EN
PL
“There are three kinds of lies: lies, damned lies and statistics.”
Attributed to Benjamin Disraeli, popularized by Mark Twain.
As the saying goes, the biggest lie of all is statistics.
But is that really true? In today’s world, where statistical tests, model outputs, and even AI responses are treated as sacred oracles, can we still say the same? Should statistical results take precedence over common sense? The answer is never clear-cut. As the classic saying goes: “It depends.”
I hope this post will encourage readers to reflect. To consider whether we should blindly trust every analysis and every study. I’ll touch on a few controversial, sometimes media-fueled topics.
“Poles commit the most crimes in the UK”: a simple data manipulation.
In 2016, Onet published articles suggesting that Poles were responsible for the highest number of crimes in the UK. These reports highlighted the raw data, but failed to mention that Poles make up one of the largest ethnic minority groups in the UK. Nor did they consider the severity of the crimes—every offense was counted equally. Once adjusted for population size, it turned out that the proportion of crimes committed by Poles was actually negligible.
The case of Zbigniew Martyka, PhD: scientific consensus vs statistical research.
The COVID-19 pandemic brought many changes to our lives—and a flood of statistical data and research findings. Mainstream media emphasized the importance of masks, vaccinations, isolation, etc. Meanwhile, others cited studies claiming the opposite. These voices were often ignored or censored in public discourse. And the same data were frequently interpreted in completely different ways.
One embodiment of these controversial opinions was Zbigniew Martyka, PhD. He questioned, among other things, the effectiveness of mask-wearing, mass testing, the real threat of COVID-19 to children and the necessity of vaccination. He cited randomized controlled trials (RCTs), which he considered the “gold standard” of scientific evidence.
In late 2022, a medical court suspended Martyka’s license to practice for one year. The verdict was not final. Prosecutors cited observational studies and modeling approaches, which are generally considered lower-quality evidence. The case dragged on until March 2025, stirring further controversy—including a verdict allegedly issued without evidence. Ultimately, the court overturned the suspension and issued only a formal reprimand.
I won’t go into the details here—I encourage you to review the case and sources yourself. I’m citing it to highlight how statistical research can be interpreted in vastly different ways and sometimes used instrumentally to support one’s position. Was Martyka right? Or was it the “scientific consensus”? I leave that for readers to ponder.
Traditional herbal medicines vs synthetic drugs
Another area often accused of statistical manipulation is the debate between natural medicine and synthetic drugs. Recently, there has been increasing criticism and restriction of traditional remedies, often justified by claims of insufficient research or lack of statistical proof of efficacy. Here are a few examples:
- Chinese herbs for COVID-19
During the pandemic, China approved three traditional Chinese medicine formulations for COVID-19 treatment. The Lancet published a commentary titled “Use of herbal drugs to treat COVID-19 should be with caution,” emphasizing the lack of rigorous evidence and potential risks.
In response, a piece in the Journal of Integrative Medicine argued that such caution reflected bias and cited observational evidence in favor of the herbs. It also pointed out that standard treatments were also introduced with limited data early in the pandemic. - Artemisia annua (anti-malaria herb) vs synthetic drugs
Used in China for centuries to treat fever, this plant’s active compound—artemisinin—became the basis for modern anti-malaria drugs. WHO strongly advises against using traditional Artemisia annua extracts or tea due to concerns over dosage standardization and efficacy. They recommend only approved, standardized drugs like ACT.
Yet some studies show that the raw plant may help fight artemisinin-resistant malaria strains and slow down the development of drug resistance. It’s worth noting that malaria-affected regions are often poor and may not have access to expensive pharmaceuticals. - Ashwagandha
Used in Ayurvedic medicine for over 3000 years to treat stress, insomnia and aging. In 2023, Denmark completely banned ashwagandha, citing potential hormonal effects and risk of miscarriage. However, clinical evidence supporting these concerns remains limited and inconclusive. Solid scientific proof is lacking.
While many natural remedies have long history of use, only treatments that undergo rigorous clinical testing provide real assurance of efficacy and safety.
“One rabbi says yes, another says no”: interpreting statistical results
The impact of social media on youth mental health is one of those hot-button topics that periodically inflames public discourse. Some researchers warn that smartphones are ruining teens’ lives. Others argue the data don’t really support that. So… what’s the truth?
In 2018, American psychologist Jean Twenge published a widely discussed study using large U.S. datasets (Monitoring the Future, Youth Risk Behavior Surveillance System). The conclusions were alarming: the more time teens spent on social media, the worse they felt. Twenge even suggested social media could be a primary cause of rising depression and anxiety—especially among girls.
But in 2019, Amy Orben and Andrew Przybylski from Oxford reanalyzed the same data. They used a more advanced method (specification curve analysis) to show how different methodological choices can change outcomes. Their conclusion? Yes, there is a relationship—but it’s tiny. Statistically significant, but practically meaningless—explaining less than 0.4% of the variance in well-being. For comparison, sleep and whether a child ate breakfast had a bigger impact.
Here’s the kicker: both interpretations are statistically valid. The difference lies in how the analysis was designed, what questions were asked and what assumptions were made. Twenge focused on broad societal trends. Orben and Przybylski took a strictly mathematical approach, showing that what seems dramatic may not be under closer scrutiny.
In other words: one rabbi says yes, another says no. And statistics? Statistics can support both stories—if you choose the right tools.
That’s why it’s so important to read studies critically—and always ask: What was really measured here?
What’s the takeaway?
These examples aim to show that while statistics are indispensable across many fields—allowing us to observe and interpret complex phenomena and make informed decisions—they are not a purely mathematical discipline. Many tools are sensitive to subjective choices of methods or variables. And results require expert interpretation. Every researcher views data through the lens of their own knowledge and beliefs. Interpretations are always shaped, to some extent, by that perspective.
So what can we do to guard against manipulation? How can we form our own opinions and learn to critically assess studies?
There’s only one answer: we need basic statistical literacy. Understanding how statistics work, their strengths and weaknesses.
To that end, I recommend various training opportunities—such as those provided by StatSoft Polska.
P.S.
Reading this post, you might form opinions about the author’s personal beliefs.
Please note: the author has purposefully highlighted less popular perspectives, which may not reflect his own views.
P.S. 2
In preparing this article, ChatGPT was used extensively to gather data and references.
The majority of the text was written by the author himself. One paragraph was written by ChatGPT. Can you guess which one?
Author: Mateusz Żołneczko, Senior Data Analysis Consultant at StatSoft