Psychology and behaviour
The mechanism is not stupidity. It is a well-documented shortcut for judging frequency — combined with an information environment that shows you every winner and none of the 292 million losers.
People are not bad at lottery odds because they cannot do arithmetic. They are bad at lottery odds because of how humans estimate frequency in the first place, and because the information they are given is filtered in one direction.
Both halves of that have names and evidence behind them.
Amos Tversky and Daniel Kahneman, Availability: A Heuristic for Judging Frequency and Probability, Cognitive Psychology, volume 5, issue 2 (1973), pages 207–232.
The paper's claim is simple and it has held up for half a century: when people judge how frequent or how probable something is, they do not retrieve a base rate. They assess how easily instances come to mind, and use that ease as the estimate.
That is usually a decent shortcut. Things that happen often are, on average, easier to recall than things that happen rarely, so ease of retrieval correlates with frequency. The heuristic fails precisely when something is easy to recall for a reason other than how often it happens — because it is vivid, recent, emotionally charged, or heavily publicised.
A lottery jackpot is all four.
Consider what you actually observe about a lottery.
You see the winner. Their name, sometimes their face, the oversized cheque, the shop that sold the ticket, the interview about what they will do with the money. Lottery operators publicise winners deliberately — in several jurisdictions publicity is a condition of payment, as set out in which countries let you claim anonymously.
You never see a losing ticket. Nobody photographs one. There is no press release. There is no interview.
This is not a conspiracy; it is what news is. But the consequence is that your mental sample of "what happens when someone plays the lottery" is drawn entirely from the winning tail.
For US Powerball the jackpot probability is one in 292,201,338 — the working is in how lottery odds are calculated.
Suppose a broadcaster decided to give equal airtime to losers, at one losing combination per second, around the clock:
| Quantity | Value |
|---|---|
| Losing combinations per jackpot combination | 292,201,337 |
| Seconds needed at one per second | 292,201,337 |
| Days | 3,382 |
| Years of continuous broadcast | 9.3 |
One winner gets ninety seconds on the evening news. Giving the losers the same per-case coverage would take nine years and three months of uninterrupted transmission. That asymmetry is the entire mechanism. Your sense of how often people win is built from a sample that has had 99.9999997% of its cases removed.
For the same idea rendered physically rather than in seconds, see picturing one in 292 million.
If availability were only a laboratory finding it would be less interesting. It shows up in real purchasing.
Jonathan Guryan and Melissa S. Kearney, Gambling at Lucky Stores: Empirical Evidence from State Lottery Sales, American Economic Review, volume 98, issue 1 (2008), pages 458–473, examined Texas Lotto retailers. In the week after a store sold a large winning ticket, that store's sales rose by 12 to 38 percent relative to comparable retailers, and some of the increase persisted for up to 40 weeks.
Nothing about the store changed. The odds of the next ticket sold there were identical to the odds everywhere else — a lottery draw has no memory of which shop printed the slip. What changed was the availability of a winning instance associated with that location. A concrete, local, recent example made winning feel retrievable, and people bought.
The authors also found the response was larger for bigger jackpots and larger in more economically disadvantaged areas. That second point connects to lottery play and financial stress.
It is worth being precise, because the bias is often stretched to cover everything.
It does not explain everything about lottery demand. People also buy for the anticipation itself, which is a separate and defensible motivation — see the cost of hope. Availability explains a distorted probability estimate, not the whole decision.
It is not a claim that players think the odds are good. Ask most players directly and they will tell you the odds are terrible. The bias operates on the felt sense of possibility, not the stated number. Those two can coexist comfortably in the same person.
It is not unique to lotteries. The same mechanism makes people overestimate deaths from shark attacks and plane crashes and underestimate deaths from ordinary illness. Lotteries are simply an unusually clean case, because the publicity filter is so extreme and the true probability is so precisely known.
Availability distorts the downside story too, and in the same direction — towards whatever gets published.
The "lottery curse" genre exists because ruined winners make better copy than winners who paid off the mortgage and kept their job. The result is a folk belief that winning typically ends badly, which the best long-run research does not support. That is the same sampling error pointed the other way, and it is unpacked in the lottery curse, fact-checked and the 70% broke myth.
Availability is not a story about players being credulous. It is a story about what gets into the sample.
There is no way to un-know a vivid example. What does work is replacing the retrieved instance with a computed one:
The odds visualiser and cost per chance exist for exactly this substitution — a computed figure in place of a remembered one.
None of that tells you whether to buy a ticket. It tells you what you are buying, which is a different and more useful thing.
Last verified: 2026-08-29