10
$\begingroup$

On SO, the percentage of unanswered questions is 27%. On cross validated, 39%.

On cross validated, there is a lot of unanswered questions. I suggest it is due to the high technical level required to answer them. I have myself huge difficulties to answer most of the questions.

Is there ways to decrease this percentage?

Should we encourage people with unanswered question to reformulate their question or to step down a level in the difficulty? Or to expand furthermore the problem?

Is there best practices in that cases, where you are willing to help but the questions are one step ahead of your competencies?

I also hope to get advice to make my questions clearer, to get response to my answers. But if these thought could help.

$\endgroup$
6
  • 6
    $\begingroup$ Related: Why is our answer rate so low?. There are perhaps some other peripherally related questions. $\endgroup$
    – Glen_b Mod
    Commented Feb 1, 2016 at 12:35
  • 8
    $\begingroup$ I don't have statistics, but I think a bigger problem is that many questions are just too unspecific to answer. For example, many approach the limiting cases "what is the best analysis for my data which I am not showing you?" or "what should I do in my project?" which are too difficult to answer. Also, in human terms, many transient visitors here have a short-term interest in doing an analysis now and minimal longer-term interest in mastering statistics otherwise. That's inevitable, and not necessarily a problem, as CV exists for such people too, but it drives down answer rates. $\endgroup$
    – Nick Cox
    Commented Feb 1, 2016 at 21:45
  • 2
    $\begingroup$ I rarely feel qualified to uptick answers to "unanswered" questions - sometimes because the material is technically beyond me (not that rare, and unsurprising since low-attention questions tend to be rather more specific, but hopefully someone can assess the quality of these answers) and sometimes because the original question is vague or poorly worded and I'm not actually sure that the answer addresses the question (not that rare either but I suspect many of these questions are lost causes). $\endgroup$
    – Silverfish
    Commented Feb 4, 2016 at 12:24
  • 3
    $\begingroup$ @Silverfish If the question is obscure but the answer looks useful I am generally happy to upvote the answer. The key is whether it might help any user of the site in future. Many answers improve considerably on rather poor questions. (Gross irrelevance can always be an exception.) $\endgroup$
    – Nick Cox
    Commented Feb 4, 2016 at 20:11
  • 1
    $\begingroup$ I come in, sometimes years after the fact and answer some that are in my areas. I have earned "necromancer" for it. I also have a few revival badges. I like having the chance to do it. $\endgroup$ Commented Feb 4, 2016 at 21:59
  • $\begingroup$ There are also too few eyes---more than 10% of the Qs on the site have got the tumbleweed bount! (low vies, no votes in a week after posting) $\endgroup$ Commented Jul 16, 2017 at 13:48

4 Answers 4

7
$\begingroup$

It is worth considering that an unanswered question is defined as a question that has no upvoted (or accepted) answers and is not closed. Also, the community user 'bumps' one unanswered question per hour to the top of the main page, which provides an opportunity for them to be addressed.

From these points, I would make a couple of tentative suggestions to any who wish to help lower our unanswered question rate: When you see an unanswered question bumped by the community user, take a look at it.

  1. Often, these threads have very low views and have answers without any upvotes, perhaps because no one has ever paid any attention to them. If there are any answers, see if they might merit an upvote (cf., CrossValidated's rather extreme up-voting problem).
  2. Also often, in my opinion, questions are not quite clear enough to really be answerable. I will typically give the OP the benefit of the doubt in these cases, possibly leaving a comment prompting them to provide needed information. However, the threads bumped by the community user are all at least one month old. It is common for a drive-by asker to leave a marginally clear / unclear question, not return to monitor the progress of the thread, and be long gone. Clicking on the OP's username will take you to their page where you can see if they are an active user and when they last showed up on the site. You can make a reasonable guess about whether they will return to their question. A question that may not be quite clear enough to be answered with an OP who will probably never return can be closed as unclear what you're asking (cf., Should the standards for closing questions be relaxed somewhat for old questions?).

These are actions that should be doable even by someone who feels the questions are "one step ahead of [their] competencies". I strongly suspect that if all the questions that do have upvotable answers were given just one upvote, and all the not really quite clear enough questions where the OP will likely never return to clarify them were closed, our unanswered question rate would be much lower.

