Survivorship Bias

a.k.a Silent Evidence

def - the bias of focusing the attention on the people who were successful in a particular endeavor, rather than focusing on the whole population, which would include a far greater number of those who are unsuccessful.

  • It is a form of seletion bias

Essentially, survivorship bias can exist anywhere that the voices of the survivors are the only ones existing. Those that did not make it past the gate are not able to add their voices to the conversation to add equity to the discussion.

ex. When we analyze companies, failed companies are excluded from the analysis, giving an illusion as to why the successful companies succeeded in the first place.

Any business that focuses on giving advice is an inherent victim of survivorship bias, since of course those who failed will never have their voices heard

Examples

when you go to the hospital you'll see a lot more people will bullet wounds in their legs than people will bullet wounds in their chests. This doesn't mean people don't get shot in the chest; it means that people who get shot in the chest don't recover.

Abraham Wald's Aircraft Bullet holes

During WWII, there were a number of military planes that survived gunfire and made it back to the military bases for repair. The military's stance was to reinforce the areas that were penetrated most by gunfire. A mathematician named Abraham Wald realized that this was succumbing to survivorship bias, since the planes that did not survive were not there to tell their story. Most likely, those planes did not survive because they were not hit in the same places as the survivors. This meant that the spots that needed reinforcing were the parts of the plane that were not even hit on the survivors. The likely reason why the failed-planes were shot down was because the areas they were hit in were not the same areas that the survivors were hit in, causing them to fall.

  • ex. during a shipwreck the praying people recount that it was their praying that saved them. however, what we dont see are the stories of those who prayed and also perished
  • silent evidence is what events use to conceal their own randomness (particularly the black swan variety)
  • neglect of silent evidence is endemic when we study comparative talent (particularly in winner-take-all activities)
  • consider 2 professions: writers and doctors. there are far more writers who work at Starbucks than those who make a professional living as a writer. on the other hand, there are hardly any doctors who work at anything other than medicine. therefore we can use the second group as an effective sample of the population, whereas the first would provide a poor sample (since we'd likely be neglecting those writers working at Starbucks)
  • consider professions like doctor to be ones without superstar effects, making their sample more representative
  • this doesnt mean that a person who becomes a superstar isnt skilled, but it does imply that there are many non-survivors who are equally as skilled, but just didnt have the luck that the survivors had. in other words, they arent less skilled, but they are less uniquely skilled
  • ex. a newspaper article says "russian mafia composed of members hardened by the gulag". to see why this causative statement is a fallacy, consider a good cross section of the rat population: big, small, fast etc. shine a shot of radiation at all of them, so that the weaker ones die. what you are left with is a subpopulation of stronger rats. if we were to be ignorant of the ratiation that was shot at the rats, we would get a false impression of the average strength of the rat population.
  • survivorship bias hides best when thr impact is largest (ie. the less survivors, the more they will come to "claim" to represent the whole group)
  • dont forget that survivorship bias can work in the opposite way too.
    • ex. consider that its likely that our perception of a criminal is flawed, since the less intelligent ones are more likely to get caught.
  • consider evolution. perhaps most mutations have a negative consequence. you might be tempted to think that evolution is always improving us, but that's not the case. the negative mutations kill off the species that receive them, putting them in the silent graveyard that no one talks about and considers as part of the grander statistic
  • in order to understand success and what caused it, you must look at the failures and study the traits present in those
    • ex. studies of rich and famous people focuses exclusively on them. they take a population of hot shots, those of big titles and big jobs, and study their attributes. they look at what those big guns have in common: courage, risk-taking, optimism etc. it is then inferred that these traits help you become successful. take this scenario and apply it to a gym, or anything where there will be people who invariably drop out along the way. the resulting population skews our perceptions of reality

The expression “what Does not kill me makes me stronger” might actually mean “what did not kill me did not make me stronger, but spared me because I am stronger than others. It killed others, and as a result of the average population is now stronger, because the weak are gone.”

  • Here we see the transfer of Antifragility from the individual to the system. The system gets stronger at the expense of its weaker components

Silent Victims

consider when an event like hurricane katrina happens. governmental money gets redirected from places like cancer, where more deaths occur each year as a result.

  • a life saved is a statistic; a person hurt is an anecdote consider when we look at someone having built in job security as a result of union policy. we ascribe this to social benefits, but fail to consider the person who has a tough time finding a job as a result consider 9/11 and the people afterwards who were scared of flying. they instead opted to drive, thereby increasing the amount of deaths (since flying is riskier than driving)

Backlinks