17.2: The Availability Heuristic

• Do more people in the U.S. die from murder or suicide?
• Are there more English words that begin with the letter ‘r’ or words that have ‘r’ as their third letter?
• Are there more famous people from Oklahoma, or from Kansas (a state with roughly the same population)?

Many of our inferences lead to conclusions about the relative frequency or proportion of some feature in a population. Are there more Bs than Cs; e.g., are there more Jeeps (Bs) or Fords (Cs)? Is an A more likely to be a B or a C; e.g., is an OU student (an A) more likely to be male (B) or female (C)? Since we can rarely check the entire population, we must base our inference on a sample. In everyday life, we rarely have the time or resources to gather a sample in the way scientists do, so we often “do the sampling in our heads.” We try to remember cases that we know about, or to imagine cases that seem relevant.

Suppose that you want to know whether there are more Fords or Jeeps in use today. You will probably rely on the sample of vehicles you can recall. You try to think about the various makes of vehicles you have observed. Obviously, this method is somewhat vague and impressionistic, since you probably don’t remember more than a handful of specific Fords (like Wilbur’s Fiesta) or Jeeps (like Aunt Ethel’s). But at least you do know that you have seen a lot more Fords than Jeeps. You have a generalized memory about this, even though you don’t recall many specific vehicles of either kind.

In many cases, including this one, this method works. You remember seeing a lot of Fords, and you remember that you haven’t seen many Jeeps. The reason why you remember more Fords is that there are more Fords. Fords are easily available to recall largely because there are so many of them. Fords are more available in your memory precisely because you have seen a lot more Fords.

When we need to judge the relative frequency or probability of something, we are often influenced by the availability or accessibility of those kinds of things in thought. The “sample in our heads” consists of the cases we remember and, to some extent, the cases we can easily imagine. We use the availability heuristic when we base our estimates of frequencies or probabilities on those cases that most readily come to mind, on those that are most available in memory and imagination. This heuristic inclines us to assume that things that are easier to remember or imagine are more likely to occur.

Why Things Are Available

Often, we remember certain things because they really do occur frequently, and when this is the case the available sample in our head will often be a good one (or at least good enough for the rough-and-ready inferences of everyday life). When availability is highly correlated with objective frequencies or probabilities, as it often is, it is a useful guide. Are there more words beginning with the letter ‘r’ or with the letter ‘z’? Words beginning with ‘r’ are more available than those beginning with ‘z’ precisely because they are much more common. Here the heuristic works very well, leading us to the correct conclusion.

But the judged frequency and the true frequency of something may be very different. Things may be available in memory or imagination for reasons having little to do with their frequency or probability. In these cases, the availability heuristic leads us to rely on a small sample (the cases we easily remember) and one that may be biased in various ways (the cases we happen to have encountered and manage to recall). For example, things we are familiar with will be available. And since memory generally becomes less vivid and accessible over time, more recent experiences and events are more likely to be available than those that occurred longer ago.

We begin this section by asking whether there are more English words that begin with the letter ‘r’ or words that have ‘r’ as their third letter. Many of us think (at least when we aren’t primed to think there is a trick involved) that more words begin with ‘r’, though in fact this is false. So why do we think this? But it is much easier to think of, to generate, letters that begin with ‘r’ than to think of words in which ‘r’ comes third. Words beginning with ‘r’ are more available to our minds, and this greater availability leads many of us to infer that more words begin with ‘r’.

Examples of Availability

Around 1 in 30 million Americans are killed by terrorists in a year; significantly more – around 12 in 10,000 – are killed in automobile accidents. But the cases where Americans are killed by terrorists are likely to make the news, and for obvious reasons, they stand out in memory. Now suppose we are asked to estimate the number of people killed by terrorists. The sample that is readily available, the one that comes naturally to mind, can easily lead us to overestimate the threat of terrorism today. People also overestimate the rate of homicides and other stories that make the news. This is one reason why most of us suppose that there are more murders than suicides, though statistics show that there are many more suicides than murders. By contrast, the frequency of things that are not so wellpublicized, like death from diabetes, is usually radically underestimated.

On the other hand, unless we know of several people who have been killed in accidents, examples of such deaths may not be so salient in memory. Such deaths are common enough that they aren’t likely to be reported by the media unless the person killed is well-known. Examples of such deaths are not particularly available, so we may radically underestimate their frequency. To take a related example, fires make the news more often than drownings, and they may be more dramatic in various ways. So, it is not surprising that many people think that death by fire is more likely than drowning, even though the reverse is the case. The good news here is that we may underestimate the amount of helping and kindness there is, since such reports rarely make the news.

