10
$\begingroup$

This is a follow up to Should we retire the "data set request" closure reason?

My interpretation of the votes and answers on that thread is that there is some support for retiring the "data set request" option (+8/-0), but there is also a feeling that there should be a concrete option that will replace the "data set request" option (+6/-0). I want to put forward this post as a suggestion for how we could use the slot freed up if we choose to retire the "data set request" option.

Right now, we have two distinct closure reasons occupying a single slot:

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.

My suggestion is that we divide this into two distinct reasons:

  1. This question appears to be off-topic because it is not about probability, statistics, machine learning, data analysis, data mining, or data visualization.

  2. 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.

I think this is a helpful change because it provides question-askers with more precise feedback about what caused a question to be closed. While it might be obvious to experienced users why a given question was closed, I think that it can be confusing to newcomers when a question is closed with the EITHER/OR closure message because the newcomer is not clear on the boundaries for reasons (1) and the boundaries for reason (2). This means that the newcomer won't be certain if the question was closed because it's not about statistics (broadly conceived), or if was closed because it's about routine programming tasks.

Newcomers are the users we should have foremost in our minds when writing closure reasons because they're the users whose questions are most likely to be closed, and they're the users who have the most information to learn about how the site works. By definition, they're the least-experienced users.

By contrast, the more experienced users who are doing the closing have a clear reason fixed in their minds (either they're closing it for reason 1, or they're closing it for reason 2), but the limitations of the closure dialogue make that communication difficult because the new user is only told that the question was closed "ETIHER for reason 1 OR for reason 2." This means that the new user has to learn two things about how the site works (which quantitative topics are on-topic here, and which computing topics are on-topic here) in order to revise their question for reopening. But it's likely that only one of these reasons caused the question to be closed, so they have to learn the details of the 2 reasons just to ameliorate the single problem. I think this is an obstacle to new users engaging successfully with the SE paradigm.

$\endgroup$
12
  • 2
    $\begingroup$ At 1. You could add probability. $\endgroup$ Mar 8 at 1:39
  • $\begingroup$ @kjetilbhalvorsen Good point! Splitting up the closure reasons means we're not as tightly limited in character count. $\endgroup$
    – Sycorax Mod
    Mar 8 at 1:42
  • $\begingroup$ Thanks for asking this. Having a concrete alternative makes a lot more sense than just getting rid of dsr. That said, I'm not sure if this is better. Is there a positive case for this option? Eg, if we were granted an extra close option, is this what we would do with it? How much is this believed to help? What is the situation where having these would be good, but the status quo would be inferior? $\endgroup$ Mar 8 at 13:51
  • $\begingroup$ @gung-ReinstateMonica I think it's clear that we will not be receiving an additional closure reason right now, for the reasons that Catija laid out. This is addressed in the previous thread. I think the only path to getting an additional closure reason (brining the total to 4 custom reasons) is if we (i) retire "data set request" and (ii) use the freed slot and (iii) demonstrate a credible need for an additional reason. I don't think we have a need for an additional reason right now, but I do think that clarifying the EITHER/OR option would be helpful, as I've elaborated in my revision. $\endgroup$
    – Sycorax Mod
    Mar 8 at 19:58
  • $\begingroup$ @gung-ReinstateMonica That said, I'm not married to dividing the EITHER/OR reason into two. If there's another suggestion that makes more sense, I'm open to the suggestion and think we should make a thoughtful comparison of the alternatives. $\endgroup$
    – Sycorax Mod
    Mar 8 at 20:09
  • $\begingroup$ I know we aren't getting another closure reason, I'm asking hypothetically to help frame the what the positive case for this change it. At any rate, the revision does provide a coherent positive case for the proposed change. +1, I'm OK w/ the chage $\endgroup$ Mar 8 at 20:23
  • $\begingroup$ I like the idea. Which closure reason would we use for requests for data sets? I would assume 1, but then again, finding good data is "about statistics" in a sense. So would it make sense to add ", or it requests a specific data set (you may want to look at OpenData.SE)" to the first reason? Or is that again too cumbersome? $\endgroup$ Mar 14 at 7:58
  • 1
    $\begingroup$ @StephanKolassa When I had first proposed this, I had assumed that closure with the "Other - leave a comment" reason would be an acceptable replacement; we could even have a meta thread (perhaps even this one) that has a boilerplate "data set request" reason that a person could copy/paste into the Other - leave a comment space. But I can see an argument for appending an instruction about data set requests to the "not about statistics" closure reason. $\endgroup$
    – Sycorax Mod
    Mar 14 at 17:48
  • 1
    $\begingroup$ @StephanKolassa How would you feel about putting "data set request" with (2) instead of with (1)? See also: the comment thread below my answer. $\endgroup$
    – Sycorax Mod
    Mar 22 at 17:41
  • 1
    $\begingroup$ I could live with putting it under (2). $\endgroup$ Mar 22 at 18:33
  • $\begingroup$ I see we have new closure reasons. Good! Unfortunately, the relevant reason now reads "Questions about how to obtain data sets are not on topic at stats.SE, but are on-topic on the Open Data site." Can we be a little more explicit than "the Open Data site", e.g., explicitly say "OpenData.SE" and link there? $\endgroup$ Apr 5 at 19:49
  • $\begingroup$ @StephanKolassa The Open Data SE link is provided in the "post owner guidance." $\endgroup$
    – Sycorax Mod
    Apr 6 at 15:06
5
$\begingroup$

