Plain-language answer
An anecdote is a report of a single case. Data, in the relevant sense, is a systematic record of many cases collected in a way that lets you count how often something actually happens, including the instances that were unremarkable or didn’t fit the pattern. Both can be entirely accurate as far as they go — the problem with treating an anecdote as data is not that the anecdote is false, but that a single case (or a handful of hand-picked cases) cannot answer a question about general frequency, no matter how detailed or vivid it is.
Why it matters
Coincidence claims are built almost entirely from anecdotes, because a coincidence is, by definition, a single memorable event. That makes this distinction the single most important discipline for evaluating any coincidence story: the anecdote can establish that an event happened and what it felt like, but it cannot, on its own, establish how surprising the event actually was, because that requires a systematic account of how often similar events occur across everyone who could have had one — exactly the kind of denominator an anecdote never includes.
Worked example: why compelling anecdotes are still weak evidence for frequency claims
Consider a chain of research claims: a single vivid case study; a small, uncontrolled series of similar cases; and a large, pre-registered, properly controlled study. Even setting hidden causal or paranormal explanations aside entirely, formal research on evidence quality documents a persistent gap between how solid a finding looks in early, small, or informally selected reports and how it holds up once tested more rigorously — analyses of the published research literature suggest a substantial share of initially reported findings do not hold up under more careful scrutiny.(Ioannidis, 2005) If findings published through peer review with formal statistical testing still often fail this way, purely anecdotal claims — collected with no control at all over how they were selected, retold, or embellished — sit on considerably weaker ground still.
Methodological treatments of coincidence collections make the same point directly: rigorous analysis of a striking coincidence requires knowing how it was found, how many similar-but-unremarkable cases were not reported, and what the actual population of opportunities looked like — none of which a single anecdote, however well told, typically supplies.(Diaconis & Mosteller, 1989)
Common misconception
“I know it’s just one story, but it’s really compelling, so it must mean something” treats vividness as if it substitutes for a missing denominator. A vivid, well-documented, entirely truthful anecdote still cannot answer “how often does this happen, compared with how often it doesn’t?” — that requires counting cases the anecdote never mentions.
Limits and open questions
None of this means anecdotes are worthless. A single case can be the first evidence that prompts a real, systematic investigation, can illustrate a mechanism vividly, or can be the entire point when the question is about personal meaning rather than general frequency (see personal meaning without causal proof). The distinction that matters is narrower and more practical: don’t use an anecdote to answer a question about how common or improbable something is in general, because that specific question needs a systematic count that an anecdote structurally cannot provide.
Related
- Selection effects covers why the anecdotes that do circulate are themselves a biased, non-random sample of everything that happened.
- Replication and publication bias covers the same underlying problem — reporting practices that favor striking results over a representative record — one level up, inside formal published research.
Key takeaways
- An anecdote can be completely accurate and still be the weakest available evidence for a claim about general frequency, because it lacks a denominator of unremarkable cases to compare against.
- Even peer-reviewed, formally tested research findings often fail to hold up under closer scrutiny — a caution that applies with even more force to uncontrolled anecdotal reports.
- Anecdotes remain valuable for illustrating a mechanism or expressing personal significance; the discipline is in not using them to answer a question about how common or improbable something is overall.