I understand that questions on AC.SE are going to be "softer" in nature than the programming questions typical of SO.SE, but I think our answers often are "softer" then needed. I often read answers and think "Citation Needed". Are we answering questions based on our personal experience when we could be providing answers based on well conducted research? Should we be up voting "easy" answers that we agree with when they are void of references?

  • Duplicate of meta.academia.stackexchange.com/q/12/102 ? – user102 Jul 29 '13 at 11:01
  • @CharlesMorisset that question asks about "how should we deal with "soft" answers to questions that demand them?" and I am asking about questions that DO NOT demand soft answers but are getting them anyway. – StrongBad Jul 29 '13 at 11:07

This is a very difficult area for our forum, as most of the answers and advice dispensed here is not borne of thoroughly research, but rather real-life (and therefore pretty localized) battle scars.

Personally, I favor the "hands-off" approach; policing this sort of thing is very tedious, with minimal benefit, as the community tends to do a good job upvoting stronger answers over the soft ones. As the distinction between strong and soft answers is typically pretty nebulous, I don't think a policy-based approach for treating this is correct. If this is something which needs to be addressed (which I'm not sure of myself), I would much prefer an education-based approach. One method could involve a community-defined comment template to leave on those sorts of answers, suggesting the commenter leave stronger answers and directing the answerer to the appropriate meta discussions. Another parallel method could be to make use of the community blog feature to publicize posting guidelines and suggestions. I would much prefer these to policy.


I think that, without forbidding experience-based answers and anecdotal evidence, we should push ourselves a little harder towards fact-based and statistics-based answers. There are quite a few questions on the site who are answered mostly with “I advise you to do this” or “I observed around me that X is more common than Y”, and which could be much improved if they were backed up with links to actual official policies and/or statistics.

Now, how to do that? I myself try, on questions where it is appropriate, to add a comment to try and remind us that fact-based answers are at the core of the Stack Exchange Q&A concept, and while we have a “softer” policy than most other sites of the network, fact-based answers are good. While I don't always add such comments (also because I don't want to cast myself into “that guy”), I think it's overall a great strategy to steer us in that direction, without policing. So, please consider adding them too!

  • As recent examples of question where I think we could have a more fact-based approach: here and there – F'x Jul 31 '13 at 16:40
  • 1
    I am not sure where the idea of policing it came in. That wasn't my intent. I think your answer echoes my thoughts exactly. – StrongBad Jul 31 '13 at 17:13
  • 2
    @F'x I don't mind seeing a fact-based answer if one is indeed available. But in the vast majority of questions here they only provide a partial solution. Your response to the second question you listed is a perfect example: for every "around 4 weeks" program, there are probably twenty that are much lower or higher, and depend on the advisor. The large standard deviation is what kills you on that question, and (I believe) a better answer is "it depends." Not as satisfying, perhaps, but it is easy to be fooled by an average in this case. – Chris Gregg Aug 1 '13 at 16:16
  • 1
    @F'x To go a bit further--if someone applied to grad school expecting 20 days vacation per year, he or she could end up very disappointed, or very happy, and would be doing him/herself a disservice by not seeing if a policy was in place before applying. You certainly stated the appropriate caveat in your answer, but this is almost always going to be necessary in these types of questions. – Chris Gregg Aug 1 '13 at 16:22
  • @ChrisGregg if your point is that most questions would be much better answer by “it depends” than actually collecting the statistics, to the extent that they exist and taken with a grain of salt, then I have to disagree – F'x Aug 1 '13 at 17:02
  • 1
    @F'x Yes, my point is that this is the case for many questions we get here. Using your example (and not to pick on you): trying to generalize the typical amount of vacation time by using a tiny sample size is misleading at best, and disingenuous at worst. I completely agree that statistics are useful when they are used properly, but I would urge against trying to shoehorn statistics into answers for the sake of trying to be scientific about it. – Chris Gregg Aug 1 '13 at 19:24
  • @ChrisGregg first, the sample can of course be expanded… and it also means “such policies do exist, and you can go look it up for the places you apply to”, which is to me an important message (more than “sorry boy, it depends”) – F'x Aug 1 '13 at 20:25
  • 2
    @ChrisGregg clearly a larger population would be better. In the absence of a large population survey, looking at a small sample of schools and their official policies is a much better answer than personal experience without any links to official policies. In the absence of a link to a survey I like the answer by F'x. While it is a small population it provides the insight into how to really get the answer. – StrongBad Aug 2 '13 at 8:44
  • 2
    @DanielE.Shub I disagree that it is a better answer -- it is misleading to assume that by looking at a few schools' policies you can get a general idea for vacation time, which is far, far more dependent on discussions with an individual advisor (and regardless of official policy). To "really get the answer" is to ask your advisor. If you don't yet have an advisor or a school, then you need to either reach out ahead of time or start working on your persuasive skills for eventually talking to your advisor. That question is the perfect counterexample to your argument, IMHO. – Chris Gregg Aug 2 '13 at 9:32
  • 2
    @F'x Ah, but the key here is that even at schools with a policy, the policy is almost certainly deferential to your advisor's whim. Quoting a policy someone put on a website actually does a disservice in this case. Academia is one place where there is so much leeway given to professors that I would argue it is more common to disregard the policy than to uphold it. I'm not disagreeing that there are lots of questions that benefit from data, but this example is (IMHO) one where it is not. – Chris Gregg Aug 2 '13 at 9:38
  • @ChrisGregg I will just state one final remark on this: I don't think you can be better off by not knowing the policy, rather than knowing it (and knowing that, as in life, actual results may not strictly conform to the policy). I think, if you're a certain type of professor (not an accusation against anyone, just a statement of fact), you may prefer that the policy is not known :) – F'x Aug 2 '13 at 10:03
  • @F'x BTW, “such policies do exist, and you can go look it up for the places you apply to” -- I absolutely agree that this is a good answer to the question. – Chris Gregg Aug 2 '13 at 10:03
  • @F'x To your last point--and I will drop this, too! :) -- I agree with your comment that it is better to know the official policy than to not know it. – Chris Gregg Aug 2 '13 at 10:09

I agree with eykanal, and also prefer an "hands-off" approach. We already have mechanisms for dealing with "undocumented" soft answers: down-votes, comments and even deletion. In extreme cases, there is even a mod mechanism allowing us to put a post notice indicating that a post requires some citations or some further explanations.

  • 3
    But we are not using the down votes and it seems we are actually up voting them. – StrongBad Jul 31 '13 at 17:14
  • @DanielE.Shub: I agree with you, I'm just saying that we somehow have all the tools already, and that if the community does not down-vote them, then I'm not sure there is anything we can do. – user102 Aug 1 '13 at 12:07

You must log in to answer this question.

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