11
$\begingroup$

Almost four years ago there was a discussion on Meta that lead to formulating the "not about statistics" default close reason description as following:

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.

I think that we should discuss this close reason again. My feeling is that this close reason is unclear for some users because of the "... EITHER ... OR ..." formulation. I've seen people arguing "but it relates to statistics" in case of software questions, where they read "OR" as "AND" and the decision was unclear for them.

Moreover, we have a large number of software questions, including questions asking for code review. It often seems to be confusing for users that the question needs to be supplemented with reproducible example for us to consider migrating it to StackOverflow.com.

I would suggest that we extracted from the "not about statistics" the strictly "software" reason that mentions:

  1. software/programming is off-topic on our site,
  2. code review is off-topic on the whole SE network,
  3. questions without reproducible example are off-topic on the whole SE network.

My suggestion is to split the current close reason into two reasons like (I don't insist on such wording):

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

and

This question appears to be off-topic because it asks for code review, or focuses on programming, debugging, or performing routine operations within a statistical computing platform. Programming questions with a reproducible example should be posted on StackOverflow.com. For help with such questions try the support links we maintain.

The proposal was declined

The follow up thread can be found here.

$\endgroup$
11
  • 5
    $\begingroup$ NB We squished 4 custom close reasons into 3 because that's the most we were allowed: now SE can give us an extra one if we show them the need for it. $\endgroup$
    – Scortchi Mod
    Commented Jan 29, 2019 at 11:35
  • 3
    $\begingroup$ I am sympathetic to rewriting this, as it is so often misunderstood. But note that there is a Code Review site on SE. $\endgroup$
    – Nick Cox
    Commented Jan 29, 2019 at 11:46
  • 1
    $\begingroup$ @NickCox yes, but it focuses not on checking code with bugs, but improving working code codereview.stackexchange.com/help/on-topic , so "my code doesn't work" would be off-topic there as well. $\endgroup$
    – Tim Mod
    Commented Jan 29, 2019 at 11:50
  • 3
    $\begingroup$ Still true that "code review is off-topic on the whole SE network" is not the case. $\endgroup$
    – Nick Cox
    Commented Jan 29, 2019 at 11:51
  • $\begingroup$ @Scortchi If we can get one more custom close reason, perhaps 'abandoned by OP after request for clarification was ignored' would be more useful? I encounter that quite often and it does not fall into any existing category. $\endgroup$
    – mkt
    Commented Feb 7, 2019 at 10:48
  • 1
    $\begingroup$ @mkt: If that clarification's needed to answer the question, it should be closed as "unclear what you're asking"; if not, why close it at all? To put it another way, whether a question's clear enough to be answered doesn't depend on whether or not specific clarifications have been requested. Anyway, I believe we would be allowed yet another custom close reason, if we could convince SE we need it. $\endgroup$
    – Scortchi Mod
    Commented Feb 7, 2019 at 11:05
  • $\begingroup$ @Scortchi Hmmm, you're right. I thought that the relevant comments do not appear in the VTC queue. But I am now seeing comments when reviewing, so I guess I was wrong about that. $\endgroup$
    – mkt
    Commented Feb 8, 2019 at 7:28
  • $\begingroup$ @Tim Re your update: Not sure why you say that we did not reach any consensus. There is only one answer there and is has 9 upvotes. Looks like a consensus? $\endgroup$
    – amoeba
    Commented Apr 4, 2019 at 21:32
  • $\begingroup$ @amoeba Nick suggests leaving both "not about statistics" and "datasets" reasons, and we can't do this given that our proposal was dismissed. So even though we agree on that, this can't be done. $\endgroup$
    – Tim Mod
    Commented Apr 4, 2019 at 21:51
  • 1
    $\begingroup$ @Tim No he does not. "That was a useful reason to close, but sacrificing that reason seems practical." $\endgroup$
    – amoeba
    Commented Apr 4, 2019 at 21:53
  • $\begingroup$ @amoeba I must have not read carefully enough, sorry. $\endgroup$
    – Tim Mod
    Commented Apr 4, 2019 at 21:55

2 Answers 2

7
$\begingroup$

Certainly it'd be good to split this close reason into two. How about "particular software" rather than "such questions"?

