The difference between an interesting observation and knowledge
May 2027 prescribed title 4 asks how far you accept a claim attributed to Agatha Christie about noticing coincidences and discarding them later. The required area of knowledge is mathematics, together with another area that you select. The complete prompt, including its exact punctuation and title number, belongs on your school's authorized title sheet. See the title list and verify it with your coordinator rather than trusting another website's reproduction.
There are at least three different decisions: notice an observation, spend resources testing a possible explanation, and accept a claim as justified. The first can be inexpensive; the second can be costly; the third requires a standard of evidence. Saying that every coincidence is worth attention makes no distinction between these stages.
Why matching birthdays are less astonishing than they look
In a room of 23 people, the probability that at least two people share a birthday is about 50.7% under the simplified model of 365 equally likely dates and independent birthdays. A student who notices a match might mistakenly treat it as too unlikely to be chance. The error is often comparing the probability of two specified people sharing a birthday (1/365) with the probability of any pair matching across 23 people.
Any pair shares a birthday
50.73%
There are 253 possible pairs, but their matching events are not independent.
Two people specified in advance
0.27%
The probability of a match for one specified pair is only 1/365 in this simplified model.
Assumptions: birthdays are independent and uniformly distributed across 365 dates. Leap days and population seasonality are ignored. Changing those assumptions changes numerical probabilities; the difference between a specific pair and any pair remains crucial.
The correct complement calculation multiplies the probability that each successive person avoids all preceding birthdays: P(at least one match) = 1 − (365/365)(364/365)...((365−n+1)/365) for 0 ≤ n ≤ 365. The formula is an exact deduction from the model assumptions. Real birthdays are not perfectly uniform; that limitation does not invalidate the model's core lesson about comparing the right events.
The mathematical insight is not “never investigate coincidences.” It is that a pattern can be expected when there are many opportunities for it to appear. An investigator must ask how many patterns could have been noticed, not only how surprising the observed one sounds after the fact.
A natural-science coincidence has an extra burden
Consider a hypothetical laboratory class that records a spike in reaction rate on the same day that the room becomes unusually warm. The coincidence may suggest a temperature mechanism, but a causal conclusion would require investigating the sensor, reactant concentrations, mixing conditions, timing and repeat trials. The initial co-occurrence alone cannot distinguish a genuine thermal effect from a miscalibrated instrument or an unnoticed reagent change.
One targeted test would hold reactant concentrations and method constant, deliberately vary only temperature within an appropriate safe range, replicate independent trials, and examine whether the measured rates change in a reproducible direction. Predict the result before collecting the new data; otherwise the investigator can repeatedly redefine what counts as a meaningful coincidence. This scenario is a fictional illustration, not a report of an existing scientific study.
What changes across mathematics and the natural sciences?
| Knowledge question | Mathematics | Natural sciences |
|---|---|---|
| What counts as a discovery? | A conjecture or mathematically unexpected relationship that can be defined precisely | A repeatable observation demanding a plausible model or an alternative explanation |
| How is a claim checked? | Proof or a mathematically specified probability model, with assumptions explicitly stated | Measurement design, controls, replication, uncertainty and discriminating predictions |
| When can it be discarded? | When a pattern is mathematically unsurprising under the correct sampling model | When it fails well-designed independent tests or has stronger mundane explanations |
| What does persistence mean? | A proven theorem holds within its assumptions, independently of further sample collection | Confidence grows through converging empirical evidence but remains revisable |
Construct a thesis that can survive a counterexample
A plausible original planning hypothesis is: noticing a coincidence can have exploratory value, but investigating every coincidence is not equally rational; the decision depends on the prior plausibility, the cost of testing, the risk of error and the standards of justification within the area of knowledge. Mathematics can explain why a particular coincidence is expected under a precise model. Natural sciences must also decide how to design an observation that can distinguish causes.
A counterargument deserves equal care: surprising events have occasionally motivated useful inquiries. If early researchers had discarded all unusual results immediately, some fruitful hypotheses would never have been tested. Your response should therefore distinguish recording an anomaly from treating it as an important scientific discovery. Avoid presenting birthday-matching as a full TOK essay; your final argument needs independently verified real-world examples, critical comparison and your own assessment of limits.
TOK essays have a maximum length of 1,600 words. Consult the exact May 2027 prescribed-title sheet and your school's academic-integrity rules before producing your own response.
Sources & verification
Primary-source references used to verify key assessment facts on this page. Last source check: October 14, 2026.
Interactive probability assumes 365 independent uniformly distributed possible birthdays; natural-science scenario is explicitly fictional. Neither constitutes a submitted TOK essay or marks prediction.
- IB Theory of Knowledge courseOfficial overview of TOK, essay assessment and knowledge questions.
- May 2027 reported prescribed titlesDate-specific title wording reported to schools, to be confirmed with the authoritative school copy.
- Birthday problem probability derivationMathematical reference for the complement probability of no matching birthdays.