The headline that was true

'Cycling deaths rise by fifty per cent' is a true statement about a town where the number went from two to three.

Nothing in the headline is false. The percentage is correct, the deaths are real, and the newspaper could defend every word in front of a regulator. The reader, however, takes away something the figures do not support: that cycling has become substantially more dangerous. With numbers this small, a single event moves the percentage enormously, and next year's figure will probably move it back.

The honest version is harder to write. 'Cycling deaths rise from two to three' is accurate and sounds trivial, which is a problem if the underlying trend is real and the paper wants anyone to read about it. Journalists face this every week, and the usual compromise is to give the percentage in the headline and the raw figures in the third paragraph, where fewer people reach them.

Some publications now have a rule: if the base number is under about twenty, the headline uses counts rather than percentages. The rule costs them attention and it prevents a specific, repeatable mistake. It also has the useful side effect of making the newsroom say out loud how many cases they are talking about, which sometimes ends the story before it is written.

Readers can apply the same rule from the other side. The question to ask a striking percentage is not whether it is accurate but what it is a percentage of. If the answer is not in the article, the article has decided not to tell you.