14

Back in 2012, it was established that answers on Politics.SE need to be backed up with references. I am all for that, but I cannot properly explain why that is the case. I regularly ask users to back up their answers with authoritative sources and they sometimes reply that it's not needed because you can just attribute the words to yourself.

I would argue that sources help others verify statements presented as facts, but I think there are more arguments in favour of extensively using sources.

I think it would be helpful to have a canonical post of reasons why answers should be backed up by (authoritative) references. For one thing, it helps avoid having discussions on the real site because we can just refer to the Meta post.

  • prime counter-example: politics.meta.stackexchange.com/questions/3512/… – Fizz Sep 3 '18 at 2:25
  • @Fizz it's a plausible answer, but it's just anecdotal evidence. The answer doesn't show that this actually is the case or that this ever has been the case. If you're writing your paper about felony disenfranchisement, you probably want to point to facts (i.e. historical cases, reports, etc.) rather than make a story that seems plausible. For example, take a look at the introduction of this paper. – JJJ Sep 3 '18 at 10:50
  • @Fizz in this case the plausible story is in line with the facts and evidence. In many cases, we find that this is not the case. Take for example fake news during recent elections. They used what seemed like plausible stories to their audience (making use of what's called confirmation bias) and many people believed it without verifying. – JJJ Sep 3 '18 at 10:53
12

One important property of a good answer is that people have reason to be confident in its correctness.

We need a reason to believe that the answerer's premises and reasoning are valid, and references are a good way of validating the premises and reasoning in an answer.


Most people here seem to agree that references should be required for an answer, but I am not one of those people. My position is, and has always been:

  1. Although References are a good way to validate your answer, it's the validation that's important, not the way it's validated. If an answer can be validated in some other way, then I am satisfied.
  2. This validation is a measure of an answer's quality, and SE measures quality with upvotes and downvotes. There is a flag reason for "Very Low quality", but that's for answers that are nearly gibberish, or otherwise totally irredeemable.
  • 1
    Good answer. In the context of politics, what sort of answer would be possible without references? In mathematics you can rely on common knowledge (you assume grad students are familiar with undergraduate calculus). In politics, I am not sure if that's often the case, so I think it's more difficult to validate an answer by providing an argument without references. – JJJ Sep 2 '18 at 23:14
  • 5
    @JJJ It's up to the readers to decide what they're satisfied by, and upvote and downvote accordingly – Sam I am Sep 2 '18 at 23:16
  • 3
    I find this stance detrimental to the health of the site. "let them vote" as a form of validation seem to bring just partisan voting often not correlated to the quality of the answer. It might not be what this SE is used to so far, but being a bit more strict on asking references might reduce the problem (that I think I observed in my limited time spent here) – Federico Sep 5 '18 at 12:45
  • @Federico yes, scores can be misleading (on any SE site). The reason is that once questions reach the HNQ list they get many more views and upvotes. Therefore, some mediocre answers get a ton of votes whereas very good answers may be viewed by few and receive only a few upvotes. This has been extensively discussed on Meta. – JJJ Sep 5 '18 at 14:49
  • @JJJ I was not referring to the general problem of HNQ voting, but to the problem specific to this SE, given the topic involved. – Federico Sep 5 '18 at 15:24
  • @Federico in that case references won't really help. If people are going to upvote bad content without sources they will also do it with bad or highly partisan sources. And as with many opinions, you can always find someone else (or some publication) to agree (so you can cite them). – JJJ Sep 5 '18 at 15:28
  • @JJJ "And as with many opinions, you can always find someone else (or some publication) to agree (so you can cite them)." But then you know at least from where the thoughts are coming from. This is kind of additional information and may help with estimating the quality of the contribution. – Trilarion Sep 10 '18 at 15:32
7

References help avoid needless disputes, and are appropriate whenever users are relaying facts, findings, and ideas which did not originate with them. References are very useful, since a user might be misquoting something, or have misunderstood what it means; or it might be an authoritative reference that leaves little in doubt, or at least less doubt than no reference at all.


Not everything needs a reference however. What's 2 + 2 could of course have a reference of some 400 page tract on The Foundations of Mathematics, but an answer of 4 is usually better.


Some folks confuse deductive reasoning with opinion. But syllogisms, algorithms and logic are objects, not opinion. Opinion might be the input data we plug into the logic, or perhaps the choice of a particular algorithm, but if the input data is:

  • commonplace factual data, then valid deductions from those are no less true than the factual data.

  • questionable data which the OP takes as a given, then applying valid logic to that is more of an extension of the question. If the OP implies X and Y, but those can be shown to prove that X is impossible, that would prove the question is not useful, and would therefore be an excellent answer.

  • a complete set of mutually exclusive possibilities, then enumerating their various logical consequences may be useful. Those consequences can then be compared to the actual state of affairs, and those possibilities which never lead anywhere real may be eliminated.


References without any implication, quote, summary, or estimate of their content are basically just clickbait, which is too often more trouble than it's worth.

3

Thanks for joining us on Meta!

Why we reference

All stacks have an on-going concern with providing high-quality answers to users. Accordingly, every Stack I've seen uses the back-it up principle.

Each site tailors this principle to its own environment. Here, one of the chief concerns is that there is a very low entry to barrier. Most people don't believe they can answer machine learning questions on CrossValidated.SE or explain religious concepts on Islam.SE, but nearly everyone has a political opinion and believes they can answer questions here.

The back-it up principle is what allows the community to distinguish between personal opinion and fact. When you back-up an answer you are claiming that it is more than just your opinion.

As a secondary, but related, benefit: backing-up an answer requires a certain amount of honesty about what has influenced the answer. Everyone has their own personal opinions, but answers need to be more substantial than personal opinion. By asking answers to be backed-up we are requiring transparency about the sources an answer is based on.

When we reference

All that being said, there are two situations in which no external references are necessary.

First, no references are needed when an answer can be constructed entirely deductively. This is a rare occurrence. Even in political theory, where such questions are possible, it almost never happens. It's never true for empirical questions, because you can never control for all possible intervening factors.

Second, no external references are needed when the answerer is the expert and they are relying on their own expertise. Certainly the user should mention their expertise in their answer, but it wouldn't be reasonable to require them to provide an external reference demonstrating their expertise. Of course, by making this claim you are inviting others to determine whether your experience is applicable to the question you are answering.

You must log in to answer this question.

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