fallacy

Texas sharpshooter fallacy

Also called: Clustering illusion

The Texas sharpshooter fallacy finds a pattern in data after the fact and treats it as meaningful, like firing at a barn and then painting a target around the tightest cluster of holes.

How it works

Random data always contains clusters. Choosing the cluster after seeing the data, and ignoring everything outside it, makes chance look like design.

Examples

"Five of the people who got sick lived near the mast, so the mast caused it."

The cluster was picked after the fact; similar clusters appear by chance everywhere.

Listing the predictions an analyst got right and ignoring the many that missed.

The target is drawn around the hits after they happened.

How to spot it

The hypothesis was formed after looking at the data.
Misses and non-clusters are not counted.
No test on fresh data.

How to respond

Ask whether the pattern was predicted in advance.
Ask how often such clusters appear by chance.
Test the claim on data that was not used to find it.