Accurate facts can be just as misleading as outright falsehoods when selectively presented or misinterpreted, says London Business School finance professor Alex Edmans.
In a speech titled “May Contain Lies: Navigating Deceptive Data” at the 2026 World Knowledge Forum in Seoul, South Korea, Prof Edmans argued that even educated people can be misled by data, particularly when it confirms something they already want to believe.
“It’s not that we’re bad people; it’s that we’re people,” he told the audience.
He said that his interest in the subject began after the birth of his son, when he and his wife sought advice on infant feeding.
A respected parenting course in the UK, citing World Health Organization recommendations, stressed the benefits of breastfeeding, including claims that it could improve intelligence, he said.
Prof Edmans said he checked the research himself, and the results appeared convincing, with reputable media outlets reporting studies linking breastfeeding with higher IQ.
But the original studies told a different story, he said.
One said its sample was too small to support a firm conclusion, while another involving thousands of children found no significant effect, he said.
A widely cited study showed breastfed children scoring about five IQ points higher than formula-fed children, he said.
Once researchers adjusted for factors such as maternal intelligence, education and smoking, however, the gap fell to about half a point and was no longer statistically significant, he noted.
The experience became the basis of his book May Contain Lies and what he calls the “ladder of misinference”.
From facts to proof
The first problem, Prof Edmans said, is that a claim is not necessarily a fact.
He cited the opioid crisis, noting that a widely repeated assertion that addiction was rare among patients given narcotics could be traced to a short letter in the New England Journal of Medicine about hospital patients receiving limited doses. Its finding was later applied far more broadly, he said.
Even accurate facts may not amount to meaningful data, he said.
Prof Edmans said individual success stories can mislead when they are selected to support an argument.
Claims that organisations thrive because they have a strong sense of purpose, for example, often highlight successful companies while ignoring purpose-driven ventures that failed, he said.
A proper comparison must include both successes and failures, he said, adding the next distinction is between data and evidence.
Prof Edmans used a courtroom analogy.
“If Tom, Dick and Harry all had an opportunity to kill Sarah, those facts alone do not identify the murderer. Because the information is consistent with several suspects, it is data rather than evidence pointing towards one of them,” he said.
The same principle applies to research, he said.
If breastfed children score higher on IQ tests, that does not mean breastfeeding caused the difference, he said, adding that children from more stable and better-resourced families may be both more likely to be breastfed and more likely to score highly.
Unless competing explanations can be ruled out, correlation should not be treated as evidence of causation, he said.
Prof Edmans said people can also behave like investigators who already have a preferred suspect, interpreting new information as confirmation while overlooking evidence pointing elsewhere.
To counter this, he suggested imagining that a study had produced the opposite result.
Readers should ask what explanations they would use to challenge it, then apply those same objections to the result they prefer, he said.
The final step is recognising that evidence is not proof, he said, as a finding may be valid in one setting but fail when applied elsewhere.
Prof Edmans cited research showing that “grit” predicted success during training at West Point, the US military academy.
But those admitted had already passed demanding physical requirements, meaning fitness levels among them varied relatively little, he said.
He also referred to a satirical study finding that parachutes did not improve survival when people jumped from aircraft.
The result was technically correct, he said, as the aircraft never left the ground.
“Even if a fact is 100% accurate, it might still be misleading,” he said.
Question the evidence
For Prof Edmans, this is why conventional fact-checking is not enough.
Readers should ask not only whether a claim is true, but where it came from, who was studied, what comparison was made and whether another explanation could account for the same result, he said.
That means tracing claims back to original studies rather than relying on headlines or summaries, and checking whether headline effects remain once other relevant factors are considered, he added.
Most importantly, people should examine evidence they agree with as critically as evidence they dislike, he noted.
“We can’t rely on other people to check the facts for us,” Prof Edmans said.
His broader message can be reduced to a simple chain of caution, he said.
“A claim is not necessarily a fact. Individual facts are not necessarily meaningful data. Data becomes evidence only when it helps distinguish one explanation from plausible alternatives. And evidence is still not proof that a conclusion will hold in every setting,” he said.
The challenge, he argued, is to ask better questions about what any newly found information actually shows.