18
$\begingroup$

There have been many threads on CV-meta about whether questions are on-topic, off-topic, are too-localized, etc.

For example:

I offer this question as a clearing house where we can keep a running and persistent tally regarding the criteria that we use to establish whether something is on-topic, off-topic, too-localized, needs to be migrated, and/or low quality. These criteria should only be applied to contemporary posts at the time they are made. Criteria for deleting/closing/migrating questions made prior to the date of this post should be discussed elsewhere.

Please provide only one criterion per answer. Please title your criterion so that it may be referred to in other criteria. Please provide examples when possible.

Please liberally up-vote and down-vote proposed criteria.

Each criterion that crosses a certain threshold of up-votes (I propose 5) should be considered to be in-force.

Additional Rules

  • If two criteria conflict, then the criterion with more upvotes applies, unless another criterion with more than 5 upvotes specifically addresses the case of conflict (in which case please cite the titles of the specific criteria implicated in your conflict resolution criterion).
  • If two conflicting criteria have an equal number of votes, but no de-conflicting criterion has crossed the threshold of acceptance, then the criterion with the later original post date applies.

When possible discussions should occur in chat.

Active Chats

  • None
$\endgroup$
  • $\begingroup$ @Stephan Do you think this list of Qs is helpful in this thread? One can as well link to the on-topic tag which by now has over 100 threads. If you think this thread should be maintained as a useful resource (I am not sure about that actually; it's a bit old), then maybe we could edit the Q to shorten/simplify it quite drastically. $\endgroup$ – amoeba says Reinstate Monica Nov 22 '17 at 12:11
  • $\begingroup$ I think a central repository for what is and what isn't on-topic makes sense. 100 questions in the on-topic tag are a bit too many, especially since some of those questions are very specific. So yes, I think this question should be kept and maybe edited. $\endgroup$ – S. Kolassa - Reinstate Monica Nov 22 '17 at 12:16
  • $\begingroup$ @StephanKolassa I did not mean listing all 119 threads in this question. I meant removing the list altogether and simply linking to the tag. $\endgroup$ – amoeba says Reinstate Monica Nov 22 '17 at 13:43
  • $\begingroup$ @amoeba: that makes sense, if we keep the answers below that summarize "general" guidelines on what is on-topic. $\endgroup$ – S. Kolassa - Reinstate Monica Nov 22 '17 at 14:39
  • $\begingroup$ @amoeba As the original author of the post I'd be pro shortening up the question. In particular the examples were topical at the time, but are now dated. I do think the thread should be maintained as a useful resource, unless there is another thread that does as good a job of expressing clear standards as voted for by the community. In fact, unless there is a better candidate, it probably should be linked to from the 'on-topic tag'. $\endgroup$ – russellpierce Nov 22 '17 at 18:18
  • $\begingroup$ I'd also be pro clarifying that this question is only a form of community guidance and that official policy as crafted by the mods is in another location (and providing the link). $\endgroup$ – russellpierce Nov 22 '17 at 18:22

10 Answers 10

21
$\begingroup$

On-topic: Terminology

As long as it's statistical terminology, I say it's on topic. Lots of statistical ideas have multiple names.

$\endgroup$
21
$\begingroup$

On-topic: Notation

Notation questions should be on-topic. Notation is critical; it is also often hard to search for. I would be generous with notation that is used in multiple fields such as math and statistics.

$\endgroup$
19
$\begingroup$

On-topic: Machine learning questions

Note: This criterion is currently accepted per this post and this post unless it falls below 0 votes.

  • No special criteria regarding this topic have been specified.
  • No criteria have been excluded for this topic
$\endgroup$
14
$\begingroup$

On topic: Program specific questions (e.g. R, SAS, SPSS, Matlab, C++, python, etc) with any statistical content

Note: This criterion is currently accepted per this post and this post unless it falls below 0 votes.

  • Answer the question unless it clearly has no statistical content.
  • No other special criteria regarding this topic have been accepted
  • No criteria have been excluded for this topic
  • Related criterion under review:
    • On topic: Questions that require statistical reasoning to be answered correctly
$\endgroup$
  • 13
    $\begingroup$ +1. Even when a question has no explicit statistical content it may be on topic if you can determine that it requires statistical reasoning to be answered correctly! (An example would be someone asking for help in debugging code that obviously is the wrong solution to their actual problem.) Many such questions unfortunately are migrated away to SO before anyone points out their statistical nature. Often, leaving a quick comment asking for clarification will help decide the matter. $\endgroup$ – whuber Jun 14 '13 at 17:48
  • 2
    $\begingroup$ +1. I think it is too easy, when looking at the numbers of program-specific questions on SO vs. CV, to underestimate the effects of (a) SO being far larger than CV, (b) SO being older, more established, and more well-known than CV, and (c) users of SO being biased toward wanting to do whatever they can on their "main" account. Without all of those factors, I think it would be more clear that the better home for most of those questions is here at CV. $\endgroup$ – A.M. Jul 2 '13 at 3:26
12
$\begingroup$