Things that occur reasonably often (e.g., fatal automobile accidents) are rarely reported and are easily forgotten, whereas events that are rare but dramatic (e.g., terrorism) make for good news, and stand out in memory. In such cases, frequency is not closely related to availability in memory, and the use of the availability heuristic will lead us astray. For example, 100 times as many people die from disease as are victims of homicide, but newspapers carry three times as many articles about murders.

Media Effects

Here are some further examples. The media and advertisers often tell us about people who strike it rich by winning a state lottery. This can make such cases more available to us in thought, leading us to overestimate the probability of winning a lottery (we all know the probability of winning is low, but it is much lower than many people suppose). Again, many more people die of diabetes each year than in accidents involving fireworks. The latter get more press, however, and many people think more deaths really are caused by such accidents.

Partly because they are reported and partly because of the success of the movie Jaws, shark attacks seem vivid, easy to imagine, and easy to remember. In fact, they are extremely rare, and you are much more likely to be killed in many other (less dramatic) ways. This is a fine opportunity to put to work the research skills you learned in past chapters. How common are shark attacks? How often are those fatal? Compare those results to human deaths caused by a much more mundane animal, like a pig or a horse.

Surprising Events are Memorable

There are many other cases in which unlikely events may be particularly available. A few people are very likely to recover from an illness that is fatal to most people who contract it. Since the tiny minority who recover will probably be under some sort of treatment (call it treatment X), and since miracle recoveries make for good news, we may hear about the miracle cure due to treatment X. This will be available to memory, and so we overestimate the probability that X can be effective in curing the disease.

Indeed, in all but the most extreme conditions, almost any miracle cure or quick fix (for losing weight, kicking cigarettes, quitting gambling, etc.) will seem to work for some people (perhaps because of a placebo effect, perhaps thorough sheer coincidence). In such cases, we may hear an endorsement, perhaps in an infomercial, from people who sincerely believe that they have benefited from the treatment. Such testimony can be very compelling, and it is often easily available in memory. In such cases, the availability heuristic can lead us to spend a lot of money on quick fixes that don’t fix anything at all (except the financial condition of the person selling them).

Salience

One or two examples may be so vivid or salient that they lead us to discount much better evidence. Cases “close to home” can be especially compelling. Your Aunt Ethel had a Toyota Prius that was a real piece of junk (though ‘junk’ wasn’t exactly the word she used). This single case is likely to loom very large in your memory. Then you learn that some consumer group you trust (e.g., Consumer Reports) did a survey of thousands of car owners and found the Prius to be more reliable than most other makes. If you are like most people, the one case close to home will stand out more (be more salient); it will be more memorable. Hence, it will have a much great influence on what you buy than the careful and detailed study by the consumer group.

Our Everyday Samples are Often Biased Many of the samples we encounter as we go about our lives are biased. Our age, gender, race, job, friends, interests, and where we live all mean that we will be exposed more to some things than others. If you live in Boston, Massachusetts you will be exposed to a different range of things than if you live in Belleville, Kansas. In many cases this is obvious, and it’s relatively easy to discount for it. You realize that it’s not safe to predict the general public’s tastes in music based on the musical tastes of the people you know; they don’t provide a representative sample. But in other cases, the biased nature of the samples we normally encounter may be less obvious. It can be tempting, for example, to form beliefs about the general public’s political views on various issues based on the views we hear expressed most often. But these may not be representative of people’s views in general.

Problems with Availability

In earlier chapters, we encountered several phenomena which suggest that the availability of things in memory is not always a good guide to how things really are. Perceptual set will incline us to notice certain things while overlooking others, thus, influencing what makes it into memory in the first place. Then elaboration in memory can affect what we remember, as can the context in which we remember it. Further biases may enter because of primacy, recency, or halo effects. In short, the sample in our heads is often based on limited experience, and it can then be further distorted in a variety of ways.

Prejudices and stereotypes are an especially insidious example of this. If you have a negative stereotype of members of a certain group, you are likely to notice some things (e.g., cases where a member of the group fails) than others (e.g., cases where a member succeeds). You will also be more likely to remember such cases and find it easier to imagine them. When you then must predict how typical members of that group will do, the negative cases will be more available than the positive ones, and you are likely to conclude that they will probably do poorly. We will return to this topic in a later chapter.

We can’t abandon the availability heuristic. It is deeply ingrained in the way we reason, and it often works very well. But we need to be aware of the ways it can lead to fallacious reasoning. We need to realize that the samples in our heads (and in the heads of others, often even those in the heads of experts) are biased in one way or another.