Texas sharpshooter
Also called: Painting the target afterwards
Finding a pattern in the data after the fact, and treating it as if it had been predicted.
How to spot it
Was the claim chosen before looking at the data, or after?
Example
“Look, every disaster happened within a year of a comet.” (After searching a century of disasters for comets.)
How to answer it
Ask what was predicted in advance, and test it on new data.
Related
- Cherry-picking: Showing only the evidence that supports your case, and leaving out the rest.
- Post hoc: Assuming that because one thing followed another, the first caused the second.