READGLOBE

Survivorship bias

Reasoning from data

Survivorship bias is focusing on the people or things that made it through a selection process while overlooking those that didn’t — usually because the failures are invisible. It makes success look more achievable, and its causes clearer, than they really are.

Why it happens

Survivors are visible and vocal; failures drop out of the dataset and out of view. Any analysis that studies only what remained is silently conditioning on success, so its conclusions are drawn from a biased sample.

Examples


  • Studying only successful founders for “the habits of success,” ignoring identical habits in those who failed.
  • Abraham Wald’s WWII insight: reinforce planes where returning survivors were NOT hit — the planes hit there never came back.
  • “They built things better in the old days” — the flimsy old buildings are simply gone.

How to counter it


  • Ask: “Where are the failures, and what would they show me?”
  • Seek the full base rate, not just the winners’ stories.
  • Treat any success-only sample as evidence about survival, not about cause.

The deeper point

The most dangerous data is the data that isn’t there. This isn’t a reasoning error you can think your way out of — it’s a sampling error baked into what you’re allowed to see. You have to go looking for the silence.

Frequently asked


What is a classic example of survivorship bias?
Abraham Wald advising the WWII military to armor the parts of returning bombers that had no bullet holes — because planes hit in those spots never returned to be counted.
Why is survivorship bias dangerous in business advice?
It studies only winners, so it mistakes traits common to survivors for causes of success — ignoring that failures often had the same traits. The lesson is unfalsifiable.
How do you avoid survivorship bias?
Deliberately look for the invisible failures, ask for the base rate of attempts, and treat success-only datasets as conditioned on survival rather than proving causation.

Related


Editorial synthesis © ReadGlobe 2026, drawing on Kahneman’s Thinking, Fast and Slow, the Tversky–Kahneman research program, and the primary cognitive-science literature. · Last reviewed 2026-05-29.