16
$\begingroup$

Question https://stats.stackexchange.com/q/222702/58675 shows a classical errors-in-variables model (and in particular, since the calibration curve looks polynomial, total least square may suffice). What is the right way to answer here? A link to the Wikipedia article is clearly not an acceptable answer. An answer which repeats, more or less, the Wikipedia exposition doesn't seem great too (it's total least squares...there isn't a lot more you can say, apart from the standard derivation). Should I simply point to the Wikipedia link in a comment? Or should I ask for actual data (a reproducible example), and give a real answer based on that? This would be better than just "copying" Wikipedia, because I would give an answer with actual numbers, maybe illustrating the use of some R package.

$\endgroup$
  • 2
    $\begingroup$ Good question! This example, though, really wants an answer explaining how to make the choice between different approaches. Note the application - prediction of volume from distance measured with the same sensor. I think it may be a duplicate, as I seem to remember some very good post on this. $\endgroup$ – Scortchi - Reinstate Monica Jul 8 '16 at 9:56
  • 1
    $\begingroup$ Since the OP seems to have some idea about what s/he is doing but is missing a crucial pointer in this case a link to some reliable source like Wikipedia might well be enough. If they seemed less clued up then you may need to give some basic detail in your reply as well as the link. I do not think there is a one-size-fits-all answer to your question which is why this is a a comment not an answer. $\endgroup$ – mdewey Jul 9 '16 at 9:29
  • $\begingroup$ @Scortchi, thanks for the input. Not sure I got your point about prediction of volume with distance from same instrument - could you elaborate here or in that question? $\endgroup$ – DeltaIV Jul 9 '16 at 11:47
  • $\begingroup$ @mdewey thanks for your suggestion. I will start with a link in the comment. $\endgroup$ – DeltaIV Jul 9 '16 at 11:52
  • $\begingroup$ On second thoughts, the OP question might be much more complicated than I thought. Maybe I'll have to ask a question on CS myself ☺ $\endgroup$ – DeltaIV Jul 9 '16 at 14:13
  • $\begingroup$ Just that it's important to distinguish between estimating the relationship between the unobserved 'true' variables & predicting observations of one from observations of another. $\endgroup$ – Scortchi - Reinstate Monica Jul 11 '16 at 8:45
  • $\begingroup$ Ouch! The question has been deleted...maybe it was a duplicate, as @Scortchi suspected. Is there a way to know who the OP was? I would like to see if she/he posted a different question, or find out which is the post this question was a duplicate of. $\endgroup$ – DeltaIV Jul 13 '16 at 9:42
  • $\begingroup$ With enough reputation, one can see who the OP was. I checked: he does not have any other posts on CV (neither Qs nor As). $\endgroup$ – amoeba says Reinstate Monica Jul 13 '16 at 11:42
  • $\begingroup$ thanks @amoeba . $\endgroup$ – DeltaIV Jul 13 '16 at 12:02
9
$\begingroup$

Usually I comment, and ask for clarification what part of the question is not already answered by Wikipedia. Then I vote to close as "unclear what you are asking" or "too broad".

unclear what you're asking Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking.

If Wikipedia has a nice answer already, people should read that first, and then clarify where they did not understand Wikipedia.

too broad: There are either too many possible answers, or good answers would be too long for this format. Please add details to narrow the answer set or to isolate an issue that can be answered in a few paragraphs.

Often good Wikipedia articles are "too long for this format".

Here is an example of such a question that IMHO is "too broad" (and answered by Wikipedia):

https://stats.stackexchange.com/questions/223875/can-anyone-help-me-to-find-out-the-working-of-expectation-maximisation-clusterin

Usually I consider closing such questions to be appropriate. If they are well answered by Wikipedia, they are usually not well posed questions. If we don't close them, people will just waste their time on adding yet another slightly different answer.

In some cases, I also post an answer that cites only the relevant paragraph from Wikipedia with an prominent link to hint that they should have read that page... I doubt that Wikipedia "changing" is a major problem for us; and users can go back in time with Wikipedia. Answers have time information that is accurate enough to find the Wikipedia page at that time, usually. I doubt that for this kind of questions, Wikipedia ever will change to a substantially worse article. If this ever happens, only one person needs to investigate this, and edit the question with the permanent revision URL. But in my opinion, we have more incorrect answers on SE than incorrect information in Wikipedia...

$\endgroup$
  • 3
    $\begingroup$ And you can always permalink to a particular revision of a Wikipedia article if you want. $\endgroup$ – Kodiologist Jul 16 '16 at 19:59
12
$\begingroup$

Setting aside the specific thread on the main site that motivated this question, I think we can address the title question more generally.

Namely, I think we should be mostly indifferent to what is posted elsewhere on the internet. In this case, I think it shouldn't matter if a question is answered on Wikipedia, but we could apply the same principle to online statistics textbooks (e.g., The Engineering Statistics Handbook), tutorial sites, (e.g., UCLA's statistics help site), etc.

  • A good deal of the information on CV is likely to be duplicated somewhere on the internet. Saying that we should only provide unique information would be too constraining.
  • The internet is an unstable place. What happens if the information elsewhere changes? This is particularly an issue with Wikipedia, which may not say tomorrow what it says today. (It is possible to permalink a particular historical version of a page, but this isn't commonly done.)

A question should not be closed because it seeks information that exists elsewhere. (It should be closed if it seeks information that already exists on CV.) If you have an answer that amounts to the same thing that is written elsewhere, I think you should be free to post it. (That is, if it is your answer. If you copy and paste, you should cite at a minimum, and under most circumstances you probably shouldn't post an answer that is only copied without any additional contribution.)

$\endgroup$
  • 1
    $\begingroup$ I would prefer to reference a high quality answer on CV compared to an article on wikipedia. I would, however, check scholarpedia for answers. $\endgroup$ – EngrStudent - Reinstate Monica Jul 15 '16 at 13:28
4
$\begingroup$

I think an answer that only copies and pastes from Wikipedia, with appropriate attribution, is fine if it answers the question completely. Originality is nice, but shouldn't be required. I believe Wikipedia's license is compatible with Stack Exchange's. Copying and pasting more than a few sentences from sites without free licenses could be legally problematic.

Any answer to a question, even a very short one, should be posted as an answer rather than a comment, because otherwise Stack Exchange can't recognize the question as answered.

Here's an example of an answer I just wrote that's basically just a copy-and-paste from Wikipedia.

$\endgroup$

You must log in to answer this question.

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