$\endgroup$
4
  • 2
    $\begingroup$ Doesn't community user only bump questions with non-upvoted answers (but does not bump questions with zero answers) -- i.e. de facto answered but de jure unanswered questions? $\endgroup$
    – amoeba
    Commented Feb 3, 2016 at 22:29
  • $\begingroup$ Interesting question, @amoeba. I thought, they could be any unanswered Q (w/ or w/o non-upvoted answers). However, scanning through the last dozen, all had answers, so you may be right. $\endgroup$ Commented Feb 3, 2016 at 22:45
  • 4
    $\begingroup$ Yes, see meta.stackexchange.com/questions/217884. $\endgroup$
    – amoeba
    Commented Feb 3, 2016 at 22:49
  • 4
    $\begingroup$ This is exactly my personal activity policy to help with the site statistics: pay attention on community bumped questions, and try to either upvote the answer, or flag the question for closing (not always possible, though). Once in a while I also risk writing an answer. $\endgroup$ Commented Feb 5, 2016 at 20:31
6
$\begingroup$

Should we encourage people with unanswered question to reformulate their question or to step down a level in the difficulty? Or to expand furthermore the problem?

Yes we should encourage people to make their questions easier to follow. In many cases more explanation of the background, or of the aims of their analysis, or the addition of an example to illustrate the issue, would help. On the other hand, extraneous material & excessive detail is off-putting. But we can't really tell them to ask an easier question just because the one they want to ask is too difficult.

Is there best practices in that cases, where you are willing to help but the questions are one step ahead of your competencies?

Well you can do some research, or perhaps provide a partial answer if appropriate. But if you can't answer, you can't answer—keep an eye open for questions you can.

If you think you know the answer, but aren't completely sure, give it anyway. Of course don't pretend to a certainty you don't have, explain your reasoning, & provide any relevant references. Even if you happen to be quite wrong you might provoke someone into writing a better answer or suggesting corrections to yours—the person who asked the question, you, & future readers all win.

Minimal working examples can be as useful in Statistics as in programming.

‡ A good place to start researching is, of course, Cross Validated; I suspect there are many duplicates & near-duplicates among unanswered questions. Sometimes a question just needs to be flagged as a duplicate; sometimes a short answer explaining just how answers to a linked post or posts relate to it is needed.

$\endgroup$
5
$\begingroup$

(In this answer I leave aside the strategy of "find questions you know the answer to and answer them" ... since I take that as sufficiently obvious to everyone.)

  1. A lot of unanswered questions could probably close as unclear, too broad, off-topic or duplicate. So can vote-to-close or flag (as appropriate to what privileges you have) when you come across them.

  2. A lot of questions have answers or part-answers in comments. If you see a question that - between all the comments - seems to be largely answered, you should feel free to put together the relevant parts and post an answer. (If you're essentially just quoting a comment or two, you need to credit the author or authors, but I think answers in comments are otherwise fair game. Some other SE sites delete any comment that has even a small part of an answer as a matter of course; that's not how we work, but we could be a lot better about taking answers in comments to actual answers.)

  3. A lot of new questions are really not clearly explained; anything you can do to encourage or guide new posters to produce answerable questions will help. (I expect many of us could answer more questions if the effort involved in getting a lot of our questions to a state where they can be answered at all were not taking so much time.)

$\endgroup$
1
-2
$\begingroup$

Some questions are unanswered even though they are interesting, well informed, well expressed, and well posed. I assume they are unanswered because the expertise to answer the questions does not exist on this forum. So, "How to decrease the number of unanswered questions?" Answer: CrossValidated somehow needs to attract more experts who could answer these questions. How to do that? These experts need to know that CrossValidated exists, so advertising to them might help. They need it to be worth their while to participate. The most obvious reward is money. CrossValidated pays with reputation points, but experts know they're good already and may not be motivated by that.

$\endgroup$
3
  • 1
    $\begingroup$ Paying actual money is a radical suggestion! But thanks for the update, I removed my downvote. $\endgroup$
    – amoeba
    Commented Feb 7, 2016 at 14:46
  • 2
    $\begingroup$ I think the idea of actual monetization is short-sighted both in theoretical as well as practical terms. From a practical stand-point: Amazon's Mechanical Turk had problems taking off even within the USA; to expect something like this will work with a globally distributed work-force is naive. (I don't even touch on pricing /work regulations questions) From a theoretical stand-point: The main reason most users come here is: pride and fun. Fun and pride mix really bad with putting price-tags on them. $\endgroup$
    – usεr11852
    Commented Feb 8, 2016 at 21:23
  • $\begingroup$ Well, @user11852, feel free to come with another reward! I was just suggesting that something other than brownie points might motivate better. What that something is, I don't know. $\endgroup$
    – stan
    Commented Feb 8, 2016 at 21:29

You must log in to answer this question.

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