The one country whose PISA comparability the OECD has publicly qualified is the one most often held up as a model

Viet Nam’s results are cited worldwide as proof that a low-income system can perform at OECD level. The OECD’s own country note states that its 2022 reading results are not strongly comparable with other countries and that 2022 cannot be compared with earlier cycles at all.

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

Download the full note (PDF)

Summary

Viet Nam is the standing proof text for the claim that national income does not determine learning outcomes. Its students sit at a socio-economic level far below the OECD average and perform close to it. That comparison is made constantly, and it rests on the assumption that Viet Nam’s numbers mean the same thing as everyone else’s.

The OECD does not make that assumption. Its own PISA 2022 country note for Viet Nam records that the country met the technical standards, and then adds two qualifications that most citations omit: that in reading, the comparability of Viet Nam’s results with those of other countries was not strong, and that the 2022 results are not comparable with past results at all.

What the data shows.

• Viet Nam’s mathematics mean is 469.4 points (3.9) at a mean socio-economic index far below the OECD average, which is the fact that makes the country famous.
• It is also a country of very large internal gaps. Village and rural students scored 51.2 points below city students (8.2), 26.4 net of socio-economic status, and 28.4 net of socio-economic status and region together.
• Students speaking a language other than Vietnamese at home, 6.3% of the cohort and largely from ethnic minority communities, scored 47.9 points below the majority (9.8). Socio-economic status accounts for most but not all of it: 19.3 points remain.
• The 5.8% of 15-year-olds who had not reached upper secondary scored 94.9 points below those who had. Who is still in the system, and at what level, is doing visible work in the national figure.

None of this makes Viet Nam’s achievement unreal. It makes the standard cross-national sentence about Viet Nam less safe than it sounds, and it makes the internal analysis the more useful one: the gaps inside Viet Nam are large, measured on a single national scale, and free of the comparability caveat that attaches to the international ranking.

Why this is the question for Viet Nam

Every country that cites Viet Nam is making a cross-national comparison, and cross-national comparison is exactly what the OECD has qualified here. That is an unusually clean example of the general problem this series exists to point at: a number can be technically sound within a country and still not bear the weight of the comparison being made with it.

The qualification is not obscure or contested. It is printed in the OECD’s own country note, in the section that describes what the results are. It is simply not carried through into the citations, and by the time the claim reaches a policy document three steps downstream it has become an unqualified fact.

There is a second reason this matters for Viet Nam specifically. A country whose measured performance is far above what its income predicts is a country where any measurement question has an outsized effect on the story. If the estimate is right, the achievement is remarkable. If it is partly a comparability artefact, a global policy narrative has been built on it. Both possibilities deserve to be visible.

The data, and how it was handled

6,068 students in 178 schools, representing a weighted population of 939,459 15-year-olds, from the OECD’s PISA 2022 student and school Public Use Files. National mean in mathematics: 469.4 points (SE 3.9).

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: Viet Nam, published 5 December 2023, https://www.oecd.org/en/publications/pisa-2022-results-volume-i-and-ii-country-notes_ed6fbcc5-en/viet-nam_a727c3a8-en.html (read 6 August 2026)

Inside Viet Nam: the gaps are large and are not a comparability question

Whatever the cross-national caveat, comparisons between Vietnamese students sit on one scale administered in one language under one national design. These are the comparisons the data best supports, and they are also the ones a Vietnamese ministry would most want.

Socio-economic status accounts for roughly half the urban advantage. Adding region, so that the comparison is between rural and urban schools in the same part of the country and at the same socio-economic level, leaves 28.4 points.

This is a large gap by any standard, and it is measured with a precision the international comparison cannot match. It is also the number that a rural education programme would be evaluated against, which makes it more useful to Viet Nam than its OECD rank.

The ethnic-minority language gap deserves separate mention because socio-economic adjustment moves it so far: from 47.9 points to 19.3. Most of the disadvantage carried by minority-language students in Viet Nam is the poverty of the districts they live in rather than language as such, which points at a different intervention than a language programme.

The three regions

Viet Nam is adjudicated regionally in PISA, so northern, central and southern estimates are supported by the sampling design rather than being an analyst’s slice.

The regional differences are smaller than the urban and rural difference, which is worth stating because regional targeting is the more common policy instrument and these data suggest it is aimed at the smaller gap.

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.

Given that the note opens on a comparability caveat, it would be incoherent to then compare a questionnaire index across Vietnamese regions without testing it. Mathematics anxiety is the index most frequently invoked to explain East Asian performance, usually as an unexamined cultural claim.

Measurement invariance of the mathematics anxiety block across region. N = 5,935. Estimator MLR. Decision rule after Chen (2007).

Note what this does and does not establish. It establishes that the anxiety index can be compared between Vietnamese regions. It says nothing about whether Viet Nam’s anxiety scores can be compared with, say, Germany’s, and the OECD’s own reading caveat is a reminder that cross-national comparability is a separate and harder question that a within-country cascade cannot answer.

What follows

Three statements are supported and narrow enough to defend.

• Citations of Viet Nam’s international standing should carry the OECD’s own qualification, which is that the 2022 reading results are not strongly comparable with other countries and that 2022 is not comparable with earlier Vietnamese cycles.
• The urban and rural gap is Viet Nam’s largest measured internal divide, at 28.4 points net of socio-economic status and region, and it is measured on a single national scale where no comparability caveat applies.
• The minority-language gap is mostly socio-economic. Adjustment reduces it from 47.9 to 19.3 points, which argues for targeting district poverty rather than language alone.

The analysis that would settle the comparability question is not a secondary analysis at all: it requires the item-level cognitive data and a cross-country differential-item-functioning study, which the public cognitive file supports. That is a research project rather than a note, and it is the single most valuable piece of work anyone could do on Viet Nam’s PISA results.

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.
• PISA covers 15-year-olds who are in school. In systems where a substantial share of the age cohort has left, the measured mean describes those who remain, and this note makes no adjustment for that.
• The comparability qualification quoted here is the OECD’s, reported as published. This note does not independently verify it and does not extend it to mathematics, which is the subject analysed.

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