What Is Availability? Judging Frequency by How Easily Examples Come to Mind

Judging frequency directly would require counting, which nobody does, so the mind substitutes a related question that is easy to answer, namely how easily examples can be retrieved. That substitution is usually sensible, since common things are encountered more often and are therefore easier to recall, so retrieval fluency genuinely correlates with frequency. The failure occurs when something other than frequency makes an event memorable, and several things do. Vividness, emotional impact, recency and personal involvement all increase how easily an event is recalled without increasing how often it happens. Repetition in media does the same. When any of those operate, the estimate is inflated in proportion, and the person has no sense that anything has gone wrong, because the judgement feels like a memory of how common something is rather than an inference from retrieval speed.

The pattern of misjudgement is predictable enough to be listed:

• •Dramatic causes of death are overestimated and mundane ones underestimated, which has been measured repeatedly
• •Risks that are reported are judged more common than risks that are not, regardless of actual rates
• •Recent events dominate, so risk estimates rise after an incident and decay over months
• •Personal experience outweighs statistics, so one known case changes a judgement more than a large study
• •Rare and frightening events attract attention and resources disproportionately
• •Common and slow harms receive less concern than their numbers warrant

The consequences are visible in individual decisions and in public policy. People avoid flying after an air disaster and drive instead, which is measurably more dangerous, and the increase in road deaths following such events has been estimated and exceeds the deaths in the original incident. Spending on security against rare threats is far higher per life saved than spending on common causes of harm, which is defensible on grounds other than efficiency and is rarely defended explicitly. Insurance purchase follows recent events rather than actual risk, with flood cover bought after floods and dropped as the memory fades. Medical decisions are influenced by vivid anecdotes about side effects more than by the underlying rates. In each case the error is not carelessness, since the people involved are responding to a genuine feature of their own memory, and correcting it requires deliberately consulting numbers.

This is one of a family of mental shortcuts identified in the same research programme and the family resemblance is instructive. Representativeness judges probability by how closely a case matches a stereotype, which produces neglect of how common the categories actually are and generates the well-known error of finding a detailed description more probable than a general one. Anchoring pulls estimates towards a number encountered beforehand. Each works by substituting an easy question for a hard one, which is the unifying mechanism, and each is usually adequate, which is why they persist. The research emphasised that these are not defects introduced by carelessness but the normal operation of a system that trades accuracy for speed, and that the errors are predictable precisely because the substitutions are systematic. That predictability is what makes them correctable in situations where accuracy matters more than speed.

The interaction with news coverage produces a self-reinforcing cycle worth stating explicitly. Unusual events are newsworthy precisely because they are unusual, which means coverage is systematically biased against frequency, and a viewer who calibrates their sense of the world to what they see will be wrong in a consistent direction. Coverage then follows attention, so the events people are already worried about receive further coverage, which increases availability further. Online systems amplify this by selecting for engagement, and engagement correlates with the same features that make events memorable. The correction available to an individual is modest, consisting mainly of checking base rates before forming a view, noticing when a judgement rests on vivid examples rather than counts, and treating the absence of coverage as uninformative rather than as evidence that nothing is happening.

Frequency is judged by how easily examples come to mind, which correlates with frequency until something else makes an event memorable. Vividness, recency, emotional impact and repetition in media all inflate the estimate without the person noticing. Avoiding flights after a disaster and driving instead has caused more deaths than the incidents themselves. News covers events because they are unusual, which biases the calibration systematically.