisdiction to another," but are not a substitute for state and district assessments when districts need actionable information to change instruction. He said the NAEP 2026 data collection would wrap up in spring and that national and state results would be released in early 2027.

The committee’s line of questioning highlighted tradeoffs between NAEP’s value as a standardized national benchmark and its limits for local diagnostic use, especially in a state with many remote schools and a high share of correspondence and opt‑outs. West recommended combining NAEP findings with state and district data when diagnosing problems and considering policy responses.