This is a problem I encounter relatively frequently when trying to tidy up the unanswered questions list.

Let's take this case as an example: What do my residuals say about my data?

The OP posted a relatively meaningless plot and asked for an interpretation. @Glen_b left a comment 2 years ago pointing out that there are far better plots to use, but there's been no response and the question was never edited.

Now that comments have been left and ignored, it seems reasonable to close. But the question is not obviously off-topic, or too broad, or unclear - it's just misguided. How should I flag it? Does the particular flag matter much in a case like this?

An alternative would be to strictly answer the question in the title, but such an answer wouldn't really be helpful - the main purpose would be to remove this from the unanswered questions list.

up vote 7 down vote accepted

As a non-moderator, I think it's sufficient to write a comment asking for clarification & register a close vote at the same time. If a comment already asks for clarification, and you agree with it, then you only have to up-vote the existing comment.

The reason I say "non-moderator" is that it takes 4 other non-moderator votes to close the question; perhaps it's unclear to you but clear enough to other reviewers.

I say that you should vote to close at the same time because it helps precisely no one to have an open & unclear question. If OP cares enough, they will revise it. If the question is revised before it's closed, you can retract your close vote. If the question is revised after it's closed, you can vote to re-open. I try to check on the status of the questions I've recently voted to close as unclear (sometimes the edits help a lot but the re-open vote gets stuck at 4 votes) but this is admittedly imperfect.

The trick is to write a comment that is precise enough that OP knows what kind of clarification to add. Sometimes I get this wrong and I ask for clarification in a way that invites OP to add a bunch of irrelevant context.

Sometimes I also delve into old, unanswered questions that are not closed and which bear a particular tag. For example, I might query [neural-networks] answers:0 closed:no and then sort by date, starting with the oldest questions. The purpose of this exercise is to remediate the question in some way -- upvote & answer it, find a duplicate, or close it. Less commonly, if it's on-topic and not a duplicate and I don't know the answer, then I just up-vote it and move on; not everyone can answer every question!

To do this efficiently, I've outlined a mental procedure that I walk through for each question. These are not in any particular order.

  • Is this an answerable question? (For example, questions that ask "is a neural network or logistic regression better?" or "what threshold should I choose for my classifier?" is not an answerable question.)
  • Can a person who is familiar with this topic write a clear, concise answer to it? Or is critical information missing?
  • Does this question ask about statistics or machine learning as discussed in the help center?
  • Does answering this question rely on writing code or debugging code?
  • If I search stats.SE for the key words, do I find a duplicate question?
  • 1
    Thank you, this is helpful. – mkt Aug 14 at 9:19

Not answering Glen_b's comment does not seem to have anything to do with this question being a candidate for closing. I'd say that this question is simply too vague/too broad, so I closed it (if someone disagrees, feel free to vote or comment).

As about general question, if the question had some quality issues and the comment asked for clarification so that it is not closed, and there was no response, then it should be closed. The story is different if the comment asked for clarifying something, but nonetheless the question itself wasn't unclear, then there's no reason for closing it. Nobody is obliged to respond to any of the comments.

Personally, I ask for clarification without voting to close if the question is "almost" on-topic, can be easily made on-topic after edits (yes, this is very subjective!). If the question crosses the boundary of being on-topic more significantly, then I vote to close and comment, since we already have lots of low quality, "abandoned" questions with no answer. From my experience, usually if OP cares about his question they edit it, while in other cases we never see any response, so OP did not seem to care about answer to their question.

  • 4
    Yes, but also: downvote! This particular Q is old and now closed. However it won't be deleted because it's upvoted. If one sees an unhelpful question, downvote! I downvoted this one and if both mkt and Tim downvote further, it will eventually get automatically deleted => profit. – amoeba Aug 13 at 13:26
  • 1
    Thanks for chiming in, Tim (& @amoeba). I wouldn't have interpreted asking how to interpret a specific, simple plot to be too broad (as opposed to poor/misguided). But it's good to know that this is how that flag is used in practice. I sometimes wish the text of the flags was less specific because of these edge cases, but perhaps I'm too focussed on the wording. – mkt Aug 13 at 15:06
  • 3
    @amoeba I agree with the downvoting suggestion and have now done so; the problem of course is that this question would have lasted much longer if it wasn't raised here (and this is true of others as well). – mkt Aug 13 at 15:08

You must log in to answer this question.

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