Fooled by Randomness #2
The null hypothesis asks what you would expect to see if the proposed explanation were false. A surprising result becomes evidence only after you compare it with the right baseline and account for how many chances randomness had to produce it.
Rohit Sharma lost 15 coin tosses in a row. Match-fixing sounds tempting—until you stop asking whether the streak is individually strange and ask how many matches offered some streak the chance to occur.
E1 R1Count the chances, not just the coincidence
Begin with a world in which nothing unusual is happening: the coin is fair, the lottery is random, or the treatment has no effect. Then ask how often that ordinary world could still produce the observation. Four consecutive heads has a probability of about 6% in one specified run—uncommon, but hardly a conspiracy. Even a far rarer sequence becomes less astonishing when many matches, players, and possible starting points create repeated opportunities for one to appear.
The moving parts are the baseline process, the probability of the observed result under that process, and the number of opportunities available. Ignore the last part and you become Fooled by Randomness: you notice the spectacular survivor while overlooking all the uneventful trials that made it possible.
E1Where it shows up
Multiples of nine
The Philippine lottery drew 9, 18, 27, 36, 45, and 54. The pattern looks designed because humans find orderly sequences suspicious. But its visual neatness is not itself evidence of rigging; the relevant question is how this draw compares with all the other equally specific combinations randomness could have produced.
R1The guaranteed male child
A ₹5,000 Ayurvedic treatment claims to guarantee a male child and arrives with testimonials, celebrity photos, acquaintances who report success, and even a money-back guarantee. Those successes answer the wrong question. The null asks how many male children would have been born without the treatment—the baseline needed before crediting its mechanism.
R1Rare does not mean impossible—or innocent
A null hypothesis does not prove that a coin is fair, a lottery honest, or a treatment useless. It disciplines suspicion. The conclusion depends on choosing the right baseline and counting the relevant opportunities; a distorted baseline can explain away genuine manipulation just as easily as a vivid coincidence can manufacture it.
E1 R1Write the no-effect prediction first
Before evaluating tomorrow’s striking claim, write one sentence completing: “If this explanation were false, I would still expect…” Then estimate the ordinary success rate and count how many attempts could have generated the showcased result. Judge the claim against that number, not against how uncanny its story feels.
E1 R1Episodes that teach this
- Science Explains Rohit Sharma Losing 12 tosses in a Row? - Future IQ start here 5,873 views