No. 16 Statistical Fallacies
Cherry Picking
In Turkish: Sümen Altı Safsatası
You showed me three cherries and hid the tree.
We won the last three matches! The title’s ours this year!
Count the ten before those too, Temel.
He shows you three cherries and hides the tree. Every year the same.
What is it?
Putting forward the data that suit you and burying the data that don’t. Everything shown is true; the lie is in what isn’t shown.
The pattern
- I have ten data points; three support my claim.
- I show those three and hide the seven.
- “See? The data support me.”
Why it fails
The cleanest way to mislead without lying: you invent nothing, you just select. The listener can’t know about the data they never saw; three examples feel like the whole. Ads, fan arguments and reports run on this. Three well-chosen truths mislead better than one lie.
How to spot it
- All the examples point the same way; there isn’t a single counterexample.
- There are numbers but no total: “three matches”, never “three out of thirteen”.
- “And the rest?” gets an evasive answer.
What to say
“These are selected; where’s the rest?” Ask for the total: “How many out of how many?” If no total comes, the selection means nothing. Then put a counterexample of your own on the table; two are enough to show the story was curated.
When it’s not a fallacy
Every account selects; telling everything is impossible. If the selection represents the whole and counterexamples aren’t hidden, there’s no problem. Producing a single counterexample to refute a claim isn’t cherry picking either; against “they’re all like that”, one exception is enough. The fallacy is deliberately leaving out the data that would change the picture.
Examples
- Ads“Ninety percent of users are satisfied.” The survey went to the people who didn’t return the product.
- WorkThe presentation shows three cities where sales rose; the seventeen where they fell are “in the appendix”.
- Health“This diet worked for three people.” The twelve who quit aren’t counted.
- Social mediaThree seconds are clipped from a speech; the remaining twenty minutes would have reversed the meaning.