Confirmation bias, in full
The English psychologist Peter Wason gave the effect its name and its best-known demonstration. In an experiment published in 1960 he showed people the triple 2-4-6 and asked them to work out the rule behind it by proposing further triples, each of which he would mark as fitting or not. The actual rule was simply any ascending sequence. Very few participants found it, and most announced something far more specific with far too much confidence.
The reason was the kind of test they chose. Having guessed something like even numbers rising in twos, they offered 8-10-12, then 20-22-24, and collected confirmation after confirmation. A single try at 3-7-11, a triple their own guess forbade, would have settled the matter in one move. Wason's later selection task found the same habit with logical rules. People reach for the case that fits and rarely for the case that would break it.
Applied to number-spotting, the mechanism is bookkeeping. A day on which 777 turns up is a day you remember, and a day it does not is a day with nothing to file. Nobody keeps the denominator. The same asymmetry makes a lottery system look effective to the person using it, because the near-misses that felt close get retold and the ordinary losing tickets do not.
- Pattern-seeing
- Confirmatory bias, Myside bias
A worked example
Ask someone convinced their birth date is lucky to list the times it paid off. They will produce several. Now ask how many draws they entered with it and won nothing. That figure is almost never known, because it was never recorded. In a 6/49 game the odds on one line are 1 in 13,983,816, so a lifetime of weekly play, around 2,600 tickets, buys an expected jackpot count far below one.
Sources
Peter Wason published the conceptual-rule experiment behind the effect in 1960, in the Quarterly Journal of Experimental Psychology 12(3), 129-140.
Quarterly Journal of Experimental Psychology — On the Failure to Eliminate Hypotheses in a Conceptual Task
See also
Where it comes up on this site
Confirmation bias: frequently asked
Participants were shown the triple 2-4-6 and asked to work out the rule behind it by proposing further triples, each marked as fitting or not. The rule was any ascending sequence. Most people guessed something far narrower, then tested only triples their own guess allowed, collected yes after yes, and announced the wrong answer with considerable confidence. Wason published it in 1960.
It corrupts the record. Wins and near-misses get remembered and retold. The long run of ordinary losing tickets leaves nothing behind, so a system can feel like it works on a sample nobody ever counted.