Fallacies: S (Part 3 of 4)

Core idea

Part 3 of the letter S leans heavily on numbers and on the things we think numbers prove. A famous statistic about swallowing spiders in your sleep turns out to be a deliberate hoax. A famous landslide poll predicted the wrong president. An old country adage reads winter severity off a squirrel's larder. The thread here is statistical literacy — knowing that a big sample, a confident percentage, or a tidy correlation can all be wrong in specific, identifiable ways.

Why it matters

We trust numbers because they feel objective. But a sample can be large and still biased; a percentage can be reported and still misunderstood; an adage can sound ancient and still be backwards. The squirrel myth and the Literary Digest poll fail for the same reason — both read a signal that was never really there.

Key takeaways

Mental model

Copy sourceZoom

Practical application

Before trusting a statistic, ask two questions. First, who is in the sample, and who got left out? — the Literary Digest had millions of responses but they all came from people willing to mail a ballot back. Second, which direction does the causation run? — the squirrel stores nuts because food was abundant; it is not forecasting anything.

Example

A company boasts that a new feature "increased engagement by 300 percent." Apply the arithmetic check: 300 percent more is four times the original, not three. Then apply the sample check: engagement measured among users who already opted into the feature is self-selected — exactly the Literary Digest trap, scaled down to a dashboard.

Related lessons