Related meta questions:

I was spurred to ask because of this answer. Skilvvz seems to think that anecdotal evidence is not sufficient, but the author of the answer doesn't agree, so I thought I'd see what the consensus is.

I think a question which is more specific to the example I linked might be "Is anecdotal evidence sufficient for answers to _________ questions?", but I think that would overly localize the question. Plus, I'm not even sure how to define that scope (hence the blank).

We have discussed at length the quality of answers and the need for evidence here before, but AFAIK, not specifically anecdotal evidence -- I didn't find any matches on meta when I searched for anecdot[e|al]. If I didn't search thoroughly, and we indeed have, this question will hopefully just become a useful duplicate.

  • 7
    Q: Do ghosts exist? A: I've seen one once, sort of.
    – Sklivvz
    Apr 2 '11 at 23:12
  • 2
    Note: If you decide to downvote the linked answer after reading this question, at least be thorough and take a look at the other anecdotal answer to that question :) Apr 3 '11 at 3:59
  • 2
    Based on some of the arguments below, perhaps we need a better distinction between 'anecdotal', 'personal experience', and 'personally tested'.
    – oosterwal
    Apr 5 '11 at 19:04
  • 1
    In this case, does "anecdotal" mean "without supporting reference", or "referring to a single incident"? Apr 7 '11 at 2:53
  • @David Good point. I failed to make the distinction above, but I consider it to mean "referring to a single incident". Also, TBH, I consider claims referring to a single incident to lack supporting reference by default. And now that I re-read it, the answer I linked in the comments isn't anecdotal at all, but it is uncited. Apr 7 '11 at 5:42
  • @Jason Plank: However, there are questions here that can be answered with a backed-up anecdote. There was one that asked if it's possible to reach outer space, and one example is enough to show its possibility. Apr 7 '11 at 13:38
  • @David If the example is "Sputnik 1 went there", I'm with you, but "I've launched a model rocket into space and you'll have to take my word on it" is what I mean by "lacks supporting evidence". Apr 7 '11 at 15:31
  • @Jason Plank: Which sounds like you're basing it on citations and external evidence, not whether it's a single incident or not. Apr 7 '11 at 16:16
  • @David Hm, well, like I said earlier, I consider a claim involving a single incident to lack evidence by default. If there is evidence to back it up (e.g. there's loads of it for Sputnik), then yes, it's fine. IMO, a claim can lack supporting evidence without necessarily being anecdotal (consider "tap water causes cancer") so I think maybe the definition must include both conditions (referring to single incident and without supporting evidence). Apr 7 '11 at 16:34
  • Can anecdotal evidence not be verifiable? For example, the telekinesis answer below could be debunked by asking the answerer to document via a video. Similarly, any verifiable source on telekinesis would essentially be an anecdotal account, since it would be of the form:researched that person X who claimed to perform telekinesis failed to do so.
    – apoorv020
    May 5 '11 at 12:53

It actually depends. Not all anecdotes are scientifically or logically invalid. Sometimes, anecdotes are used, for example, to launch further study in an area. The famous Newton apple, for example, is an anecdote which helps people to understand a concept. The details of the story aren't entirely accurate, as most people hear or read about the story, but there is some truth to it and it is something that led to further study. So, if an anecdote is presented as an anecdote, but in such a manner that it can benefit the ongoing examination of a question, especially if that is all the evidence that we have, then it should be allowable to use an anecdote. I would say, though, that anytime an anecdote is used, that a statement should be made which clarifies how it is relevant to the topic and discussing the limitations.

Also, anecdotes can be used to debunk generalizations. Anecdotes that have repeatable conditions are especially good for this. For example, if a claim is made that everyone in my profession is stupid and I say that I am not, it may be an anecdote that I use to tell you why I think I am not stupid. However, you can test my intelligence to either disprove or confirm my claim.

The reason the anecdote given in the answer you linked to is not useful is because it is not the deepest examination we can have of the subject, which is testable; it doesn't open up a path to further examination and it isn't testing a generalization.

