Survivorship bias
Survivorship bias is a cognitive bias that occurs when an analysis only considers the 'survivors' or successful entities of a group while overlooking the failures, thereby skewing the results and leading to erroneous conclusions. This bias stems from the human tendency to draw inferences based on incomplete data, largely due to the absence of information about non-survivors.
How it works
Survivorship bias works by filtering out non-survivors or failures and focusing only on successful cases, often leading to misleading conclusions. This results in a distorted view because the data analyzed represents only those who 'survived' or succeeded, rather than a complete picture that includes both successes and failures.
Examples
- During World War II, damage assessment of returning aircraft led to recommendations for reinforcing areas that survived attacks. Statistician Abraham Wald suggested reinforcing areas with no bullet holes, realizing that aircraft hit in those areas didn’t return at all.
- In finance, evaluating mutual funds based only on the current winners ignores funds that have performed poorly and closed, leading to an overestimation of industry-wide success rates.
- In entrepreneurship, focusing on stories of successful entrepreneurs but ignoring those who failed can create a misleadingly positive picture of success probabilities.
Consequences
Survivorship bias can lead to overly optimistic expectations and underestimation of risks. By focusing only on successful outcomes or surviving entities, important lessons from failures are often disregarded, which can result in flawed strategies or the misallocation of resources.
Counteracting
To counteract survivorship bias, it is essential to seek out data on both success and failure cases, ensuring a comprehensive analysis. Including diverse perspectives and fostering an awareness of omitted data can help mitigate the effects of this bias.
Critiques
One critique of survivorship bias is that it oversimplifies complex systems by ignoring a multitude of factors leading to success or failure. Critics argue that while acknowledging survivorship bias is beneficial, real-world scenarios often involve variables that are not easily accounted for.
Fields of Impact
Also known as
Relevant Research
Historical patterns of aircraft survivorship: An application of survivorship bias
Piatkowski, T. M., & Krug, E. J. (2019)
Journal of Historical Economics and Econometric History
Does the stock market overreact? The Journal of Finance, 40(3), 793-805
Bondt, W. F. M., & Thaler, R. (1985)
Performance persistence
Brown, S. J., & Goetzmann, W. N. (1995)
The Journal of Finance, 50(2), 679-698
Case Studies
Real-world examples showing how Survivorship bias manifests in practice
Context
A mid-sized venture capital firm set out to replicate the traits of recent billion-dollar startups after a string of friends and partners celebrated high-profile exits. The firm's investment committee distilled a shortlist of traits (technical founder, rapid initial user growth, strong network introductions) and used those as hard criteria for new deals.
Situation
Over two fundraising cycles the firm screened 1,200 early-stage startups and only actively tracked the 40 that matched the 'unicorn profile.' The investment team built internal scorecards and dashboards reflecting patterns observed in those 40 winners and used them to make follow-on investment decisions.
The Bias in Action
The partners' analysis relied almost entirely on the successful cohort and ignored the 1,160 startups that either failed, stagnated, or pivoted away from those traits. Because only survivors were analyzed, correlations between traits and success were inflated — for example, they concluded that having an Ivy League founder increased exit probability by 4x, when in reality many Ivy founders had failed and simply weren't in the tracked dataset. The firm's decision rules began rejecting promising nonconforming teams and overallocating to companies that fit the 'profile' but showed fragile unit economics. Feedback loops (more funding, more introductions) amplified perceived success factors while hiding the many counterexamples.
Outcome
After five years the fund's performance fell short of expectations: the portfolio produced an aggregate 0.85x return (net, across the fund) instead of the targeted 2.0x, and the follow-on reserve allocation concentrated in profile-fitting companies generated most of the losses. Team morale suffered as junior partners realized many rejected companies later achieved modest success, and LPs questioned the diligence process.