Truth, probability and interpreting a single experiment

A single experiment is one contribution to an argument already under way. The red-wine paper joins earlier papers, different populations, different doses, different methods and substantial uncertainty. Its proper effect is modest: informed belief shifts a little.

Popular reporting routinely changes the claim. “Was associated with” becomes “protects”. “In this sample” becomes “for everyone”. The dose vanishes. The limitations appear below an advert for funeral insurance. By breakfast, a narrow result has become permission to pour another glass for medicinal reasons.

This happens because probability sounds weak and certainty sells. A headline saying “Coffee saves your life” has a protagonist, danger and rescue. A headline saying “One estimate moved slightly inside a disputed literature” attracts less attention. Editors choose the first sentence, readers remember it, and the next contradictory paper creates another dramatic story instead of correcting the accumulated picture.

The honest article would locate the new result among the old ones. It would name the population, the size of the observed effect, the plausible alternative explanations and the degree of disagreement. Readers could then decide whether anything in their lives should change. Without that context, successive reports give contradictory health instructions and leave the underlying level of confidence unclear. Sensible personal decisions become harder to make from the coverage.

Behavioural principles

Behavioural ideas at play in this post

Short, plain-English explanations of the principles behind this post, with links to related books and examples in the archive.