Survivorship bias

What it is

Survivorship bias happens when we draw conclusions from the cases that made it through some filter, while the cases that did not are invisible. Because the failures are missing from view, the survivors can make a strategy look safer, more effective or more predictable than it really is.

An everyday example

A newspaper feature profiles five people who quit stable jobs to open restaurants and are now thriving. Each one says the same thing: “Trust your gut and go all in.” Readers come away feeling that boldness is the secret ingredient.

What the feature can’t show are the people who also trusted their gut, also went all in, and closed within two years. They had the same courage; they just aren’t around to be interviewed. Without them, it is impossible to tell whether boldness helped at all.

Why it happens

Failures tend to vanish from the record. Closed businesses stop advertising, merged or closed investment funds drop out of databases, and people rarely write books about plans that did not work. What remains is a sample that has already been filtered, and our minds naturally treat what we can see as the whole picture.

How to counter it

• Ask “where are the others?” For any success story, picture everyone who started on the same path, not just those still standing.
• Look for the base rate. How many people or projects tried this, and what share of them succeeded?
• Prefer complete data. In investing especially, check whether performance figures include funds or strategies that were closed along the way.
• Treat advice from winners as a starting point, not proof. What they did may be common to losers as well.

What the research shows

In finance, Brown, Goetzmann, Ibbotson and Ross (1992) showed that studying only funds that survive can make performance appear predictable even when it is not. Funds that take more risk are more likely to either do very well or disappear, so a survivors-only sample can show patterns that are partly an artifact of who was left.

A well-known historical case comes from World War II. The statistician Abraham Wald developed methods for estimating how vulnerable different parts of an aircraft were, using damage seen on planes that returned while accounting for the planes that did not (Mangel & Samaniego, 1984). The key insight: areas that rarely showed damage on returning planes might be exactly where a hit was most likely to bring a plane down. Popular retellings of this story are often simplified, but the statistical idea is genuine.