The reason a rising rate can mean a shrinking problem

A city reports that the rate of bicycle accidents per cyclist has risen by fifteen per cent. The headline warns of worsening road safety. Left unmentioned: the number of cyclists tripled over the same period, and the absolute number of accidents actually fell. More people are cycling more safely in total, but the rate-based story tells the opposite.

What everyone sees

Readers accept the rate as the natural unit of alarm. A rising rate reads as deterioration, and because rates are normalised — per capita, per mile, per user — they feel more scientific than raw counts. The normalisation is treated as a correction for scale, so the corrected number is trusted more than the uncorrected one, even when the correction obscures the improvement it was supposed to clarify.

What is actually happening

Rates and counts can move in opposite directions whenever the denominator changes faster than the numerator. This is a mathematical identity, not a fallacy, but it becomes misleading when only one metric is reported. Gigerenzer has argued that confusion between rates and counts is one of the most common sources of statistical illiteracy, and that presenting both together — along with the denominator — is the minimum standard for honest reporting. The bicycle example is structurally identical to cases in healthcare, crime and education where expanding the base population inflates the rate while the absolute burden decreases.

Why it stays hidden

The confusion hides because rates and counts answer different questions and are rarely displayed together. The rate answers “how risky is cycling for a cyclist?”, and the count answers “how many accidents is the city dealing with?” Both are legitimate, but a headline can only carry one, and the one that sounds like a crisis gets the headline. The shrinking problem is invisible because no one asked the shrinking question.