Concept
Survivorship Bias
cognitive-bias evidence decision-making
Judging what leads to success by studying only the survivors or winners of a process, while the far larger set of failures that never became visible goes uncounted — systematically distorting the apparent lesson. Melina Palmer traces the name to a WWII analysis: engineers examining returning bomber planes for bullet-hole patterns initially proposed reinforcing the spots where holes were most common. A statistician pointed out the flaw — those planes survived being hit there; the real lesson was to reinforce the spots with no bullet holes, since planes hit there likely never made it back at all (see WWII Bomber Survivorship Bias Analysis).
Entrepreneurship application. Business advice overwhelmingly comes from successful survivors (a college dropout who became a billionaire), while the much larger population of people who tried the same thing and failed goes unheard from and unstudied — following "what worked for them" is therefore unreliable, since the very selection of whose story gets told is already biased toward success. Palmer illustrates this with a business-simulation classroom exercise: students who won using a specific strategy one year, then reused the exact same inputs the following year in a different competitive field, often lost badly — success the first time didn't mean the strategy was actually sound, only that it worked in that particular field.
A further, non-business demonstration: a 1987 study reported that cats falling from below six stories were injured more severely than cats falling from six stories or higher, seemingly because falling further let cats reach terminal velocity and right themselves. The more likely explanation is survivorship bias: cats that die in high falls are far less likely to be brought to a vet and recorded at all (see 1987 Falling Cats Survivorship Bias Study).
A closely related but distinct statistical trap: restriction of range (Episode 225). Alex Edmans critiques Angela Duckworth's Grit on this basis: her famous West Point study found "grit" predicted cadet success better than fitness did, but every cadet in the sample had already been admitted to West Point — meaning fitness had already been filtered to a narrow, uniformly high range before the study began, artificially suppressing its apparent predictive power relative to grit. Unlike survivorship bias (studying only the survivors, with failures entirely absent from the sample), restriction of range studies a population where a variable's range has been artificially narrowed by a prior selection step, distorting which factors appear to matter (see Duckworth Grit West-Point Restriction-of-Range Case).
Discussed in
- Episode 45 — 45- 3 Crucial Lessons for Entrepreneurs
- Episode 225 — 225-why-most-bestselling-business-books-are-bs (restriction of range, a related trap)