The United Arab Emirates does not have one school system, and its PISA mean is an average over seven

Students in British-curriculum private schools outscore students in Ministry of Education public schools by more than 145 points inside the same country. Adjusting for socio-economic status and emirate removes almost none of it.

Dr Milos Kankaras | ORCID 0000-0002-3190-7751 | Center for Psychology, Podgorica | PISA 2022 country notes, Middle East | August 2026

Download the full note (PDF)

Summary

The OECD’s sampling design for the United Arab Emirates crosses emirate with sector and with the curriculum a school follows: Ministry of Education, American, British, Indian and other. That is not an analyst’s construction, it is the country’s own description of its school system, and it is the single most important fact about interpreting an Emirati PISA number.

Reported as one mean, the country sits below the OECD average. Reported by curriculum, it contains schools performing at the level of the strongest OECD systems and schools performing more than a standard deviation below.

What the data shows.

• British-curriculum private schools averaged 511.0 points; Ministry of Education public schools averaged 365.3. The difference is 145.7 points (2.0).
• Socio-economic status does not explain it. Net of the PISA index the gap is 136.2 points, and net of socio-economic status and emirate together it is 123.4 points (2.6).
• Indian-curriculum schools sit close behind at 490.6 points, and are the most striking case: their socio-economic profile is far below the British and American schools at 0.26, and their advantage over the public system grows to 116.4 points under adjustment.
• Immigrant students outperform nationals, which reverses the pattern seen in most of the world. First-generation students are 48.9 points ahead of native-born students net of socio-economic status and curriculum (2.6), and immigrant students are the majority of the cohort.

The practical consequence is that almost every sentence beginning "students in the UAE" is either meaningless or misleading. The relevant units are the curriculum systems, and they differ from each other by more than most countries differ from each other.

Why this is the question for the United Arab Emirates

The UAE is the largest PISA participant in which a majority of students are in private schools and a majority are not nationals. Roughly half the cohort attends schools following a foreign curriculum, taught largely in English, staffed by internationally recruited teachers, and answerable in part to a foreign inspection regime. The public Ministry of Education system teaches a different curriculum in a different language to a largely national student body.

These are not variants of one system. They are parallel systems that happen to be inside one set of borders, and PISA’s own stratification says so. Any analysis that averages across them and calls the result a national performance level is describing a statistical artefact.

That makes the UAE the clearest case in this series for the point the series exists to make: the number is only as meaningful as the comparability of the things being averaged.

The data, and how it was handled

24,600 students in 840 schools, representing a weighted population of 60,765 15-year-olds, from the OECD’s PISA 2022 student and school Public Use Files. National mean in mathematics: 431.1 points (SE 0.9).

Achievement is imputed rather than measured, so estimates are pooled by Rubin’s rules over all ten plausible values. The sample is clustered in schools, so standard errors come from the 80 Fay-adjusted replicate weights the OECD ships for the purpose. Omitting either understates the uncertainty, usually by a factor of two or more.

Source: OECD, PISA 2022 Results (Volume I and II) Country Note: United Arab Emirates, published 5 December 2023, https://www.oecd.org/en/publications/pisa-2022-results-volume-i-and-ii-country-notes_ed6fbcc5-en/united-arab-emirates_74e92cf9-en.html (read 6 August 2026)

Seven systems, one mean

The full breakdown, with socio-economic profile alongside performance so the obvious explanation can be checked rather than assumed.

Socio-economic status is the first explanation to reach for and it does not survive contact with the table. Indian-curriculum schools sit well below British and American schools on the socio-economic index and above the American schools on performance. Ministry of Education public schools sit above Indian-curriculum schools socio-economically and 125.4 points below them in mathematics.

Holding emirate constant as well changes very little, which rules out the other obvious explanation: this is not simply Dubai being different from Fujairah. Within the same emirate and at the same socio-economic level, the curriculum a school follows is worth 123.4 points at the extreme.

What the data cannot tell us is why. Curriculum, language of instruction, teacher recruitment, selectivity of admission and parental choice are entirely confounded here: a British-curriculum school differs from a Ministry school on all five at once. The honest statement is that the system a child attends predicts their result to an unusual degree, and that PISA cannot say which feature of the system does the work.

