Survivorship Bias
You'll learn how ignoring failures and only studying successes can trick you into dangerously wrong conclusions.
Survivorship bias occurs when you analyze only the cases that "survived" some selection process while overlooking those that didn't make it. This creates a distorted view because failures often contain the most important information.
The Classic Example
Imagine studying successful startups to learn what makes businesses thrive. You interview founders of 100 thriving companies and notice they all took big risks early on. Should you conclude that big risks lead to success?
Not so fast. You've ignored the thousands of failed startups that also took big risks but went bankrupt. By focusing only on survivors, you've mistaken correlation for causation and reached an overoptimistic conclusion about risk-taking.
Why It's a Sampling Problem
Survivorship bias is fundamentally about biased samples. When failures drop out of your data collection process (because they went out of business, dropped out, or stopped responding), your sample no longer represents the true population. You're essentially doing non-probability sampling where the probability of being included depends on success.
Real-World Consequences
- Investment advice based only on winning stocks ignores all the losers that went to zero
- Medical trials that only track patients who completed treatment miss those who got too sick to continue
- Product reviews skew positive when only satisfied customers bother to write them
This connects directly to selection bias and nonresponse bias you've already learned—survivorship bias is a specific type where the "non-survivors" physically can't be in your dataset.
Key Takeaway: Always ask yourself: "Who or what is missing from this data?" Survivors alone tell an incomplete, misleadingly optimistic story.