data
Question the numbers
The rule
A big headline number is itself a claim, like a shopkeeper's price tag. Check how it was made before you trust it.
Where it flips
But do not swing the other way and call all numbers fake. That is just as lazy as trusting them fully. The goal is to understand how a number is built, not to throw away every number.
Huff showed that a number can be true and still fool you. It can hide the truth by how it is counted, what it is compared to, or what it leaves out. A big economic number is polished many times before you read it. So ask how it was built and what it skips. Do not trust it just because it looks official.
A worked example
A reader sees an Indian inflation number called low. She looks closer. It uses a cheap month to compare against, and its basket has little of the food she buys. The number was true, and it still misled her. [illustrative]
How to spot it
- ·a low month chosen to compare against, to flatter the result
- ·an adjustment quietly doing a lot of the work
- ·a definition or basket that leaves out what matters to you
- ·a number that will be changed later, quoted as if final
Darrell Huff · How to Lie with Statistics