Availability heuristic
If examples come to mind easily, we assume they're common.
What it means
The availability heuristic is the mental shortcut by which people judge how likely or frequent something is by how readily instances of it come to mind. It substitutes an easy question - 'how quickly do examples surface?' - for the hard one - 'how common is this, really?' - so anything that boosts retrieval can inflate perceived probability. Vivid, recent, emotionally charged or heavily reported events are recalled more easily and so feel more probable than they are, which is why coverage warps risk perception. The shortcut is usually serviceable, because ease of recall does broadly track real frequency; it misfires when memory is fed by attention rather than incidence. A long-standing refinement - that felt ease of retrieval matters more than the number of examples recalled - has failed large preregistered replications and is better treated as contested. It matters most in risk communication, insurance, public health and security.
Why the shortcut usually works
Availability is not simply a defect. In most environments, how easily an example comes to mind tracks how often it occurs: common things are encountered more and retrieved faster. Retrieval fluency is a cheap proxy for frequency, and leaning on it saves the cost of keeping running tallies. Tversky and Kahneman framed it this way themselves — availability correlates with real frequency, and bias appears only where that correlation breaks, where something inflates retrieval independently of how often the event happens. Media coverage, personal involvement, recency and emotional charge all do exactly that. The heuristic fails not because memory is unreliable but because the information environment samples the world non-randomly, oversupplying the dramatic. Expect errors wherever coverage and incidence come apart.
What the evidence shows
The evidence splits into two claims that often get conflated. The broad claim — that retrievability shapes judgments of frequency and risk — has wide support. The narrow claim rests on the few/many paradigm of Schwarz and colleagues, in which listing twelve examples of your own assertiveness should leave you feeling less assertive than listing six. That paradigm is in serious trouble. Weingarten and Hutchinson's meta-analysis pooled hundreds of effects and reported a small-to-medium average of r = .25, but the literature carries heavy publication bias, and several of the strongest supporting studies come from Lawrence Sanna, who has eight retractions following findings of research misconduct. Two large tests found nothing: a demographically representative national survey experiment by Yeager and colleagues, and Groncki and colleagues' preregistered direct replication of the assertiveness study (N = 661).
Where the classic demonstrations break down
The letter task is the cautionary tale. Tversky and Kahneman asked whether more English words begin with K, L, N, R or V or carry them in third position; most people said first, and the error was read as availability at work. Sedlmeier, Hertwig and Gigerenzer later measured the two things availability could mean — how easily instances come to mind, and how many — independently of the frequency judgments themselves. Neither predicted what people estimated; actual letter frequencies did. The demonstration that helped launch the concept turned out not to need it. This is the recurring complaint: availability is usually inferred from the very judgment it is invoked to explain, which makes it hard to falsify. Measure ease separately and the mechanism often thins out.
Using it in practice
Assume your audience's sense of frequency is built from what they have seen, not from your table: give base rates as concrete frequencies, and expect one vivid counter-anecdote to outweigh them. The tell is a gap between what people fear and what base rates say, moving with recent salience — flood cover bought after a flood and lapsed as the memory fades, security budgets aimed at the attack that made the news rather than the misconfiguration behind most breaches. If you are the one deciding, ask what would change your estimate other than a story you can recall, then find the denominator. Do not build interventions on the few/many trick: asking people to list many reasons, hoping the difficulty backfires, rests on the least replicable part of this literature.
Examples
After seeing news of a plane crash, people overestimate the risk of flying — even though it is statistically far safer than driving.
After a burglary two streets away, neighbours rush to buy alarms; the vivid, close-to-home story crowds out the flat statistic that break-ins on their road have been falling for years.
Ask whether more English words start with the letter r or have r in third place; most people say the first, because words beginning with r are simply easier to fish out of memory.
A hiring manager rejects a candidate from a university that produced one bad hire two years ago; the single memorable failure stands in for a base rate she has never looked up.
A doctor who caught a rare diagnosis last month orders the workup for it again on the next patient with vague symptoms, while the far commoner differential goes unchecked; the memorable case has quietly replaced the prevalence figure.
First described in Tversky & Kahneman (1973).
Key references
- Groncki, R., Beaudry, J. L., & Sauer, J. D. (2021). Investigating the ease-of-retrieval effect in an eyewitness context. Memory, 29(2), 234-254. doi.org/10.1080/09658211.2021.1882502
- Yeager, D. S., Krosnick, J. A., Visser, P. S., Holbrook, A. L., & Tahk, A. M. (2019). Moderation of classic social psychological effects by demographics in the U.S. adult population: New opportunities for theoretical advancement. Journal of Personality and Social Psychology, 117(6), e84-e99. doi.org/10.1037/pspa0000171
- Weingarten, E., & Hutchinson, J. W. (2018). Does ease mediate the ease-of-retrieval effect? A meta-analysis. Psychological Bulletin, 144(3), 227-283. doi.org/10.1037/bul0000122
- Sedlmeier, P., Hertwig, R., & Gigerenzer, G. (1998). Are judgments of the positional frequencies of letters systematically biased due to availability? Journal of Experimental Psychology: Learning, Memory, and Cognition, 24(3), 754-770. doi.org/10.1037/0278-7393.24.3.754
- Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka, H., & Simons, A. (1991). Ease of retrieval as information: Another look at the availability heuristic. Journal of Personality and Social Psychology, 61(2), 195-202. doi.org/10.1037/0022-3514.61.2.195
- Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207-232. doi.org/10.1016/0010-0285(73)90033-9