Our recent proposal for adding one more close reason was declined. The reason was that we could re-use the "datasets" reason

Questions about obtaining specific datasets are off-topic on Cross Validated. You are welcome to post this question to the Open Data site instead, thank you!

as it is used only in 0.5% of the close reasons.

Should we split

This question appears to be off-topic because EITHER it is not about statistics, machine learning, data analysis, data mining, or data visualization, OR it focuses on programming, debugging, or performing routine operations within a statistical computing platform. If the latter, you could try the support links we maintain.

into two reasons while re-using the slot from "datasets" close reason?

What is our decision on this? I personally see no problem, but I'm curious about your opinions.

  • 2
    $\begingroup$ Could we in place include the Open Data SE site in the list of migration targets? $\endgroup$ – kjetil b halvorsen Mar 28 '19 at 16:09
  • 2
    $\begingroup$ @kjetilbhalvorsen No. We don't generally establish migration targets to/from beta sites. You can always flag something for migration and let the mods decide whether it's appropriate or not, though. $\endgroup$ – Catija Mar 28 '19 at 16:38
  • 5
    $\begingroup$ Personally, I'd much rather keep the data sets close reason and expand the number of reasons by 1 even though that means some additional effort. $\endgroup$ – Glen_b Mar 29 '19 at 10:58
  • 1
    $\begingroup$ I do not like the datasets reason for closure: the etymology of statistics is information about States and many professional statisticians' work is related to this so it should be on-topic here $\endgroup$ – Henry Mar 30 '19 at 15:50
  • 2
    $\begingroup$ @Henry, maybe we could italicize specific? I think there probably is room for questions like "What are the benchmark datasets for X?" but I'm much less excited about questions regarding where to download TinyImages or what paperwork you need to complete to download mental health data sets from the NIH. $\endgroup$ – Matt Krause Apr 1 '19 at 13:48
  • 1
    $\begingroup$ @Glen_b But this was explicitly declined by SE admins, so I guess this option is not available and we have to choose from the alternatives, right? We either do nothing, or split this closure reason in two and get rid of the datasets close reason. Which way do you prefer? $\endgroup$ – amoeba Apr 4 '19 at 21:29
  • $\begingroup$ My preference is to continue to argue the toss, which I did (though as a comment). Of these two I'd slightly prefer the status quo, to be honest; the data reason isn't that common but it's very useful; there are good arguments for the other way, though and I don't disagree with the general points being made. $\endgroup$ – Glen_b Apr 5 '19 at 1:51
  • 1
    $\begingroup$ @glen_b maybe dataset-obtaining questions could be handled via "belongs to another SE site" and finally getting some choice there (data science, opendata)? $\endgroup$ – Has QUIT--Anony-Mousse Apr 7 '19 at 7:37
  • 4
    $\begingroup$ @Anony-Mousse this was discussed, the problem is that they are beta sites and can't be added. You can flag them and moderators can transfer them. $\endgroup$ – Tim Apr 7 '19 at 7:43

I like the idea of splitting. I don't know the exact history here, but it shouldn't determine the best set of reasons to use henceforth. The EITHER/OR structure of that reason has long seemed awkward to me: now it appears that it was a way to get round the limit on number of reasons allowed. What is important here, simply but fundamentally, is that reasons given are clear not just to those who vote for one, but also to the posters on the receiving end, often people new to CV or SE in general.

This question appears to be off-topic because it focuses on programming, debugging, or performing routine operations within a statistical computing platform. You could try the support links we maintain.

could be an example of a crisper, more focused reason to close often used here.

This question appears to be off-topic because it does not relate to the theory or practice of statistics, machine learning, data analysis, data mining, or data visualization.

could be an example of a crisper but catch-all reason to close (otherwise put, irrelevant to what we discuss here, while trying to avoid the circularity of saying this is irrelevant because it is irrelevant).

Asking for datasets may not now be very common, but we do see it done. That was a useful reason to close, but sacrificing that reason seems practical. But I suggest also adding nuance to https://stats.stackexchange.com/help/on-topic where we already say

Questions about obtaining particular datasets are off-topic (they are too specialized). The GIS site welcomes inquiries about obtaining geographically related datasets.

