In the 1980s, a shift toward the social sciences as valid sources of data on education led to the creation of the Programme for International Student Assessment (PISA), an exam for 15-year-olds created by Andrew Schleicher in 2000 and run by the Organisation for Economic Co-operation and Development (OECD). It is administered every three years in 81 countries around the world, including the United States, Germany, Mexico, Spain, Singapore, Zambia, the United Kingdom, and Japan. 

Ripley chose the PISA as a basis for her international comparison because it provides a standard measure of achievement in key subjects and skills. The test was designed to measure students’ aptitude in math, reading, and science, as well as their ability to think creatively, reason with data, interpret texts, solve problems, and apply learning to real‑world situations. The PISA’s design emphasizes these skills as essential in modern economies. Consistent with the philosophy of lifelong learning, the PISA is constantly being improved, expanded, and updated.

Because the PISA does not have set answers, it must be graded by people rather than machines, and this has drawn controversy for its perceived subjectivity. For example, a 2025 study by Norwegian researchers compared question features (such as cognitive processes, text source, format, and text type) across 71 countries to see how these features affect the exam’s difficulty and whether the framework is valid across diverse cultural and linguistic contexts. The study found that while many item features show consistent patterns across countries, some do not, suggesting that cultural or linguistic differences can shift how difficult students find certain tasks (Marcq, Kseniia, and Johan Braeken. “From Framework to Functionality.” Large-Scale Assessments in Education, vol. 13, no. 26, 2025). A 2024 study in the journal Engineering Applications of Artificial Intelligence used PISA data along with interpretability techniques to show how socioeconomic status, resource availability, gender, and regional differences contribute to variation in math performance (Gomez-Talal, Ismael, et al. “Understanding the Disparities in Mathematics Performance.” arXiv, 29 Jan. 2025).

There is also concern about selection bias regarding which students participate and whether systems exclude or undercount those who drop out or do not take the test. Adjusting for bias has effects on a country’s ranking, as shown in a 2025 study (Boussim, Onil. “Correcting Selection Bias in Standardized Test Comparisons.” arXiv, 11 Feb. 2025). When using the PISA as a benchmark, educators and scientists must attend to how testing conditions, sampling, cultural context, language, and item design might influence results, rather than just paying attention to raw scores.