Anecdotes
Ron Graham
An anecdote is a story. You'll use that story to reinforce some point you're making in an argument. Sometimes, you'll find that a story makes the point more understandable to your audience. It can make the abstract seem real, the complex simple, and the far off near.

How it does that exactly I'm not sure. Maybe we can't say. I know I will choose an anecdote because it made an impression on ME. On the other hand, the strength of the anecdote may lay in the storyteller's ability to relate it to the audience more than in its content alone.

Anecdotal evidence is a bit risky in arguments: you run the risk of not having sufficient evidence if you defend your point on the basis of anecdotes alone, or if those you use aren't representative enough. There are related risks as well:

  • You repeat a story you heard elsewhere and you get the details wrong.
  • You don't "change the names to protect the innocent."
  • You consider the teaching of your parable to be so much more important than its setting that you give no thought to the setting at all.
  • You tell a story you like because you like it, instead of telling it because it brings home a point you're trying to make.

Murphy's Law insists that there'll always be someone in the audience to catch you. Ramage, Bean and Johnson, in Writing Arguments, give the following rules of thumb for using anecdotes in argument:

  • Distinguish facts and data from inference and opinion. [To me, this doesn't say anything about the truth of the anecdote itself; only about separating it as evidence from more "factual" source material.]
  • Use up-to-date info for up-to-date subjects. [But what's "up-to-date" can vary wildly from one subject to the next.]
  • Avoid bias by using "representative" examples. [No recommendations given on how to seek out such examples.]
  • Avoid "hasty generalization" - abstraction from a small number of examples.
  • Use trusted sources. [To me, this doesn't say anything about whether we can use a second-hand anecdote, provided we heard it from someone we trust.]

The Scientific View

When scientists talk about anecdotes or anecdotal evidence, they don't try very hard to hide their scorn:

Anecdotes are also entirely too susceptible to being unverifiable.
-- http://www.cs.colorado.edu/~lindsay/skeptic/ ...anecdotal.html

The argument draws a conclusion from cases specifically chosen to support the argument, while ignoring cases that might tend to undermine the argument.
-- http://www.cuyamaca.net/bruce.thompson/Fallacies/ ...anecdotal.asp

When dealing with statistical data, no matter how strong the evidence is for one conclusion, there are always a few examples that do not fit the rule. The fallacy of anecdotal evidence is when one uses these "exceptions to the rule" to prove their point.
-- http://www.santarosa.edu/~dpeterso/permanenthtml/ ...propaganda/prop_anecdotal.htm

Scientists rightly reject the use of anecdotal evidence in deriving theory, largely due to its lack of logical rigor or consistency, but also because the "facts" and events which comprise anecdotal accounts are dubious, fleeting, and often preclude replication.
-- http://www-pam.usc.edu/volume3/v3i1a2s1.html

Here's what bothers scientists like these:

  1. Opposition. Scientists experience opposition to theory and experiment based on anecdotal evidence. Their frustration lies in the fact that anecdotal evidence will convince people at times, despite the apparent strength of scientific data.

    This leads them to fall on their own fallacies: blame the audience for accepting the weaker anecdotal evidence; blame opponents for presenting them.

    Neither of these strategies is likely to strengthen the "scientific view" in the eyes of an otherwise neutral audience. A better strategy would involve

    • communicating scientific data more clearly, including the uses of money in future research
    • approaching this communication less combatively -- why does it have to be "the scientific view" vs. "anecdotal evidence" in the first place? why aren't there solutions where everybody wins?

  2. Public Belief in the Unmeasurable. Skeptics will cite typical fountains of anecdotal evidence as reasons to avoid anecdotes as sources:
    • scientific creationism (it may not be true, but we have good citizens in our society who believe in a Creator, and who pay taxes to support scientific research)
    • UFOs (they may not exist, but we have SETI and a human desire to explore as far away as we can)
    • holistic medicine (it may not work based on what we know, but part of what heals us is not measurable or verifiable anyway)
    They will say that millions of people believing in those ideas does not make the ideas true. Still, when millions of people believe these ideas, that means viewpoints coming from science have not been convincing to a larger audience. Once again, who's to blame for that? It's not the existence of anecdotes that we should blame.

  3. A Limited Scientific View of Anecdotes. So anecdotal evidence is of little use in deriving theory. As if that were its only use. Scientists actually use anecdotal evidence themselves, even if they won't admit it, in the following ways:
    • deciding how and to whom to apply for research grants
    • deciding directions for new and unstarted research
    • deciding what questions to ask human subjects in gathering empirical data
    • deciding what and when to publish
    The quotes above suggest that when one holds a "scientific view," that leads to the assumption that anecdotal evidence can only be used one way: to oppose the "scientific view." Pardon me for saying so, but that's a really weak assumption. We can see readily that anecdotal evidence isn't always used as an "exception to the rule."

The Difference Between Engineers and Scientists

Scientists tell us a person's word is of little use but they want us to believe their word.
-- http://www.geocities.com/freee80/anec33.html

  1. "What-if" Games. Engineers are trained to play "what-if" games. Those scenarios are not true by definition, and may not be representative. Still, they often direct our strategy -- either because of the chance that they might become true or because of the penalties we pay if they do. You can't use a double-blind test procedure to analyze the causes of structural failure. You have one data point, and anecdotes from all observers.

  2. Knowing Who's Interested. The "scientific view" as given above suggests that theory can only be brought forward by those who are disinterested. Unfortunately, that describes nobody. Everybody has an interest in something, even scientists. The sooner we all see it, the sooner we can move forward.

Apparent weaknesses of anecdotal evidence :-) :-) include

  • A statistician can have his head in an oven and his feet in ice, and he will say that on the average he feels fine. (Unknown)
  • Do you hear about the statistician who drowned in a lake averaging only 2 inches in depth? (Unknown)
-- http://www.bamaed.ua.edu/~rlomax/LOMAX/HUMOR.HTM

What You Can Do

  • If the lesson from an anecdote is critical to its success, you should try to make sure your audience gets it.
  • Don't represent an anecdote as a "true story" unless you can demonstrate it. (Either through it being your own experience or through someone else vouching for its authenticity.) A little embellishment for drama or comedy's sake may not hurt much.
  • Anecdotal evidence must be both true and representative for its use to be acceptable in establishing theory.
  • Remember that your anecdote may be repeated. If it is, Murphy's law insists that they'll screw it up. Will that change your approach?
[Table of Contents] [Previous] [Next]