What is already yours, and mostly unanalysed
Every country in PISA, PIAAC, TIMSS, PIRLS or ICCS receives, on release day, a rank, a headline, and an international report written for dozens of education systems at once. Almost none receives an analysis of its own data at the level its own ministry, press and parliament argue about. The data has been collected and paid for. The analysis that would make it useful domestically usually has not been done.
The part that is rarely offered
National commentary on these surveys routinely treats scale scores and questionnaire indices as though they were comparable across countries, across groups and across cycles, and then builds policy arguments on differences that are partly measurement artefacts. Every deliverable here states which of your comparisons will survive scrutiny and which the first competent critic will take apart. That is the specialism behind this work and the reason to commission it here rather than from a general research firm.
Two levels of work
A rapid response brief, built from the published results tables within days of a release, saying what the result actually means for a minister and for a press officer, what moved and what sits inside the confidence band and is therefore not news, and which comparisons currently being made are defensible.
A full national deep dive, run on published microdata over several weeks. Within country equity decomposition by region, language, socio economic quartile, migration status, school type or gender, with both the raw gap and what survives adjustment. Plausible values and replicate weights handled correctly, so every standard error carries the imputation component as well as the sampling one, which is the single most common omission in national commentary. And a measurement comparability audit of every construct the report compares, run as a staged invariance cascade, with the verdict stated plainly: this comparison you may make, that one you may not.
What is not included
No primary data collection, no fieldwork, no access to data the OECD has not made public, and no analysis that depends on a microdata release that has not happened yet. Availability follows the data calendar rather than the sales calendar, and a deliverable will not be promised against files that do not exist. Every piece of work carries a written statement of what the analysis does not claim, which is as much a part of it as what it does.