The emirates, for comparison

The geographic spread is real but is smaller than the curriculum spread, and it is partly a consequence of it, since the emirates differ in how much of their provision is private and foreign-curriculum.

Dubai leads, and the distance from Dubai to Fujairah is 99.3 points, 90.1 net of socio-economic status. Any federal target set against a national mean will be met or missed for reasons that differ entirely between emirates.

Can this comparison be trusted? A measurement audit

Test scores are placed on a common scale by design. Questionnaire indices are not. Comparing an index across groups assumes the items mean the same thing in every group being compared, and that assumption is testable. It is usually not tested.

This is where the UAE stops being an interesting case and becomes an urgent one. The questionnaire indices are administered in Arabic to some students and English to others, in schools with different pedagogic cultures and different national reference points for what a school is supposed to feel like. Comparing a belonging index across those groups without testing it is not a small methodological sin; it is the whole risk of the exercise.

Measurement invariance of the sense of belonging at school block across curriculum and sector. N = 22,321. Estimator MLR. Decision rule after Chen (2007).

The cascade stops at partial metric, and this is the most consequential result in the note. Equal loadings across the curriculum systems could not be sustained without freeing parameters, and equal intercepts were never reached. The belonging index therefore cannot be compared across the UAE’s curriculum systems at all: the items do not function the same way for a student in an Arabic-medium Ministry school and a student in an English-medium British-curriculum school.

This is exactly the comparison that inspection reports, school marketing and federal dashboards make routinely. It is not supported.

The contrast with the achievement analysis is the lesson. Test scores are placed on a common scale by design and can be compared across these systems; questionnaire indices are not and, here, cannot. Absolute fit is poor as well, with a configural CFI near 0.76, so the index is not a clean single factor in any of these systems.

What follows

Three statements are supported and narrow enough to defend.

• The national mean is not a meaningful description of any school system in the UAE. The relevant units are the curriculum systems, which differ by up to 123.4 points net of socio-economic status and emirate.
• The gap is not socio-economic and not geographic. Both explanations were tested and neither survives; Indian-curriculum schools in particular outperform far wealthier systems.
• Immigrant students outperform nationals by a wide margin even within the same curriculum and socio-economic band, which inverts the pattern most countries plan around and makes the national-student outcome the policy question rather than the immigrant one.

The analysis that would follow this one is a within-curriculum comparison of national students only, which would isolate the outcome the federal system is actually accountable for from the composition effects that dominate the headline. The public file supports it.

What this note does not claim

• No trend statement. Comparing 2022 with an earlier cycle requires the published link error for that cycle pair to be carried in the variance. No such comparison is made here.
• No causal claim. Every difference is an association measured at one point in time. “Net of ESCS” means one measured index is held constant, not that other things are equal.
• The invariance cascade is fitted without the replicate-weight design, which is standard practice and is stated rather than left implicit.
• Curriculum, sector, language of instruction, admission selectivity and teacher labour market are completely confounded in these data. The note reports the association with curriculum system and does not attribute it to curriculum content.
• The curriculum and emirate groupings are read from the published stratum labels. Strata are a sampling construct: they describe the frame accurately but a school’s actual curriculum authority can change between sampling and testing.
• Fee level is not observed. Private schools in the UAE span an enormous fee range, and fee is likely to be a better predictor than curriculum label; the public file does not carry it.

Method and reproducibility

Data: OECD PISA 2022 Public Use Files, downloaded from webfs.oecd.org on 6 August 2026. Point estimates pooled over ten plausible values by Rubin’s rules; sampling variance from 80 Fay-adjusted replicate weights with a Fay factor of 0.5, computed per plausible value and averaged; imputation variance inflated by (1 + 1/M). Invariance tested as a staged configural, metric, scalar and strict cascade in lavaan with full-information estimation for the rotated questionnaire design, plus a categorical sensitivity re-run on pairwise-present data.

Subgroups are derived from variables the OECD codes identically in every participating system, so this note was produced without country-specific recoding. Where a country’s own sampling strata carry a more policy-relevant structure, that structure is used instead and the note says so.

Enquiries about reproducing the analysis are welcome at milos@centerforpsychology.me. Commissioning the equivalent for another country is consulting work, handled by AdriaMont Consulting DOO at milos@adriamont.me.

All country notes in this series