A field guide to the ways reasoning breaks
Evidence and sampling
Also called Clustering illusion, post hoc pattern fitting
Drawing the target around the hits, so a pattern is defined after the data is seen.
The prediction hit on three of forty forecasts, so it is accurate.
Ask whether the pattern was specified before the data was collected.
Named for a tall tale about a marksman who fires into the side of a barn, then paints a bullseye around the tightest cluster of holes. The image goes back to John Venn's 1866 Logic of Chance, and the Texas version surfaced in statistical writing in 1977 before Atul Gawande popularized it in 1999. It is the statistical face of apophenia — pattern-seeking after the fact.
A newsletter subscriber counts back through an astrologer's fifty weekly forecasts and reports only the three that 'came true'.
An analyst tracks which of his hundred stock picks beat the market and publishes a report titled 'My Seven Proven Winners'.
A city's crime map shows a cluster of incidents in one district and the evening news declares a 'crime corridor' — on data from a single quarter.
A sports bettor posts screenshots of his last three wins and calls himself a sure-thing tipster; the full ledger is private.