In the United Arab Emirates, British-curriculum private schools and public Ministry schools sit 147.6 PISA points apart in mathematics.
- Research
Ten programmes and thirty-seven studies on how psychological and educational constructs are measured.
- Instruments
Twenty-six instruments with their own scales, scoring and reference data, published free.
- Services
Assessment, programme evaluation and research advisory for organisations and researchers.
Published research
Analyses that can be checked, not simply believed
Nine independent analyses of PISA 2022 microdata are published here, one for each world region, and each leads with the question peculiar to its own country rather than a template applied to all of them. Latvia’s language of instruction. Germany’s tracking. Brazil’s regional gap. The seven parallel curriculum systems operating inside the United Arab Emirates.
Two of the nine conclude that the comparison they set out to make is not supported by the data, and withhold the figures instead of printing them under a caveat. That is the part worth reading. Every analysis also begins by reproducing statistics the data provider has already published for the same population, because if those do not come out right then nothing downstream is trustworthy, and a reader should not have to take that on faith.
Where the method comes from
The instruments on this site are scored and interpreted using methods worked out on other people’s data first, for the OECD, UNESCO, Eurofound and Cedefop, and in the peer-reviewed literature.
The Center’s director led the design and implementation of the OECD Study on Social and Emotional Skills across ten participating sites. His work across international studies collectively involved 250,000+ participants. The doctoral work underneath all of it was on measurement equivalence in cross-cultural survey research, at Tilburg University.
Why this is becoming urgent
Assessment evidence is turning into a legal requirement
For most of the past twenty years, whether a selection instrument could demonstrate measurement invariance was a question of professional standards. Under the EU AI Act it is becoming a question of compliance. The transparency obligations in Article 50 have applied since 2 August 2026. The high-risk obligations covering employment and worker management under Annex III apply from 2 December 2027, which is far enough away to feel comfortable and close enough that a validation study producing usable invariance and adverse-impact evidence has to be designed and commissioned well before it.
Most vendors and employers who will need that evidence currently hold validation documentation written for a different purpose, and a fairness claim that was asserted rather than generated. Reading it honestly, saying what is missing, and designing the studies that close the gap is the work described under scientific advisory and expert witness.
Work with us