4

An "Accepted Answer" means something different on each Stack Exchange site. On some sites it's a functional question; this solution worked, full stop. We have a more subjective case.

When should a user accept an answer? When their curiosity is satisfied? When the community seems to have come to a consensus on the validity of an answer? Only when peer reviewed research supports an answer?

Obviously when to accept is up to the user, but I think it make sense to have a guideline so we can evaluate whether it's acceptable for users to have very low accept rates.

  • 3
    Any question that is not closed as not-a-real-question should have a possible answer that can be accepted. If one can't imagine what sort of answer would answer a given question, then the question should be closed as not-a-real-question. – Artem Kaznatcheev Feb 3 '12 at 4:03
  • @Artem I basically agree, but I am still a fan in some instances of the occasional well selected big-list style question, but at least on this site I may be in the minority (see meta discussion) – Jeromy Anglim Feb 3 '12 at 4:06
  • @JeromyAnglim yeah, good big-list questions are definitely fun (especially when clearly marked as community wiki, is that possible anymore?). It was really unfortunate that this question was closed for instance. However, these sort of questions should be uncommon. Right now we seem to be having an issue with lots of open-ended almost self-help questions that are not precise enough to have a definite answer. If the majority of the material on the site is like this, then I think we will have a hard time attracting researchers and experts. – Artem Kaznatcheev Feb 3 '12 at 4:10
  • @ArtemKaznatcheev (a) I did vote to close that question; I think a similar question with a narrower scope might have been acceptable (all of cog sci seemed a bit broad to me, in addition to it being a big-list format); (b) I think the issue of attracting experts is a big one; there's some meta discussion already about that, but there's scope for more; perhaps you want to ask a particular question on meta about that. – Jeromy Anglim Feb 3 '12 at 4:18
  • relevant discussion of acceptable accept rates from meta.so: meta.stackexchange.com/questions/16721/… – Jeromy Anglim Feb 10 '12 at 2:01
  • I'm all for the discriminating palate in selecting answers, but from a practical standpoint having too low of a rate could discourage would-be responders to one's questions. It shouldn't happen that way, but it does. – Chuck Sherrington Feb 16 '12 at 5:28
  • @jonsca I think that depends on the site too, IMO that shouldn't be as much of a problem on a science-minded site (though I'm aware it might be). Low accepted on Stack Overflow generally means you don't care/know to accept answers, low accepts on our site might simply mean you haven't gotten acceptable answers or that they don't exist (which is possible in science). – Ben Brocka Feb 16 '12 at 14:27
  • 1
    @BenBrocka I agree that it does depend on the site. I went over to TCS (which is a great model for sites aimed at academic/research level questions), and I found varying percentages of acceptance. I understand that you want to protect the integrity of the answers on the site, and I do want to respect that notion, but I'm just offering the idea that seeking perfection for answers will stagnate the site. I don't think that's an immediate danger, though. We should be modeling an atmosphere where answers are fluid, updated with pertinent information, and marked/unmarked as appropriate. – Chuck Sherrington Feb 16 '12 at 20:49
4

Good question. I agree that it is largely a personal decision for the person asking the question. That said, here are a few criteria that I'd consider:

  • Completeness: The answer is sufficiently complete. It provides an answer to the core question that is asked. In some sense saying that a question is answered suggests that no more answers are required by the person asking the question. In many occasions on cogsci.se a person will make a few points and add a reference or two. Such answers are often useful, but more could be said.
  • Accuracy: The answer is considered accurate and true by the person asking the question.
  • Evidence: The answer provides reasonable evidence supporting the claims made. In cognitive science and psychology this often requires reference to empirical research; and sometimes references are often used whereby the arguments in the reference are used to justify claims.
  • One clear winner: Where there are multiple answers that are both good, there probably should be one clear winner. I suppose if both are good, and one is a little bit better, then that would be fine; however, in such cases I'm often more inclined to let the voting play their part to determine the best answer.
  • Not own answer: Sometimes you can ask a question, get no responses, and then work out an answer yourself and post that. When this happens to me, I prefer not to award myself the "correct answer" on the basis of conflict of interest.
  • 2
    I think these are all great points except the last one. If you found the correct answer yourself, accept it! That's your way of telling others that you feel you found the best answer. There's no shame in that at all! – Josh Feb 3 '12 at 18:37
  • @Josh okay. That's good to know. Personally, I like the idea of encouraging people to answer their own questions. A slightly fuzzier situation can emerge where I have my own answer and there is another reasonable answer. Anyway, this comment is getting longer, perhaps there is a possible meta question on encouraging people to answer their own question and etiquette about awarding yourself "answered" to your own answer. – Jeromy Anglim Feb 4 '12 at 1:10
  • 1
    In the case where I have my own answer and others have given reasonable (but maybe not perfect) answers, I tend to accept the other user's answer and still post my own. – Josh Feb 4 '12 at 20:09
  • 1
    That sounds polite and reasonable. – Jeromy Anglim Feb 4 '12 at 23:50

You must log in to answer this question.

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