If you’re determined to find data that confirms a crisis, it’s easy enough: just keep adjusting until the numbers fit your narrative.

In the West, there is rightly much focus on ending violence against women and girls. According to the UN, an estimated 736 million women globally have experienced physical and/or sexual violence at least once in their lives. To effectively combat this issue, reliable data is essential.

In November 2022, European media sounded the alarm over an epidemic of gender-based violence in universities across the continent, citing findings from the EU-funded UniSAFE survey. The startling statistic: 62% of respondents reported experiencing some form of gender-based violence. The implication was stark — European campuses appeared more perilous than a midnight walk through Kabul.

But how solid was this evidence? At first glance, UniSAFE looked impressive. It boasted that it was “the largest survey of its kind”, with 42,055 respondents from 46 universities across Europe. But the headlines conveniently skipped the embarrassing detail that only 3.9% of those contacted actually replied to the questionnaire, with this figure dropping to just 2.7% among students. Normally, responsible researchers would quietly discard such dismal data.

However, those at UniSAFE pressed ahead, using vague measurements that counted nearly anything short of an explicit “no” as a case of gender-based violence. An awkward joke, a lingering glance: almost anything could qualify. Crucially, this survey lacked documented psychometric validation and had no clear timeframe, making its results unreliable.

Lisa, a Swedish psychologist and senior member of UniSAFE’s statistical team, quickly spotted the flaws. Alarmed by how casually the team treated its shaky numbers, she voiced concerns internally. She wasn’t alone: senior researchers, including Karin Dahlman-Wright from the Karolinska Institute and UniSAFE’s own quantitative analyst Christina Björklund, shared her discomfort.

When Lisa pressed her bosses at Sweden’s National Secretariat for Gender Research about why they highlighted the sensational 62% figure rather than more credible data (like the validated COPSOQ measure, which found just 3.5% reported harassment), her questions were brushed aside.

“I’ve researched women’s vulnerability my whole career,” Lisa told me. “But it has to be done correctly. When I saw how the numbers were presented, I knew something was fundamentally wrong.”

Ironically, Fredrik Bondestam, the senior Swedish researcher overseeing UniSAFE’s methodology, had previously warned about precisely this issue. In a 2018 government-sponsored report, Bondestam cautioned that harassment prevalence figures varied wildly from 2% to 93% depending entirely on methodological choices. Yet by 2022, he enthusiastically promoted UniSAFE’s inflated numbers without caveat, conveniently securing around €2 million in additional EU funding for a follow-up project aptly named “GenderSAFE”.

But none of these inconvenient truths made it into the polished PowerPoint slides for Brussels. At a high-profile EU conference in Prague in November 2022, Bondestam and colleague Sofia Strid unveiled their findings. Bondestam declared: “We have the data on prevalence in place […] there are no more excuses for keeping silent or passive.” Strid insisted that gender-based violence had become “a near-normalised part” of academic culture, again invoking the “two in three” statistic.

Bondestam ratcheted up the rhetoric further: “Combating gender-based violence as a mandatory EU requirement needs to be put in place now. This is not optional.” The actual UniSAFE report projected behind them, but which the audience couldn’t properly read, quietly admitted the real response rate was 3.9%.

Institutions such as the European Commission swiftly incorporated this statistic into its Zero-Tolerance Code of Conduct on Gender-Based Violence for universities, and UNESCO prominently cited it in its global Safety of Scientific Researchers report. It did not acknowledge the glaring methodological problems, despite UNESCO’s own guidelines explicitly cautioning against exactly these pitfalls.

To illustrate exactly how absurd this statistical charade is, imagine inviting 25 people to your birthday party and only one person shows up. You then confidently announce that “100% of attendees loved the cake” and immediately decide that every cake at every family party henceforth must be this cake. This, essentially, is how UniSAFE’s 3.9% response rate was transformed into official EU policy.

Lisa paid the price for refusing to play along, becoming increasingly marginalised, excluded from meetings, and eventually pressured out altogether. “We should let data speak — not our political wishes,” she said, looking back. “But it was never about what was true — it was about what was convenient to say.”

Yet the EU’s magic figure of 62% continues to reverberate, shaping policy and public perception across Europe and beyond. What began as important but very questionable research now governs millions of euros in funding and dictates the terms of acceptable debate. Apparently, when you’re manufacturing a crisis, the data doesn’t matter as long as the headlines look good.