🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
birthday_problem
The birthday problem demonstrates that people grossly underestimate coincidence probability because they think about individual probabilities rather than the number of possible pairs. With just 23 people, the probability of any two sharing a birthday exceeds 50%. This intuition failure has serious consequences in forensic DNA matching, security system design, and coincidence reasoning in statistical claims.
In a group of 23 people, most people guess there is roughly a 23/365 ≈ 6% chance that two people share a birthday. The actual probability is 50.7%. With 57 people, the probability reaches 99%. This is because 23 people generate 253 possible birthday pairs.
A teacher tells her class of 30 students that there's probably a shared birthday in the room. The students laugh it off, each thinking their own birthday has only a 1-in-365 chance of matching anyone else's. They're stunned when two students actually share a birthday — unaware the true probability was about 70%.
At a company team-building event with 40 employees, the HR manager bets the group that at least two people share a birthday. Most employees take the bet, estimating the odds at around 10%. The HR manager wins easily — the actual probability was roughly 89% — because employees failed to account for the explosion of possible pairs.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Does the claim involve the probability of any two members of a group sharing a characteristic?
Type: binaryIs the probability of coincidence being estimated by thinking about a single individual rather than all possible pairs?
Type: binaryIs the number of possible pairings growing quadratically with group size being ignored?
Type: binaryIs a low per-pair coincidence probability being used to dismiss the overall probability of at least one coincidence?
Type: binaryThe birthday problem demonstrates that people grossly underestimate coincidence probability because they think about individual probabilities rather than the number of possible pairs. With just 23 people, the probability of any two sharing a birthday exceeds 50%. This intuition failure has serious consequences in forensic DNA matching, security system design, and coincidence reasoning in statistical claims.
The human tendency is to calculate the probability for one specific pair, ignoring the combinatorial explosion of possible pairs as group size grows.
Calculate the number of possible pairs: n(n-1)/2. Then estimate the probability of at least one coincidence as 1 minus the probability of no coincidences across all pairs. Apply this reasoning whenever evaluating claims about coincidences in populations.
Forensic labs comparing DNA profiles against large criminal databases face the birthday problem: with millions of profiles, random matches between unrelated individuals become statistically inevitable.
Use these tools to detect, analyze, or train this aspect.