After reading the comments, my revised suggestion is that we (a) retire the "data set request" closure reason and (b) divide the EITHER/OR reason into two distinct reasons:

  1. This question appears to be off-topic because it is not about probability, statistics, machine learning, data analysis, data mining, or data visualization.

  2. This question appears to be off-topic because it focuses on programming, debugging, or performing routine operations, or it asks about obtaining specific datasets. You could try the support links we maintain or the Open Data site instead.

$\endgroup$
7
  • 1
    $\begingroup$ In (2), "within a statistical computing platform" may be overly narrow, because frequently the question refers to Excel, a database platform, or something not devoted to statistics. Also, do you plan on including a link to the list of support links? $\endgroup$
    – whuber Mod
    Mar 22 at 17:21
  • $\begingroup$ @whuber Thanks for the feedback! How would you feel about removing "statistical" from "statistical computing platform"? I do plan on including links -- they just weren't copied over when I copy-pasted the text. I'll revise to fix the hyperlinks. $\endgroup$
    – Sycorax Mod
    Mar 22 at 17:32
  • 1
    $\begingroup$ Brevity is the key to writing VTC text, so why not omit the phrase "within a statistical computing platform" altogether? When that's done, it could leave room to include the dataset VTC reason as an alternative. That might make a little more sense (given we're trying to lump three reasons into two slots) because "software-only" and "data-only" are somewhat related reasons. $\endgroup$
    – whuber Mod
    Mar 22 at 17:36
  • 1
    $\begingroup$ I see what you're saying. In your suggestion, option (1) says "This is blatantly off-optic," while option (2) says "This is statistics-adjacent, but still off-topic; here are some potential resources." $\endgroup$
    – Sycorax Mod
    Mar 22 at 17:41
  • 2
    $\begingroup$ @whuber I've reworked the closure reasons to reflect your feedback & Stephan's input. LMK what you think. $\endgroup$
    – Sycorax Mod
    Mar 22 at 18:41
  • $\begingroup$ It's not entirely obvious that the context "performing routine operations" is "with analytical software". Is there space for that? Or for "routine use of software"? $\endgroup$ Mar 23 at 8:48
  • 1
    $\begingroup$ @Scortchi-ReinstateMonica The honest answer is 'I don't know'. I've looked into doing a "dry run," but it seems I'd have to deactivate an existing reason, then type the comment, to see if it's too long. I don't really want to do an experiment in the "live system" until I have to make the change for real, and I won't do that until April (to allow for more discussion). But when I make the change, I'll see if I can fit more detail into the allotted space. $\endgroup$
    – Sycorax Mod
    Mar 23 at 13:51

You must log in to answer this question.

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