(I'm doing a real post today because yesterday was substandard)
Someone trained in formal logic is going to point out that the use of anecdotal evidence is not a fallacy per se, and that is correct. However, it lends itself so readily to the construction of other fallacies (specifically of the post hoc, unwarranted generalization, or regressive variety) that I feel justified in pointing the spotlights at it.
Anecdotal evidence is simply using an example or anecdote – "I know a person who smoked 2 packs per day and lived to be 90" – to support an argument without acknowledging the extent to which it can be generalized. Often, as in the example I just used, anecdotal evidence is cited to contradict something supported by all available evidence. There's no logically sound way to argue that smoking isn't bad for you, so bring up an (unverifiable) anecdote of someone who beat the overwhelming odds. Like its friends heresay and conjecture, anecdotal evidence strikes at the heart of how good arguments are constructed and supported. If the evidence for an argument is unverifiable, unfalsifiable, or statistically improbable then the argument, regardless of whether or not it has merit, is invalid. I would be equally incorrect to argue "I know someone who smoked and died of lung cancer, so smoking is dangerous." Correct conclusion, but irrelevant to the "evidence" cited.
Talk radio and right-wing columnists are an excellent source of anecdotal evidence (not to mention heresay, conjecture, and sweeping generalizations). Deeply serious and respected NY Times columnist Bill Kristol, please step up and support my claim! (from "President Huckabee?"):
At a Friday night event at New England College in Henniker, (Huckabee) played bass with a local rock band, Mama Kicks. One secular New Hampshire Republican’s reaction: "Gee, he’s not some kind of crazy Christian. He’s an ordinary American."
Anecdotal, and totally unverifiable. How much of the population feels this way? Is it this person or millions of Republicans? To the deeply serious Bill Kristol, one guy's offhand comment is proof of Huckabee's "widespread appeal" to non-Bible thumping Arkansans. Another gem comes from Rush Limbaugh's reaction to the New Hampshire primary, which he blames on his pet theory of "Out of state buses" (i.e., candidates bus in voters from other states to illegally participate in the primary). Rush supports his theory based on callers (heresay) who "saw a lot more people than usual" at their town polling places (anecdotal) and concluded (conjecture) that this is because voters were imported from other states. That's the complete Trifecta of bullshit non-evidence categories in one argument. Thanks Rush!
Anecdotal evidence…isn't. That phrase is a misnomer. I should emphasize that evidence for valid arguments can be anecdotal; this fallacy doesn't mean that your argument is wrong, but rather that your evidence doesn't prove it one way or the other. Maybe hypnosis and miracle dietary supplements can cause weight loss. If that statement is true, it's not because you know someone who tried it and lost 30 pounds.