16 Anecdotal Evidence
(Modified on 2026-08-11)
1,100 words
The human mind is built to think in terms of narratives, of sequences of events with an internal logic and dynamic that appear as a unified whole.
— Schank & Abelson, (1995, 51)
In 1966, Peter Wason (Wason 1966) published an article showing a peculiar aspect of human reasoning. He devised the following task. Various cards are presented: each card has a number written on one side and a letter written on the other. There is a rule: any card having a vowel on one side has an even number on the other. Now suppose you are looking at four cards with the following visible characters:
- A
- B
- 4
- 7
Which of the four cards should you turn over in order to check whether the rule has been followed? (It’s important to try this task before continuing to read.)
The correct answer is…
Turn over A and 7. You need to turn over A to make sure there is an even number on the back. You need to turn over 7 in order to ensure that it doesn’t have a vowel on the back.Did you get it right? Maybe you did smarty pants, but a surprisingly large number of people don’t!
Now consider the following variant on the task. In this case, the cards now contain different information. On one side of the card is the age of a person at a bar. The other side of the card identifies what it is that the person is drinking. Once again, there is a rule: anyone drinking alcohol must be 21 years of age or older. Your task is to identify which cards to turn over in order to check whether the rule is followed:
- beer
- cola
- 16
- 25
Most people find this problem trivial. You need to check the age of the beer-drinker, and you need to check what the 16-year-old is drinking. What’s important is that the two problems are logically identical, but people have a much easier time understanding (and correctly solving) the problem that is story-based rather than abstract.
Stories
Social psychologists Roger Schank and Robert Abelson (Schank and Abelson 1977, 1995) have argued that stories and storytelling are fundamental to human knowledge. Narratives involving individual people or events are easier to understand than abstract principles or large numbers. Research persistently shows that people are not interested in numbers—We are interested primarily in other people and in hearing about the experiences of other people. When people gather together, they tend to relay stories; these stories include not only first-hand experiences, but also the recounting of stories we have heard of other people. People find People magazine much more compelling to read than reports from the U.S. Bureau of Labor Statistics. There are compelling human concerns behind labor statistics, but because these numbers are faceless, people don’t find them interesting.
Charities have learned this lesson. For example, many charities (Oxfam, Save The Children, CARE, UNICEF, Mercy Corps, etc.) are involved in trying to alleviate hunger and starvation. In their advertising, these charities have learned the importance of storytelling. A charity appeal that notes that one million people are currently experiencing starvation in Somalia has surprisingly little effect. Video footage of a single starving child (who is given a name) is far more effective than video footage showing a sea of starving people in refugee camps.
In addition to being easier to understand, human-oriented narratives are more memorable. We remember anecdotes about people better than statistical facts (Schank and Abelson 1977). In fact, people’s memories of essential facts are indexed in the brain around stories (Schank and Abelson 1977, 1995). Facts that are remembered are attached to stories; Other facts only enter short-term memory.
The Utility of Anecdotes
There are two places where anecdotes have a role in research. First, anecdotes are a legitimate part of exploratory research. Anecdotes can provide useful starting points for formulating theories which can then be tested using formal empirical methods. Second, anecdotes provide memorable illustrations or examples—often, an anecdote is the simplest and clearest way to explain and motivate a complex idea. Because narratives can be quite memorable, we should always make use of narratives in our teaching. Let the research speak, and then cement the lesson by adding an appropriate anecdote. Journalists love to do this. Even when a journalist is reporting research they will typically try to personalize the article by discussing an individual case, or a specific anecdote.
Of course, good scientific writing should only use anecdotes to illustrate a point or trend that is consistent with empirical research!
The Dangers of Anecdotes
There are at least three problems with anecdotes. First, anecdotes are mentally too compelling. A good anecdote will trump reasoned argument. Anecdotes are easily recalled when they go against expectations. For example, some people have died in automobile accidents because they got tangled in a seat belt. However, seatbelts save many more lives than they end. One or two anecdotes can have an enormous impact on people’s willingness to wear seatbelts. Unfortunately, anecdotes commonly trump research statistics.
The second problem with anecdotes is that they are not systematically or randomly sampled and so are rarely representative. A person might tell you that they ate left-over roast duck for breakfast. They won’t tell you that they ate cereal for breakfast. The first story is more interesting than the second. If we rely mostly on anecdotes, we may wrongly conclude that roast duck is a more common breakfast food than cereal. In addition, anecdotes are often “volunteered” rather than “elicited.” Often, people have an underlying motive or goal in telling the story. That motive can often be to “correct” what the story-teller regards as a questionable received wisdom.
A third problem is that anecdotes tend to attract similar anecdotes. In ordinary conversation, for example, if a person relays a positive anecdote about Jill, the story is likely to evoke further positive anecdotes about Jill. Conversely, if someone says something negative about Jill, the comment is likely to evoke further negative anecdotes about Jill. This bandwagon effect is evident on the Internet where stories on some subject tend to attract similar stories on the same subject. This bandwagon effect exaggerates a given view and makes it seem more commonplace than is the case. The tendency for similar stories to “stick together” means that anecdotes tend to amplify sampling bias.
In light of the sampling problems related to anecdotes, haphazard collections of anecdotes should not be used as data for testing a theory or hypothesis.
Mantras
Our next mantra reminds us to be wary of anecdotal information:
Anecdotes are, by their very nature, unrepresentative samples of reality. Anecdotes give us ideas and motivate us, but their value as evidence in an argument is basically nil.