PISA scores (reading / maths / science)

Known criticisms of this statistic

Every statistic embodies method choices, and the substantive critiques of those choices are part of understanding the number. These are documented positions — each linked to who makes it. Listing a criticism is not endorsing it; we take no side.

• The OECD itself flags the Dutch 2022 result: more students were excluded from the sample than PISA's technical standards allow (mostly students with a disability for whom no adapted test existed), the country is marked with a caution asterisk, and — in the OECD's words — high proportions of excluded students may bias performance results upwards.

  — OECD, PISA 2022 country factsheet for the Netherlands · OECD PISA 2022 factsheet — Netherlands (p. 4)

• The statistical model behind the scores has been challenged in the peer-reviewed literature: a reanalysis found strong evidence that the same test item is not equally difficult for equally able students in different countries, concluding the published country rankings are not robust.

  — Kreiner & Christensen (University of Copenhagen), Psychometrika 2014 · Psychometrika: A new look at the PISA scaling model

• PISA is a low-stakes test with no consequences for the student, and effort varies by country: measures of student effort explain roughly a third of the cross-country variation in scores — rankings mix ability with motivation to try.

  — Zamarro, Hitt & Mendez, Journal of Human Capital 2019 · When students don't care (abstract)

• Method changes weaken cycle-to-cycle comparison: the 2015 switch from paper to computer testing is a documented factor that strongly influences results, so part of a multi-cycle decline can reflect how the test was taken rather than what students know.

  — Sybren Spit, in ESB (2025) · ESB: Dalende leerprestaties

Source & method

Raw snapshot (JSON) · every observation keeps its retrieval timestamp and revision history in the public database.