We could give that a bit of a boost, say

Questions about obtaining particular datasets are off-topic. They usually are too specialized or raise programming or other computing issues of downloading and processing files better discussed elsewhere on Stack Exchange. The GIS site welcomes inquiries about obtaining geographically related datasets.

  • 1
    $\begingroup$ We could also add the "obtaining particular datasets" to the list of "... because it focuses on programming..."; that way we still have the "obtaining particular datasets" listed somewhere as an explicit reason to close. $\endgroup$ – jbowman Apr 1 '19 at 19:47
  • 1
    $\begingroup$ @jbowman but then we also end up with ambiguous close reason. $\endgroup$ – Tim Apr 2 '19 at 15:00
  • $\begingroup$ @tim - but if you're asking about datasets, you should be able to figure out that the "obtaining particular datasets" close reason is the one that applies to your question, whereas if you're asking "why doesn't this program work?" you should be able to figure out that it's being closed because of the "focuses on... debugging" reason, not the "obtaining particular datasets" reason. $\endgroup$ – jbowman Apr 2 '19 at 15:40
  • $\begingroup$ @jbowman Sympathetic to your aims, but I don't like reasons appearing to overlap and I think we should try to minimise the challenge to the OP of working out which nuance applies to them. $\endgroup$ – Nick Cox Apr 2 '19 at 15:44
  • $\begingroup$ You seem to suggest that we leave both reasons "not statistics" and "datasets", but given that our proposal was declined, this is impossible. Also it would be infeasible to leave the datasets reason that is responsible for 0.5% of close reasons while removing the programming one, that is surely used more commonly. $\endgroup$ – Tim Apr 4 '19 at 21:52
  • 2
    $\begingroup$ No; I am suggesting removing the datasets reason but boosting the advice against asking for datasets in the Help Center. Remove one reason; split the Either/Or reason into two; same number of reasons. $\endgroup$ – Nick Cox Apr 4 '19 at 22:06
  • $\begingroup$ +1. The datasets explanation, wherever it might appear, could use enhancement. Often it is used now for questions that don't need a statistical or ML answer because they are really asking for application-specific advice: that is, they require data and/or knowledge of neurology, genetics, engineering, etc. $\endgroup$ – whuber Feb 26 at 15:24

No, we shouldn't. If 'asking for datasets' close reason has been used, it has been useful (pretty much the same way as 'opinion-based' has, the second least used topic).

However, the idea of splitting the 'not about statistics' and 'pure programming is' good. Despite initially declined, we can try to insist with the SE staff to have a 4th close reason (as many other SE sites already have). Moreover, sooner or later that close reason will be broadened to include other types of shopping list questions here in CV SE (we can observe a shift in how users perceive such types of questions are in course; recent example).

Going either way (removing or not removing, splitting or not splitting), I also suggest we reorder the close reasons in the dialog from top to bottom according to their frequency of use (reported in Extract software-specific default close reason):

  • programming;
  • self-study;
  • not about statistics, machine-learning, etc.*
  • asking datasets;
  • blatantly off-topic (for <3k rep) / other (>=3k rep) .

*I am assuming if split occurs, 'not about statistic' would rank 3rd (but this could be reassessed after some time with real data).

Here is the current dialog order:

enter image description here

The picture also illustrates the dialog has space for adding more topics if we want to.

  • $\begingroup$ (Not mine -1.) Only the three: "self-study", "data" and "software" reasons are editable by us. So your proposal would not work for us, otherwise I agree with the first part. $\endgroup$ – Tim Apr 7 '19 at 6:58
  • 1
    $\begingroup$ The proposal that we have 4 close reasons was declined (as noticed in the question), this is not an option. $\endgroup$ – Tim Apr 8 '19 at 21:05
  • $\begingroup$ Yes, because this thread asks about alternatives to the fact that it was declined. $\endgroup$ – Tim Apr 9 '19 at 4:55
  • 2
    $\begingroup$ I wonder if by authoritative you mean authoritarian. $\endgroup$ – Nick Cox Apr 9 '19 at 7:31

You must log in to answer this question.

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