You already work with data. The test asks where the authority comes from.
If you are applying for field 1 you can clean a file, write the join, read a confidence interval. That is not what the field-related test is mostly about. It is about the system around the figure: which body may call a number a European statistic and under which Regulation, what the Code of Practice actually requires of a release calendar, when a correction is a revision and when it is an error, what a quality report has to say about accuracy, and what a statistical service may do with personal data. Under a clock, four precise definitions blur into one and the option next to the right one is the one that gets picked.
The Discipline of the Data is the PassEPSO study book for the field-related test of EPSO/AST/158/26, field 1, Data processing, analytics and statistics: 657 pages in 26 chapters, written so that the things that sound alike stop sounding alike.
The only document that governs this competition is the notice of open competition, Official Journal C/2026/4963 of 6 October 2026, together with its annexes. Its Annex II lists the duties of field 1, and every one of them is covered somewhere in the book. EPSO publishes no syllabus and no weighting between those duties, so neither this page nor the book claims one. Where the book and the notice ever appear to differ, the notice governs — read it in full.
The law behind the figure
Regulation (EC) No 223/2009 and the Code of Practice are the spine of this field, and the questions live in their detail: who the national authority is, what professional independence binds, which criteria belong to which principle. Chapters 1 and 2 are about reading those texts the way a question reads them.
Words that are nearly synonyms
Lineage and reproducibility. Validation at the input and validation at the output. Editing, imputation and harmonisation. A design weight and a calibration weight. Each pair is a mark, and each is lost to a near miss rather than to ignorance.
The half nobody revises
Personal data in a statistical service, statistical confidentiality and disclosure control, open data licences and high-value datasets, structured metadata and quality reports. Dense, memorisable, and squarely inside the duties the notice lists.
What is in the book
Twenty-six short chapters in eight parts, each one assuming the ones before it. Every chapter closes the same way: a one-paragraph spine that compresses the argument, a numbered list of the traps that material produces under time pressure, ten check-yourself questions with answers and reasons, flashcards for retrieval practice, and the primary sources with their article and paragraph numbers so you can verify anything. This is not a question bank — it teaches the material and the reasoning, and it assumes you are practising alongside it.
The statistical system and its rules
The European Statistical System and the law behind an official figure. Professional independence in practice: release calendars, revisions and corrections.
Sources, exchange and pipelines
Exchange with data providers: format checks, validation rules and the agreed standards. Pipelines that can be rerun: staging, incremental loads and reconciliation. Metadata, provenance and processing logs that make a published figure traceable.
Modelling and querying data
Relational modelling: keys, normalisation and referential integrity. SQL for aggregation: joins, grain, and the total that comes out too high.
Quality and validation
Input validation and what to tell the provider. Cleaning, harmonisation and missing values: editing that stays defensible. Output validation: plausibility checks, thresholds and the human confirmation.
Describing data
Summarising a distribution: centre, dispersion and skew. Index numbers, time series and seasonality.
Inference: estimation, precision and analysis
From sample to population: inclusion probabilities, design weights, the unbiased total. Precision: confidence intervals, significance, and what a narrower interval costs. Exploratory and geospatial analysis: area units, denominators, a traceable trail.
Protection, openness and metadata
Personal data in a statistical service: roles, purposes, minimisation and retention. Confidentiality and disclosure control, with the cybersecurity safeguards around them. Open data and reuse: formats, licences, high-value datasets and interfaces. Statistical metadata and quality reports.
Presenting data, tools, automation, AI, and the exam
Charts, dashboards and maps that do not mislead. Reports, briefings and support to expert groups. Scripts and queries you can hand over: version control, testing, reproducibility. Machine learning and generative AI with a control around them. And a chapter on the test itself.
The twenty-six chapters
The table of contents, as it is printed in the book.
- 1. The European Statistical System and the law behind an official figure
- 2. Professional independence in practice: release calendars, revisions and corrections
- 3. Data and metadata exchange with providers: format checks, validation rules and agreed standards
- 4. Pipelines that can be rerun: staging, incremental loads and reconciliation
- 5. Metadata, provenance and processing logs: making a published figure traceable
- 6. Relational modelling: keys, normalisation and referential integrity
- 7. SQL for aggregation: joins, grain, and the total that comes out too high
- 8. Input validation and what to tell the provider
- 9. Cleaning, harmonisation and missing values: editing that stays defensible
- 10. Output validation: plausibility checks, thresholds and the human confirmation
- 11. Summarising a distribution: centre, dispersion and skew
- 12. Index numbers, time series and seasonality
- 13. From sample to population: inclusion probabilities, design weights, the unbiased total
- 14. Precision: confidence intervals, significance, and what a narrower interval costs
- 15. Exploratory and geospatial analysis: area units, denominators, a traceable trail
- 16. Personal data in a statistical service: roles, purposes, minimisation and retention
- 17. Confidentiality and disclosure control, with the cybersecurity safeguards around them
- 18. Open data and reuse: formats, licences, high-value datasets and interfaces
- 19. Statistical metadata and quality reports: structured fields, harvesting, and telling a user what a figure is worth
- 20. Charts and dashboards that do not mislead: axes, scales and accessibility
- 21. Maps: choropleths, rates, and what shading an area actually says
- 22. Reports, briefings, methodological notes, and support to expert groups
- 23. Scripts, queries and automation you can hand over: version control, testing and reproducibility
- 24. Machine learning for routine work, with the quality and interpretability checks that make it reportable
- 25. New technologies in the production chain, and generative AI used with a control around it
- 26. The field-related multiple-choice test: 30 questions, 40 minutes, and the only mark that ranks
How you receive it
Two files, both on your account the moment the payment clears: a 657-page PDF to read on any screen, print or annotate, and an EPUB for e-readers and phone reading apps. You download them from inside your PassEPSO account, straight away — nothing is posted and nothing is waited for.
If you do not have a PassEPSO account yet, you do not need to make one first: type your email at checkout and the account is created for you, with the book already in it.
Already bought it? Open your copy — it is on the account you paid with.
Where the material comes from
“Everything in this book is drawn from official, publicly available primary sources that were in force at the time of writing. Each chapter lists the sources it used, with the instrument, the article or paragraph number, and the version.”
Texts of EU law are cited from EUR-Lex with their Official Journal references, so every instrument can be found in the version used: the Regulation on European statistics, the data protection and open data instruments, the NUTS Regulation, the European Statistics Code of Practice and the Eurostat methodological manuals. Where a framework is private and protected by copyright, the book reproduces none of it: it states what the standard requires in its own words and tells you exactly where to read the original. Every source it uses is free to consult.
No statistic about the competition, about candidates or about pass rates appears anywhere in the book. Worked cases with amounts, tables and sample sizes are invented to teach a point and are labelled as such where they appear. PassEPSO is independent and is not affiliated with, endorsed by or connected to EPSO, EU Careers, Eurostat or any institution of the European Union. Your Notice of Competition governs — not this book.
Before you buy, read these
Both are free and neither needs an account: the competition page sets out the tests, the languages, the scoring and the eligibility of EPSO/AST/158/26, and the checklist article walks through the supporting documents EPSO asks for.
EPSO/AST/158/26 — AST 3 Data Management · The supporting documents checklist