Survivorship Bias
Also known as: Survivor Bias, Survivorship Fallacy
Formulated by Abraham Wald (1943)
Definition
A logical error that occurs when conclusions are drawn from a data set consisting only of the 'survivors' of some selection process, while overlooking the cases that did not survive and are therefore invisible to the analysis. Because the failures leave no trace, the surviving sample looks unrepresentatively successful or resilient, leading to systematically distorted conclusions about what actually causes success or survival. The bias appears everywhere from business ('successful startups all did X') to finance (mutual fund performance data that excludes funds that closed) to everyday advice ('follow your passion,' drawn only from famous people who succeeded that way).
Illustration of a hypothetical damage pattern on returning WWII bombers, loosely based on the data analyzed by Abraham Wald. Reinforcing the areas covered in red dots, where survivors were hit, would have been a mistake, since planes hit in those areas mostly made it home. New version by McGeddon, vector by Martin Grandjean, CC BY-SA 4.0, via Wikimedia Commons.
History
During the Second World War, the U.S. military's Statistical Research Group (SRG) studied Allied bombers returning from missions over Europe to figure out where to add armor. The obvious idea was to reinforce the spots taking the most damage: engines, wings, and the tail were riddled with bullet holes, while the cockpit and central fuselage were comparatively untouched.
Mathematician Abraham Wald, a member of the SRG, reversed the logic. The damage data came only from planes that had survived and made it back. A bomber hit in the engine or fuel system was far less likely to return at all: those planes went down and were absent from the sample. The undamaged areas on the surviving aircraft were not safe; they were fatal. Wald recommended armoring the places where the returning planes showed no damage, since a hit there was the kind a plane rarely survived to report.
Wald's 1943 memorandum, A Method of Estimating Plane Vulnerability Based on Damage of Survivors, was not published for decades but became one of the founding case studies of applied statistics. The term "survivorship bias" itself came into common use much later, popularized in business and finance writing from the 1970s onward, but the underlying insight, that a sample missing its failures cannot tell you about failure, remains Wald's.