Survivorship Bias

Survivorship bias is a form of selection bias whereby persons or things that made it through a selection process of some kind are represented in a sample, and those that didn’t make it through the process are overlooked, usually because they lack visibility in the data set. This often leads to false conclusions when those failures are ignored.

People are prone to believe success within a group is due to some special attribute rather than coincidence; correlation = causation. In investing, this tendency for failed companies to be excluded from studies because they no longer exist skews performance results.

Author Nassim Taleb refers to the data left out due to survivorship bias “silent evidence”.

In what may be the earliest reference to survivorship bias, Diagoras of Melos (5th century BC) reportedly had a friend try to convince him of the existence of the gods by pointing out how many paintings there were of people being saved from storms at sea by “praying to the gods”. But Diagoras replied that there were no paintings of those who prayed and were also shipwrecked and drowned at sea.

Abraham Wald, a hero of operational research, accounted for survivorship bias when tasked to reduce bomber losses to enemy fire. He surmised the aircraft that returned could take damage to the areas that had received fire, so he recommended armoring the aircraft in the locations where “the bullets weren’t”.