Plain-language answer
Coincidence stories come in recognisable shapes: two events close in time, an unexpected encounter, a name or a number that repeats. Sorting a story into one of these shapes doesn’t explain it, but it tells you which question to ask next — because the question that matters for a chance meeting (“how dense is this social network?”) is not the question that matters for a biographical parallel (“which details were searched, and which were left out?”).
Why it matters
“Coincidence” is not one thing statistically. Treating every case with the same generic “what are the odds” response hides the fact that different kinds of coincidence depend on entirely different quantities — number of opportunities, network density, flexibility of a matching rule, or how many details were quietly dropped to make two stories line up. The table below is a set of descriptive lenses, not a set of mutually exclusive natural kinds: a single story can belong to more than one row.(Diaconis & Mosteller, 1989)
The eleven lenses
| Type | Working description | Key analytical question |
|---|---|---|
| Temporal | Similar or linked events occur near each other in time | How wide was the time window? |
| Spatial | Events or people unexpectedly converge in place | How many locations or opportunities existed? |
| Resemblance | Names, appearances, phrases, or narratives match | How flexible is the match criterion? |
| Repetition/clustering | Similar events occur in runs | Is the clustering unusual under the underlying process? |
| Personal encounter | An unexpected meeting or social connection | How dense and assortative is the network? |
| Information coincidence | A thought, dream, or statement seems to anticipate information | Was it recorded before the outcome, or only remembered after? |
| Biographical parallel | Two lives appear to share striking details | Which details were searched for, and which were quietly omitted? |
| Historical parallel | Events from different periods are aligned | Are the facts accurate, and were the comparison rules fixed in advance? |
| Numerical | Dates, numbers, or counts align | How many transformations or roundings were allowed? |
| Serendipity | A chance observation leads to a useful discovery | Which causal steps, after the chance event, actually did the work? |
| Manufactured/pseudo-coincidence | Framing or selection creates an apparent pattern | Who chose the sample, and after seeing what? |
A few of these reward a closer look.
Resemblance and flexible matching. The more ways two things are allowed to “match” — same name, similar-sounding name, same initials, matching nickname — the less surprising a match becomes, because you are really asking whether any of many loose criteria succeeded, not whether one specific one did.
Biographical and historical parallels. These are the cases most vulnerable to selective retelling: a list of twelve ways two lives were alike can look compelling until you ask how many ways they differed, and whether those differences were left out of the story.
Manufactured or pseudo-coincidence. Sometimes the appearance of a pattern is produced entirely by how a sample was chosen — for instance, only collecting or publicising cases that already fit a striking pattern. The “coincidence” here is in the selection process, not in the world.
Common misconception
Calling something a “temporal” or “numerical” coincidence is a description of its shape, not a verdict on whether it is meaningful or improbable. Two categories can also combine: an unexpected reunion (personal encounter) that happens on a significant date (temporal, numerical) is analysed by asking both sets of questions, not by picking the one that sounds most dramatic.
Limits and open questions
These eleven lenses are a practical sorting tool, not a scientific classification with fixed boundaries. Reasonable people can disagree about which lens fits a given story best, and a good case analysis usually has to work through more than one.
Related
- What is a coincidence? separates the event, the surprise, and the causal claim that any of these lenses can carry.
- Probability is a model, not a verdict develops the “how many opportunities” question that several rows above depend on.
- Cases will apply this taxonomy to specific, sourced examples as they are published.
Key takeaways
- A coincidence story’s “shape” tells you which analytical question to ask — not how improbable or meaningful it is.
- Many real cases combine more than one lens at once.
- The categories most prone to manufactured patterns are resemblance, biographical parallel, and manufactured/pseudo-coincidence — all three hinge on how flexibly a match was defined and how a sample was chosen.