James Cochran
Upon learning of the complete legal vindication of my good friend Andreas Georgiou, I allowed myself the luxury of a few moments of elation. His long and painful ordeal was finally over.
Then I began to seethe again over what had been done to him.
What follows is a confluence of fact and reflection. I have tried assiduously not to allow my anger to outrun the evidence and to distinguish between what the record establishes and what I believe that record requires us to confront.
Andreas became president of the Hellenic Statistical Authority, known as ELSTAT, in August 2010, an extraordinary moment in Greek history. By the time he took his post, the credibility of Greece’s government finance statistics had been badly damaged and the country was already in the midst of a sovereign debt crisis.
Among Andreas’ responsibilities was ensuring Greece produced credible government deficit and debt statistics consistent with European statistical standards.
In November 2010, he transmitted revised deficit and debt statistics to Eurostat. He did not first submit those statistics to a vote of ELSTAT’s governing board.
Why?
Because official statistics are not supposed to be decided by vote.
That decision helped consume nearly 15 years of his life.
Beginning in September 2011, Andreas faced a succession of legal prosecutions, suits, and investigations. Allegations that he had artificially inflated Greece’s deficit ultimately produced no conviction, and by 2019, the case was finally dropped. But a breach-of-duty prosecution based on his failure to submit the statistics to the ELSTAT board persisted.
He was acquitted at first instance. Prosecutors appealed. He was subsequently convicted and received a two-year suspended prison sentence.
At the heart of his defense was a principle that should be familiar to every statistician: professional independence.
The European Statistics Code of Practice assigns responsibility for statistical methods, standards, procedures, and the content and timing of statistical releases to the appropriate statistical authority, and specifically its head, to ensure accountability. Andreas maintained that submitting official statistics to a board vote would have surrendered a responsibility the statistical system explicitly placed elsewhere.
In 2023, the European Court of Human Rights unanimously found that his right to a fair hearing had been violated. In August 2026, Greece’s Supreme Court annulled his remaining conviction.
The statistics stood.
The principle stood.
Andreas Georgiou stood.
But nearly 15 years had passed.
That last fact is the one I cannot get past.
We statisticians speak often about ethics. We teach our students not to fabricate data, manipulate analyses, conceal inconvenient results, or misrepresent uncertainty.
All of that matters.
But Andreas’ ordeal exposes a harder form of statistical ethics.
Sometimes doing the ethical thing means refusing to surrender statistical authority to someone who should not possess it.
And sometimes doing so has a price.
Andreas’ ordeal is particularly disturbing because it is far from an isolated incident. Graciela Bevacqua in Argentina faced retaliation after resisting political pressure surrounding inflation statistics. Gordana Radojević in Montenegro was dismissed after resisting pressure surrounding the conduct of a national census; her dismissal was subsequently ruled unlawful. And in August 2025, Bureau of Labor Statistics Commissioner Erika McEntarfer was dismissed immediately following the release of a politically controversial employment report containing substantial downward revisions to previously reported job growth.
Different countries. Different statistics. Different circumstances.
The same uncomfortable question: What happens when the numbers say something powerful people do not want to hear?
A deficit does not become smaller because a government would prefer a smaller estimate. An inflation rate does not fall because acknowledging inflation is politically damaging. A population does not acquire a different composition because the actual composition is politically inconvenient. Nor does an employment estimate become wrong because its revision is politically unwelcome.
Evidence can be challenged. Methods can be scrutinized. Errors must be corrected.
But evidence is not open to negotiation.
This is why I worry that merely celebrating Andreas’ vindication allows us to learn the wrong lesson.
Andreas Georgiou is a hero of official statistics. I do not use that word casually.
But a healthy statistical system should not require heroes.
If exercising professional independence requires extraordinary personal courage, then something is wrong with the institution supposedly guaranteeing that independence. A code of ethics is not enough. A statutory declaration of independence is not enough. Independence must be protected when its exercise becomes inconvenient.
Otherwise, we create a chilling effect.
Imagine being the next director of a statistical agency. You know what professional standards require. You know what the evidence says. And you know what happened to Andreas.
Perhaps nobody needs to telephone you.
Perhaps nobody needs to threaten you.
Perhaps nobody even needs to ask you to change a number.
You already know the potential price of saying no.
The resulting distortion may never appear in a statistical diagnostic. It may instead appear in an analysis not undertaken, a revision postponed, a methodological concern softened, or a release delayed.
The most effective interference may be the interference that never has to occur.
Everyone has already learned the lesson.
That is why professional solidarity matters. Statisticians and statistical organizations around the world supported Andreas throughout his ordeal. European institutions defended the validity of the statistics and the importance of professional independence. Colleagues attended proceedings, issued statements, and refused to allow the case to disappear from view.
That support mattered.
Andreas’ final legal vindication therefore deserves celebration. But celebration is not enough, and self-congratulation can be delusional.
We should not congratulate ourselves for defending professional independence while accepting institutions in which someone can spend 15 years defending himself for exercising it.
Andreas Georgiou is a hero of official statistics. But if we stop at celebrating his heroism, we risk allowing institutions to evade the harder lesson.
Professional independence must mean more than permission to make the right decision. It must include meaningful protection for the statistician who makes it.
Individual integrity is necessary. Institutional design must make integrity structural rather than heroic. A well-designed system should not require heroic behavior merely to produce the behavior the system claims to want. Professional independence must mean more than permission to make the right decision. It must include meaningful protection for the statistician who makes it.
Sometimes the numbers will tell us something we do not want to hear, and sometimes that is precisely their purpose—sometimes we need most to hear what we least want to hear.
The test of our commitment to statistics comes at precisely that moment.
For nearly 15 years, Andreas was required to take that test again and again. He endured it. He defended the integrity of his statistics, his profession, and its principles.
And, finally, he was completely vindicated.
I am angry about this, and I think you should be angry, too.
Two questions now belong to us.
Would we have had the courage to do what Andreas Georgiou did?
And perhaps the more important question: What are we doing to build statistical institutions in which the next person who does will not have to pay the same price? n
Leave a Reply