  • I agree with your point about testability, but I dunno. I am struggling to think of a question that is both not testable and yet still on-topic for the site, not subjective, etc. Apr 3 '11 at 3:57
  • 4
    Anecdotes are fine as useful, as anecdotes. They are not fine, as whole answers.
    – Sklivvz
    Apr 3 '11 at 10:29
  • There is a difference between anecdotes and anecdotes. A lot of people can try out a simple technique to heal a hiccup, but who can test near-death-experiences? Apr 4 '11 at 1:33
  • Anecdotal evidence is considered dubious support of a claim; it is accepted only in lieu of more solid evidence. This is true regardless of the veracity of individual claims. So, by its own definition, anecdotal evidence is not solid or scientific, but dubious. It can't even stand as a counterexample.
    – Sklivvz
    Dec 4 '12 at 23:54
  • @EbenezerSklivvze: If an original post cites a writer who claims that something never happens, would a poster's claim that he sees the event in question almost daily at a specific place, and that it's so commonplace there that it wouldn't merit particular notice but for the original writer's claims, be appropriate?
    – supercat
    Dec 24 '14 at 17:41


  • 11
    [Citation needed] ;-) Apr 6 '11 at 15:05
  • @KonradRudolph: Well played.
    – Borror0
    Apr 6 '11 at 15:14
  • 1
    Before such an answer can be useful, I think "anecdote" must be defined. I think it's clear that not all anecdotes are created equal. And the dictionary defines an anecdote as a "short story"--who decides what's short? When does "anecdotal evidence" become sufficiently long that it's no longer an anecdote? Some answers may be sufficiently answered with a single case--is that an anecdote?
    – Flimzy
    Dec 4 '12 at 23:25

An anecdote is not a single point of data. An anecdote is an unreliable account which never has the strength of data by definition:


  1. (of an account) Not necessarily true or reliable, because based on personal accounts rather than facts or research.
  2. Characterized by or fond of telling anecdotes.

Oxford Dictionaries

As such, it absolutely does not constitute good enough evidence for this site!

See also the definition of skepticism:

Scientific skeptics attempt to evaluate claims based on verifiability and falsifiability and discourage accepting claims on faith or anecdotal evidence.

Scientific skepticism

I don't see how someone claiming that something happens just by heresay is acceptable here, ever.

  • In the medical world, things like case studies/reports often count as anecdotes, yet they are useful to some extent, so the general/English dictionary definition isn't the end of the story.
    – Fizz
    Aug 28 '21 at 22:49
  • (Still this answer is better sourced / reasoned than the plain "no" proclamation from another, so I'm upvoting.)
    – Fizz
    Aug 28 '21 at 23:11

Consider this question: Can you unlock a car using a cell phone?

@fred correctly points out that this is something that could be tested out. Anecdotal evidence could show a positive for a specific case of car manufacturer, cell phone manufacturer, cell network provider, etc. However, anecdotal evidence of a negative outcome would not disprove the query.

For this question, anecdotal evidence that clearly lists all the test input parameters and outcomes of repeated tests should be considered sufficient evidence to support the claim.

  • 2
    I disagree. I don't see why we should trust the answerers. E.g. "Q: Can you unlock a car using a cell phone? A: Sure, I can do it on my Renault Twingo.". We can't accept this kind of "evidence".
    – Sklivvz
    Apr 5 '11 at 16:44
  • 1
    If you can do it, show it. Film it and put in on YouTube.
    – Borror0
    Apr 5 '11 at 18:49
  • 3
    If you cannot trust answerers, how can you be sure you can trust authors of scientfic papers? If "anecdote" is described in a way that anyone can use to reproduce it, I think it is a valid answer. Imagine a question like "does apple really fall down?". Would you object to have an experiment description being provided as an answer, and would you require peer reviewed scientific paper instead?
    – Suma
    Apr 6 '11 at 19:33
  • Borror0: Yeah let's just exclude everyone who doesn't have a video camera and a bunch of free time. That'll make this site useful. May 24 '11 at 20:34
  • 1
    @Brendan: If they don't have sufficient time/resources to answer the question properly, then they don't have sufficient time/resources to answer the question at all. That's fine; we won't hold it against someone for only having time to read, but we shouldn't accept half-done answers.
    – Oddthinking Mod
    Dec 5 '12 at 0:56

I surely can do telekinesis, especially when no skeptics are present. Along with quite a few paranormal activities involving cards. I mean, really, what kind of question is that. If anecdotal evidence was ok, this site would have no right to have skeptics in its name.

  • Wow! You can do telekinesis?
    – Sklivvz
    Apr 5 '11 at 18:25
  • Apparently I can't do proper English ;)
    – user288
    Apr 5 '11 at 18:27
  • You can really do telekinesis?
    – Borror0
    Apr 5 '11 at 18:48
  • What if I asked you to upload a video of your telekinesis?
    – apoorv020
    May 5 '11 at 12:54
  • This answer isn't addressing the issue of anecdote, but that of credibility of a witness.
    – Flimzy
    Dec 4 '12 at 23:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .