Natural frequencies
Also known as: Frequency format effect
Stating risks as counts of people, not percentages, makes the math click.
What it means
A way of presenting probabilistic information as raw frequencies referenced to a common population ('10 out of 1,000') rather than as conditional probabilities or percentages, which sharply improves people's ability to reason about risk. The format works because natural frequencies preserve the base-rate information that probabilities strip out, letting the relevant counts be read off directly instead of requiring Bayes' theorem to be applied in the head. When diagnostic problems are recast this way, the proportion of people — including doctors — who correctly infer the chance of disease given a positive test rises dramatically. The finding reframes 'base-rate neglect' partly as a problem of representation rather than fixed irrationality: the mind handles frequencies, the format it evolved with, far better than abstract single-event probabilities. It matters for medicine, law, and any field that must communicate risk to non-statisticians.
Examples
Telling clinicians '10 of 1,000 women have the disease; 9 of those test positive; 89 of the healthy 990 also test positive' yields far more correct answers than the same facts given as percentages.
A jury told 'in 1,000 similar cases the match would show up in 1 innocent person' grasps the evidence far better than a lawyer's 'the match probability is 0.1 percent'.
A leaflet reading 'of 100 people with this result, about 5 do not have the condition' is understood by more readers than 'the test has 95 percent positive predictive value'.
First described in Gerd Gigerenzer & Ulrich Hoffrage (1995).