On topic: Literature/reference requests relating to topics that are on-topic on CV

  • No other special criteria regarding this topic have been accepted
  • No other special criteria regarding this topic have been excluded
$\endgroup$
11
$\begingroup$

On-topic: "find the distribution”

Note: This criterion is currently accepted per this post unless it falls below 0 votes.

"Find the distribution" topics are on-topic if:

  • They do not meet any other exclusion criteria
  • Specifically the following criteria are relevant to consult:

    • Too Localized (not yet specified, cf. the FAQ and use your common judgement): If your 'find the distribution' question is specific to your data-set then it is too localized to be a good question on CV unless:

      1. The question asker demonstrates considerable research effort and describe clearly the steps they have taken to solve the problem (distributions considered etc)
      2. The question asker clearly describes the source(s) of the data.
      3. The question asker's problem is not only about using software to answer their question in which case the question should be migrated to the appropriate stackexchange site.
    • Not a real question (not yet specified, cf. the FAQ and use your common judgement)

    • Duplicate (not yet specified, cf. the FAQ and use your common judgement; list specific duplicates here, and cite them when closing the question)

Unless otherwise noted, questions meeting this criterion should be closed.

$\endgroup$
9
$\begingroup$

On topic: Questions that require statistical reasoning to be answered correctly

For example, someone asking for help debugging code that is obviously the wrong solution to their actual problem.

Note: This criterion is atypical because it is a property based on the answer rather than the question itself. If approved this criterion may be difficult to interpret consistently.

$\endgroup$
  • 1
    $\begingroup$ A good cue about whether a question requires statistical reasoning to be answered correctly can actually be found in the question itself: Does the questioner think so? Did the question get posted on CV, presumably with the goal of being seen by people with expertise in statistics? Even questions that end up being answerable by a non-statistician might reasonably belong on CV if the questioner and others who search for the question and answers in the future think a statistician would best handle it. $\endgroup$ – A.M. Jul 2 '13 at 3:33
  • 1
    $\begingroup$ I don't agree with you here A.M. The consequence of the criterion you are proposing seems to be very broad... I think perhaps unacceptably so. The present criterion is meant to reflect some comments from @whuber that I think may be misguided (for reasons he himself had pointed out prior to suggesting the criterion). Nevertheless, your comment does not reflect the spirit of this criterion. So, I would ask that you delete your comment. If you feel that the criterion you propose should be presented to the community, please do post it as a separate answer so that the community may decide. $\endgroup$ – russellpierce Jul 3 '13 at 3:12
  • $\begingroup$ +1. I sympathise with @A.M.'s concern here, but once the judgment of any OP is regarded as paramount, the whole idea of community judgment is moot. We are on a slippery slope to "It's my question and I think it is OK, so there it is". $\endgroup$ – Nick Cox Jul 5 '13 at 23:19
  • $\begingroup$ @NickCox I just said it was a good cue, not that no rules apply to OPs. By the way, the judgement of any OP is already regarded as paramount when it comes to choosing the best answer (unilateral decision on that big green checkmark), and I think looking at the reasoning behind giving the OP that much power would be informative here. $\endgroup$ – A.M. Jul 6 '13 at 1:32
  • $\begingroup$ I've added a chat room if we want to continue this conversation: chat.stackexchange.com/rooms/9537/ongoing-on-topic-chat $\endgroup$ – russellpierce Jul 6 '13 at 2:26
8
$\begingroup$

On Topic: Questions related to survey methodology as it relates to statistical analysis

This includes "soft" concepts such as questionnaire design, non-response biases, interviewer effects, mode effects, timing of the phone calls, and other operational aspects so long as the question either directly addresses a statistical issue or the question asker indicates that their concern is in regards to later statistical analysis rather solely the practical operational aspects of conducting the survey.

$\endgroup$
6
$\begingroup$

Off-topic: "where to publish"

Questions on where to publish academic statistical or Machine Learning manuscripts are off-topic, as voted at this question.

$\endgroup$
4
$\begingroup$

Low quality: Opinions sought in regards to literature materials without descriptions of the relevant material or key discussion questions

  • No other special criteria regarding this topic have been accepted
  • No other special criteria regarding this topic have been excluded
$\endgroup$

You must log in to answer this question.

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