Survivorship bias
Studying only the winners and missing the silent evidence of everyone who failed.
What it means
Survivorship bias is the error of drawing conclusions from a sample that has passed through a selection filter while ignoring the cases that dropped out, producing distorted and usually over-optimistic inferences. The failures are invisible precisely because they did not survive to be counted, so the surviving sample looks systematically different from the original population. The mechanism is selection, not motivated reasoning: any process that quietly removes the worst outcomes will leave a residue that flatters whatever trait the survivors happen to share. It is especially pernicious in business and investing, where the visible successes — the unicorn startups, the funds still open — crowd out the far larger graveyard of comparable ventures that vanished. The correction is to reconstruct the full reference class, including the dropouts, before inferring what 'works.' It matters because advice distilled from survivors ('they all dropped out of college') confuses a trait of winners with a cause of winning.
Examples
WWII engineers wanted to armor the bullet holes on returning planes — until Abraham Wald noted the planes hit elsewhere never made it back, so those were the spots to reinforce.
Business books study the founders who made it and conclude that dropping out and betting everything works. The thousands who did exactly that and vanished never got to write a chapter.
People insist old buildings were built better because every surviving one is handsome. The shoddy and ugly were pulled down decades ago and never made it into the sample.
First described in Popularized via Abraham Wald's WWII work.