A number in a headline carries authority. It looks measured, precise and neutral. But every statistic is a choice about what to count, what to compare it with and how to present it. The same data can produce opposite headlines.
Relative versus absolute
"A new medicine cuts the risk of a condition by 50%." Impressive, until you learn the risk fell from 2 in 10,000 to 1 in 10,000. Both statements are true. The relative change is dramatic; the absolute change is tiny. Honest reporting gives both.
The missing denominator
"500 people suffered side effects." Out of how many? Five hundred out of a thousand is very different from five hundred out of ten million. A number without its total is half a fact.
Averages that hide the spread
Average income can rise while most people become poorer, if a few incomes rise enormously. The median, the middle value, often tells a more honest story than the mean.
Cherry-picked time frames
Almost any trend can be made to look like a rise or a fall by choosing where the chart begins. Ask why a graph starts where it does, and what the longer series looks like.
Correlation presented as cause
Two things rising together does not mean one causes the other. Both may be driven by a third factor, or the link may be coincidence. "Linked to" and "causes" are very different claims.
Five questions for any statistic
- Compared to what?
- Out of how many?
- Over what period?
- Measured by whom, and how?
- Is this a cause, or only a connection?
How Honest Lens treats numbers
In the Distortion Tracker, an altered statistic, for example a 1.8% decline reported as "30% wiped out", is marked in orange with a one-sentence explanation of how it differs from the source. In the Claim to Evidence engine, a claim built on a real number used the wrong way is labelled Misleading rather than simply true or false, because the number itself may be correct while the conclusion is not.