Cognitive Bias Probability

Survivorship Bias Awareness

Understanding that focusing only on successes obscures the failures that inform realistic probability assessments.

Quick Definition

Understanding that focusing only on successes obscures the failures that inform realistic probability assessments.

Definition

Survivorship bias occurs when we concentrate on people or things that made it past some selection process and overlook those that did not, typically because of their lack of visibility. By focusing exclusively on the survivors of a particular endeavor or situation, we systematically overestimate the probability of success and misunderstand the true characteristics of the population.

This bias leads to flawed conclusions because the failed cases—which are often equally instructive or more instructive—are missing from our analysis. The classic example involves WWII aircraft armor studies, where engineers wanted to reinforce planes based on returning aircraft, but the missing planes held the crucial information about where they were actually being hit.

Origin & History

The concept gained recognition through Abraham Wald's work during World War II at the Statistical Research Group. When the US Army Air Forces asked where to add armor to bombers, Wald pointed out that the data they had—bullet holes in returning planes—represented places planes could be hit and survive.

The holes in returning planes indicated where armor was NOT needed because those hits didn't bring the plane down. The missing data—planes that didn't return—showed where hits were fatal. This insight saved countless lives. The term "survivorship bias" emerged later as psychologists and statisticians studied errors in reasoning about past successes.

Key Principles

  • Identify the visible population - When studying successes, ask what population they were selected from
  • Consider the selection process - Understand what criteria determined who ended up in your visible sample
  • Imagine the counterfactuals - Ask "What would I see if I could observe the failures?"
  • Seek out failure data - Actively look for information about failures, not just successes
  • Question "lessons from success" - Be skeptical of advice unless it also studies failures

When to Use

  • Evaluating success stories and case studies
  • Learning from others' experiences
  • Estimating probabilities of success
  • Considering career paths or business ideas
  • Reading business books or biographies
  • Evaluating investment strategies

How to Apply

  1. Identify the visible population - When studying successes, ask what population they were selected from
  2. Consider the selection process - Understand what criteria determined who ended up visible
  3. Imagine the counterfactuals - Ask "What would I see if I could observe the failures?"
  4. Seek out failure data - Actively look for information about failures
  5. Adjust probability estimates - Remember that for every visible success, there may be many invisible failures
  6. Question "lessons from success" - Be skeptical unless they also study failures
  7. Consider your own selection - Recognize what selection process brought you to your current position

Real-World Example

Startup Success Advice: Successful entrepreneurs often give advice about what made them successful. But survivorship bias means we don't hear from the vast majority who followed the same advice and failed. The advice may be wrong, incomplete, or only relevant under specific conditions that can't be replicated.

Common Pitfalls

  • Studying winners only - Analyzing successful cases without understanding the failure rate
  • Invisible selection - Not recognizing that the visible population was selected through some process
  • Success-story oversampling - Overweighting memorable success stories over systematic data
  • Misunderstanding success factors - Attributing success to wrong causes because failures aren't visible
  • Ignoring base rates - Not adjusting for how rare success actually is
Back to Category All Models
Previous: Confirmation Bias Awareness | Next: Regression to Mean