🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
survivorship_bias
Survivorship bias is the logical error of concentrating on entities that passed through a selection process while overlooking those that did not, leading to overly optimistic conclusions. By studying only the 'survivors' (successful companies, published studies, living species), one misses the full picture that includes the far larger number of failures, creating a distorted view of what leads to success.
Business books study only successful companies like Apple and Google to extract 'principles of success,' ignoring thousands of companies that followed identical strategies but failed. The extracted principles may have nothing to do with actual success.
A fitness influencer attributes their chiseled physique solely to a specific supplement regimen and posts about it daily, gaining thousands of followers who buy the product. The countless people who followed the exact same regimen without notable results never post about it, so the supplement's failures remain invisible.
A film school professor analyzes the career paths of celebrated directors like Spielberg and Nolan to teach students how to break into Hollywood, without acknowledging the tens of thousands of equally talented graduates who followed similar paths and never got a single film made.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Does the analysis only consider successful cases while ignoring failures?
Type: binaryAre conclusions drawn from a sample that excludes those who dropped out or failed?
Type: binaryWould the conclusion change if non-surviving cases were included?
Type: binarySurvivorship bias is the logical error of concentrating on entities that passed through a selection process while overlooking those that did not, leading to overly optimistic conclusions. By studying only the 'survivors' (successful companies, published studies, living species), one misses the full picture that includes the far larger number of failures, creating a distorted view of what leads to success.
Failures are invisible because they are removed from the dataset by the selection process itself. Only successes remain visible for analysis, creating the illusion that the characteristics of survivors are the causes of survival rather than potentially irrelevant or even common to failures too.
Always ask 'Where are the failures?' and seek data on the full population, including those who did not survive the selection process. Compare survivors' characteristics against a representative sample of non-survivors.
Survivorship bias affects investment analysis (only surviving funds are tracked, inflating average returns), music and art appreciation (we only hear the best works from past centuries), and military strategy (the famous WWII example of reinforcing planes where returning aircraft were hit, rather than where they were not).
The anecdotal argument fallacy occurs when personal experiences, individual stories, or isolated examples are presented as sufficient evidence for a general claim. While anecdotes can be valuable for illustration, hypothesis generation, or making data relatable, they are unreliable as evidence because they are subject to selection bias, survivorship bias, memory distortion, and the representativeness heuristic. A single vivid story can psychologically overwhelm statistical evidence covering thousands of cases.
Confusing selection factors with results. The assumption that swimmers have athletic bodies because of swimming, when in reality people with certain body types are selected for (or gravitate toward) competitive swimming. A specific manifestation of the broader confusion between selection effects and causal effects.
Use these tools to detect, analyze, or train this aspect.