Availability Heuristic
Also known as: availability, avh
Judging how likely something is by how easily examples come to mind.
Share: also:
In plain terms
When you estimate how common something is, your brain rarely counts. It checks how quickly examples surface, and treats the speed of recall as a stand-in for frequency.
Most of the time this works. Common things really are easier to remember, which is why the shortcut survives. Tversky and Kahneman named it in 1973, and the interesting part is not that people use it but where it fails.
It fails whenever recall is driven by something other than frequency: how vivid an event was, how recently it happened, how much coverage it got, or whether it happened to you. That is why plane crashes and shark attacks feel like live risks while the boring causes of death, the ones that fill the actuarial tables, feel abstract.
Why it matters
Coverage is not frequency, and the relationship often runs backwards. News selects for the rare and the dramatic. Rare dramatic events therefore become the easiest to recall, which makes them feel routine, which is the opposite of what their coverage actually indicates.
Personal experience inflates the same way. One burglary on your street moves your sense of neighborhood crime more than a decade of published statistics, because the statistics don't come with a memory attached.
The practical cost is misallocated worry and misallocated money. People buy insurance against the memorable and skip the mundane, in their own lives and in their organizations' risk registers.
Canonical example
A: "I'm driving. There were two crashes in the news this month."
B: "Out of how many flights?"
A: "No idea. But I can name both crashes."
Being able to name them is the whole problem. Crashes are reported exhaustively, and they're reported exhaustively because they are rare enough to be news. The ease of recall that makes flying feel dangerous is produced by the same rarity that makes it safe.
Two examples with no denominator is not a rate. Whatever A concluded, the two headlines did not support it.
Counter-example (not the availability heuristic)
A: "I'm avoiding that particular carrier. Three of their aircraft were grounded this year and the regulator has an open investigation."
This is also a judgment built from recalled examples. It isn't the availability heuristic.
The examples here are the relevant evidence, drawn from a small and specific population rather than from whatever happened to be broadcast. Three groundings out of one airline's fleet is a meaningful proportion. Two crashes out of tens of millions of flights is not. Recall is a fine place to start; the bias lies in using recall as the measurement.
The line: are your examples a sample of the thing you're measuring, or a sample of what got your attention?
How to fix it
If you've been linked here, the fix is one question asked of your own examples: out of how many? A count of vivid cases means nothing until you know the size of the pool it came from. Then ask a second question: why do I know about these particular cases? If the answer is that they were covered, awarded, gone viral, or personally survived, your sample was selected by something other than frequency, and it will skew high in exactly the direction that feels most convincing.
If you're on the receiving end, don't open with the correct statistic. Ask for the denominator and let the other person notice they don't have it. "How many flights were there?" does more work than a table, because the gap is more persuasive when someone finds it themselves. Then supply the base rate.