When statistics become a political issue: How unfounded accusations of manipulation damage democracy

August 20, 2025 / Katharina Schüller

Statistics form the basis for important decisions in politics and economics. When allegations of statistical manipulation are made, more than just rows of numbers begin to unravel. Two recent cases show how heedlessly accusations against the Federal Statistical Office are handled in Germany - and why objectivity is now more important than ever.

If you don't like what the numbers show, you can change the numbers—or fire the statistician. This is what happened in the US. In early August, US President Trump dismissed Erika McEntarfer, head of the Bureau of Labor Statistics (BLS), because he didn't like the latest labor market data. Trump even spoke of “falsification.” His reasoning was that the labor market figures had been revised several times.

The opposition accused the President of autocratic behavior. Many see threats to the integrity of the statistics agency. At stake are the neutrality, methodological competence, and political independence of official statistics.

Statistical agencies must remain independent

Can statistical indicators that carry political weight be apolitical? Not only can they, but they have to be! However, especially after major revisions, be they due to changes in methodological standards or a result of revised data, the tension between political relevance and factual neutrality is regularly put to the test. This poses a major challenge in terms of both transparent communication of statistics and of users‘ objective criticism.

 The US administration's approach to the BLS is an example of how unfounded insinuations and accusations of manipulation destroy essential trust. The resulting damage is long-term, undermining the evidence base for important decisions in politics, economics, and public discourse.

Unfortunately, even Germany is no longer free of baseless cries of manipulation aimed at deliberately undermining the credibility of official statistics. For the sake of our democracy, two recent cases must not go unchallenged.

Case 1: The alleged GDP calculation error

In an opinion piece on Focus Online, media entrepreneur Gabor Steingart accused the Federal Statistical Office of having “miscalculated” the gross domestic product (GDP), seeing the proof in subsequent corrections to it for 2023 and 2024. His conclusion: “The statisticians are deceiving us alone, not themselves.”

Unstatistik author Katharina Schüller responded to Steingart's article in an open letter. In it, she corrects numerous errors and misinterpretations that Mr. Steingart made in his opinion piece. For one, GDP (like US labor market figures) is regularly revised—this is nothing unusual and is in line with the international guidelines of the System of National Accounts (SNA) and the legally binding rules of the European System of National Accounts (ESA).

What makes this particularly curious is that the Federal Statistical Office regularly pointed out expected major revisions in its press releases between November 2021 and February 2024: “The current GDP results also continue to be subject to greater uncertainty than usual. This applies in particular to the price-, seasonally- and calendar-adjusted quarterly figures.” It is therefore hard to understand how anyone could have been surprised by the latest revisions.

Case 2: The vanished poor

The Federal Statistical Office has changed its method to calculate the at-risk-of-poverty rate. Instead of the microcensus core program, it now uses the MZ-SILC (Statistics of Income and Living Conditions) to improve comparability with other European countries. The Office has transparently conveyed this information to its users. All previous calculations have been archived.

MZ-SILC is specifically tailored to EU comparisons and better captures income that is not derived from gainful employment, such as child benefits, student loans, care allowances, or housing benefits. In addition, the exact income in euros is requested and not just the income according to rough categories, so that overall, more reliable data are now available than previously. However, the smaller sample size means that analyses of poverty risk factors in MZ-SILC based on socioeconomic characteristics are only possible with a higher degree of uncertainty.

Thirty poverty researchers have now signed a letter of protest demanding that the at-risk-of-poverty rate continue to be calculated on the basis of the core program. They argue that the Federal Statistical Office's approach is “utterly unscientific” and “borders on administrative arbitrariness when a federal office withholds results of general scientific and public interest, thereby curtailing the entire expert discussion and public reception. Or is it intended to steer the discussion in a certain direction?” They find it fully impossible that a million people at risk of poverty have vanished overnight as a result of the changeover.

These million have not. They most likely never existed, but were the result of measurement errors due to the less precise and systematically biased recording in the core program.

As an aside, the interpretation of the at-risk-of-poverty rate is a matter of considerable debate. We have pointed out several times in the context of unstatistics that this indicator has little to do with “poverty” in the general understanding of the term. To provide a more comprehensive picture of the phenomenon of poverty, the Federal Statistical Office also publishes the EU-wide comparable indicator AROPE, which takes into account what people can actually afford and the extent to which they can participate in social life.

Much ado about nothing

Only in one line of the protest letter, which most media outlets overlooked, does it become clear what the signatories are actually criticizing: The Federal Statistical Office no longer calculates a certain variant of the at-risk-of-poverty rate. The data from the core program still exist and are still used to calculate and publish comprehensive, regionally and sociodemographically detailed at-risk-of-poverty rates.

The only difference is that the poverty risk is now based on the median income of the respective federal state instead of the federal median. This means that people in Thuringia are no longer compared with a nationwide average that also includes high incomes from the metropolitan region of Munich. This change can be interpreted as abandoning a meaningless statistical average in favor of a more differentiated approach, thereby better reflecting the different realities of life in Germany. Our very first “Unstatistik” in 2012 pointed out the problem of using the federal median to calculate poverty rates.

Facts instead of fake news: Scientific freedom is not restricted

The Federal Statistical Office has good reason to publish only an “official” at-risk-of-poverty rate. All other data remain available for research purposes, albeit with a delay of two to three years. Any scientist can apply for access via the Federal Statistical Office's research data center and carry out their own analyses.

Even from the published statistics, an experienced poverty researcher could easily calculate the at-risk-of-poverty rate based on the federal median. However, the signatories of the protest letter appear unable – or unwilling – to do so.

The accusation of “unscientificity” or a “restriction of scientific freedom” is therefore simply false.

Trust in statistics is the foundation of a functioning democracy

We are deeply concerned about how quickly, heedlessly, – and in some cases against better judgment – accusations of “manipulation” and “political influence” now abound in Germany when a few individuals or institutions dislike a number published by the Federal Statistical Office. Even more worrying is that some media outlets gratefully and often uncritically adopt these alleged scandals without questioning the political or economic intentions of their originators. In doing so, they undermine trust in an important democratic institution.

The Federal Statistics Act (BstatG), § 1, makes it unmistakably clear: “The principles of neutrality, objectivity, and professional independence apply to them [official statistics].” As members of the “Commission on the Future of Statistics” (Bauer, Radermacher, Schüller) and long-standing chair of the Statistical Advisory Board (Bauer), we have always found the Federal Statistical Office to be professionally independent of any political directives. However, this no longer seems to be sufficient. In order to secure its independence for the future, the Federal Statistical Office should finally be given a legal status similar to that of the Federal Audit Office.

In “Unstatistik des Monats,” Berlin-based psychologist Gerd Gigerenzer, Dortmund-based statistician Walter Krämer, STAT-UP founder Katharina Schüller, and RWI Vice President Thomas K. Bauer take a critical look at recently published statistics and their interpretations every month. All “Unstatistics” can be found online at www.unstatistik.de and on the Twitter account @unstatistik.