Streetlight Effect (Drunkard's Search)

Also Known As: Drunkard's Search Principle Observational Convenience Bias Lamppost Effect

Statistical Error ID: streetlight_effect

Definition

The Streetlight Effect is the tendency to search for answers only where it is easiest to look, rather than where the answer is most likely to be found. Named after a joke about a drunk searching for his keys under a streetlight (not because he lost them there, but because the light is better), this bias affects research design, data analysis, and everyday problem-solving. It leads to systematic gaps in knowledge and false conclusions drawn from incomplete evidence.

Examples

A company surveys only existing customers to understand why sales are falling. The critical information — why potential customers chose competitors — lies entirely outside their data. They conclude 'customers love our product' while the real problem (pricing) remains invisible because non-customers were never asked.

A hospital measures quality of care using metrics it already tracks: readmission rates, length of stay, infection rates. Patient experience, care coordination, and the quality of the last conversation before discharge — harder to quantify — go unmeasured and therefore unimproved.

An AI model trained on digitised books and web text performs poorly on spoken-language dialects, legal handwriting, and low-resource languages. The model learned from what was easy to collect, not from what would make it representative. Its blind spots are precisely the areas that weren't in the data.

Verification Steps

Verification Steps

Binary yes/no questions that an AI must answer to detect a reasoning pattern in a text.

Each of the 452 aspects has verification steps — simple yes/no questions designed to systematically detect whether a pattern appears in a text. For ad hominem: "Does the argument attack a person rather than their claim?" For false dichotomy: "Are only two options presented when more exist?" This ensures consistent, reproducible analysis.

View in glossary →

Binary (yes/no) questions an LLM must answer to identify this aspect: