Availability Bias: A 6-Question Evidence Check Before You Decide

How often does it happen? When you ask yourself that question about a risk, a market or a person, where does the answer come from? For most of us it comes from whatever examples arrive first. If they arrive quickly, the thing feels common. If we have to strain, it feels rare.

That shortcut has a name: the availability heuristic. Availability bias is what happens when the shortcut misleads us, when how easily something comes to mind stands in for how often it actually happens.

In 1973 Amos Tversky and Daniel Kahneman described the heuristic and tested it in a series of small experiments. In one, people were asked whether letters such as K and R appear more often as the first letter of English words or as the third. Most chose first position, even though these letters turn up more often in third position. Words that start with K are easy to search for in memory. Words with K in third place are not. In another study, people heard lists of names. When one sex was represented by more famous names, listeners judged that sex to be more numerous in the list, even when it was not.

Here is where it gets interesting. Is it the examples themselves that persuade us, or the feeling of finding them? In 1991 Norbert Schwarz and colleagues separated the two. They asked some people to recall six times they had behaved assertively and others to recall twelve. Twelve is hard. Those who listed twelve, and so had more examples in front of them, rated themselves as less assertive than those who listed six. The struggle itself became the evidence. These were students in controlled studies, not managers making real calls, so treat it as a clue rather than a rule. But it is a useful clue: the ease of remembering can carry more weight than what we remember.

Availability bias sits close to a few neighbours. Recency bias is about the latest event crowding out older ones. Availability is broader: an example can be old and still easy to recall because it was vivid, frightening or told often. WYSIATI is about building a confident story from whatever evidence is in front of you. Availability is about how that evidence got in front of you in the first place. And probability versus plausibility is about a coherent story outranking the base rate. Availability is often how the story arrived.

The fix is not to distrust memory. It is to treat what comes to mind as a starting sample, then check whether it is a fair one. Six questions for when a few examples are shaping your judgment:

1. What came to mind first? List the two or three examples behind your feeling. Name them, so you can see how few they are.
2. Why were these the easy ones to recall? Were they vivid, recent, emotional, personal or widely reported? Each of those makes a case memorable without making it common.
3. What would a fair sample look like? Pick a comparison group and look for a count. How many cases in total, and how many went each way?
4. What examples would I never hear about? The quiet cases rarely become stories: the project that simply worked, the flight that landed on time.
5. What is one example that points the other way? Find it deliberately, and write it down next to the first ones.
6. What is my judgment now, and when will I check it? Record the revised estimate and a review date in a decision journal.

Imagine a manager deciding whether to let a new team work remotely. Two cases come to mind at once: a remote hire who went quiet for weeks before quitting, and a news story about a company calling everyone back to the office. Why these two? The first was painful and personal. The second was everywhere that month. A fair sample: suppose the team records show eleven remote team members over four years. The quiet cases: suppose nine of them did steady, unremarkable work that nobody retold. The example pointing the other way: one remote designer led the best launch of last year. The revised judgment: in this imagined case, remote work has mostly worked, and one risk worth testing is a new starter with too few check-ins. So the manager allows it, with weekly one-to-ones for the first quarter, and puts a review date in the journal for three months out. Nothing in this example is a forecast. It shows the move: replacing two loud cases with the full count.

But hold on. Sometimes the vivid example really is typical, and sometimes one case is enough to change your mind, such as a serious safety failure. The six questions do not dismiss memorable evidence. They ask you to say what it represents before you let it decide. For a wider scan of the traps that travel with this one, the cognitive bias checklist is a good companion.

This is where ClearMind fits, quietly. It will not tell you how common something is or estimate a risk for you. It works as a mirror: its First Principles walkthrough asks what assumptions you are making and which of them are facts rather than conventions, then reflects back the patterns and blind spots in your answers. The evidence check above is yours to run alongside it. The judgment stays with you.

An old saying puts it well: a falling tree makes more noise than a growing forest. Memory hears the tree. A good decision goes looking for the forest.