Glossary

Terms used in the charts and notes.

Asterisk (*): The OECD marks a country’s result with an asterisk when one or more of PISA’s sampling standards were not met, for example the response rate or the share of excluded students. Such results should be interpreted with care.
Comparable over time: Results from different rounds can be compared when the OECD has linked them to the same scale and the country’s data meet its standards. Each subject’s trend starts when it was first the main subject. The OECD excludes some country results from trends, for example when the sample did not cover the population as intended; on this site such rounds are shown as separate, hollow points.
Gap: The difference between the average scores of two groups in the same country and round, for example girls minus boys. A gap describes a difference; it does not explain it. Groups also differ in other ways, for example immigrant students are on average more often socio-economically disadvantaged.
Immigrant background: Non-immigrant students have at least one parent born in the country of the test. Second-generation students were born in the country to parents who were both born abroad; first-generation students were themselves born abroad, as were both their parents. Students who did not answer, or (from 2022) answered "I don’t know", are not classified.
Language at home: Whether the language students speak at home most of the time is the language of the test. Up to 2006 other national languages and dialects counted as the test language; from 2009 countries classified languages themselves, so results before and after 2009 are not directly comparable.
Low performers: Students below proficiency level 2. The OECD sees level 2 as the baseline that students need to take part fully in society and to keep learning; it is also the UN’s indicator for minimum proficiency (SDG 4.1.1).
Margin of error: PISA tests a sample of students, and each student answers only some of the questions, so every figure is an estimate. The margin of error is half the width of the 95% confidence interval (1.96 standard errors): the true value is very likely within this distance of the estimate. It combines the uncertainty from sampling students and from estimating their scores.
OECD average: The plain average of the OECD member countries’ results, each country counting equally regardless of size. Because membership and participation change, the OECD uses fixed sets for trends: the average of 23 countries with comparable results in every round, and of 35 countries since 2015 (and in earlier rounds where all 35 have comparable results). The OECD average for a single round covers the members that took part that round and is not meant for trends.
Percentile and median: The 10th percentile is the score that 10% of students fall below; the 90th percentile the score that 10% exceed. The median (50th percentile) splits students into two equal halves. The distance between the 10th and 90th percentiles shows how unequal results are within a country.
PISA: The OECD’s Programme for International Student Assessment. Every three years (four between 2018 and 2022, because of COVID-19) it tests 15-year-olds in school in mathematics, reading and science, and asks them and their schools about their background. Each round has one main subject that is tested in more depth.
PISA score: Results are reported on a scale that was set so that the average across OECD countries was 500 and the standard deviation 100 when each subject was first the main subject (reading 2000, mathematics 2003, science 2006). Scores are not percentages of correct answers. Around 20 score points correspond roughly to a year of schooling, although this varies between countries and grades.
Plausible values: Because each student answers only part of the test, PISA does not give one score per student but several plausible values: random draws from the range of scores the student could have. Every figure is computed once per plausible value and then combined, which also measures how uncertain the scores are. PISA has 5 plausible values per subject up to 2012 and 10 from 2015.
Proficiency levels: Score ranges with descriptions of what students at each level can typically do, from below level 1 up to level 6. The thresholds are fixed for each subject, so the share of students at each level can be compared over time.
Small samples: Figures based on fewer than 30 students or fewer than 5 schools are very uncertain, and the OECD does not report them. They are shown faded here and marked in tooltips and tables.
Socio-economic background (ESCS): The PISA index of economic, social and cultural status combines parents’ education and occupations and home possessions such as books. Students are divided into four equal quarters within each country and round, so the bottom quarter is the most disadvantaged 25% of students in that country. The index has been revised over time; quarters show relative, not absolute, positions.
Statistically significant: A difference is statistically significant when it is too large to be explained by the margins of error alone (here at the 5% level). On this site, a gap is marked as significant when its 95% interval does not include zero. Overlapping bands of two lines are a rough guide only: two estimates can differ significantly even if their bands overlap a little.
Top performers: Students at proficiency level 5 or 6, who can handle complex tasks, for example modelling complex situations in mathematics or understanding long and abstract texts.
Weights and sample: In each country, schools and then students are sampled. Each student carries a weight: the number of 15-year-olds in school they represent. All figures are weighted, so they describe the whole population of 15-year-old students, not just those tested. Standard errors use the 80 replicate weights PISA provides (balanced repeated replication with Fay’s adjustment).