This question appears to be off-topic because it asks for code review, or it focuses on programming, debugging, or performing routine operations within a statistical computing platform. Programming questions with a reproducible example should be posted on StackOverflow.com. For help with particular software try the support links we maintain.


But this text has 428 characters (including the URLs), & we're limited to 400:

This question appears to be off-topic because it asks for code review, or it focuses on programming, debugging, or performing routine operations within a statistical computing platform. Programming questions [with a reproducible example](https://stackoverflow.com/help/mcve) should be posted on StackOverflow.com. For help with particular software try the [support links we maintain](https://stats.meta.stackexchange.com/q/793).

So we have to trim it down a bit, unless anyone knows a way to compress the URLs.

How about "routine use of statistical software" rather than "performing routine operations within a statistical computing platform":

This question appears to be off-topic because it asks for code review, or it focuses on programming, debugging, or routine use of statistical software. Programming questions with a reproducible example should be posted on StackOverflow.com. For help with particular software try the support links we maintain.

That gives 394 characters:

This question appears to be off-topic because it asks for code review, or it focuses on programming, debugging, or routine use of statistical software. Programming questions [with a reproducible example](https://stackoverflow.com/help/mcve) should be posted on StackOverflow.com. For help with particular software try the [support links we maintain](https://stats.meta.stackexchange.com/q/793).

So we'll ask community managers for this additional close reason in a week or so, unless anyone can think of further improvements to the text.

$\endgroup$
4
  • 1
    $\begingroup$ +1 It would be nice if we could work in error messages, package / software requests as well. (I suppose we cold say "debugging" covers error messages, but it isn't clear that people recognize that.) Also, how many characters is this & how many do we have? $\endgroup$ Commented Jan 29, 2019 at 14:25
  • $\begingroup$ @gung: Oh! ... 428 characters with the URLs. As far as I know we still only have 400 characters. [I just checked - "too long by 28 characters".] $\endgroup$
    – Scortchi Mod
    Commented Jan 29, 2019 at 14:36
  • 2
    $\begingroup$ Maybe "performing routine operations within a statistical computing platform" $\to$ "using statistical software" to reduce chars count? $\endgroup$
    – Tim Mod
    Commented Jan 29, 2019 at 14:52
  • 4
    $\begingroup$ @Tim: Good, that's 385. Or "routine use of statistical software" gives 394 - I like having "routine" in there. $\endgroup$
    – Scortchi Mod
    Commented Jan 29, 2019 at 15:03
5
$\begingroup$

I can definitely see that y'all would benefit from this close reason. It seems that you've shoehorned two close reasons into one and it's your second most-used close reason currently, accounting for almost a third of all closed questions in the last 90 days (requires 10k rep to view). For everyone's benefit, the current 90-day numbers look like this:

Reason                Qty.    %/total
---------------------------------------
unclear               779     32.30 %
stats/programming     739     30.64 %
duplicate             380     15.75 %
self-study            170      7.05 %
too broad             166      6.88 %
another site           88      3.65 %
other                  25      1.04 %
opinion-based          23      0.95 %
datasets               12      0.50 %

Splitting this into two close reasons is an excellent idea as it allows you to give more guidance to users in each specific case. Helping users understand why their question is closed and guiding them to resources that may lead to the answer they need if the question is truly off topic or how to improve the question if it's not is an excellent thing.

We do generally recommend the format

Questions about X are off-topic because Y. For more information, see [Z](link to meta post).

As such, I suggest a rephrasing:

Questions asking for code review, or that are focused on programming, debugging, or routine use of statistical software are off topic. Programming questions with a reproducible example should be posted on StackOverflow.com. For help with particular software try the support links we maintain.

Questions asking for code review, or that are focused on programming, debugging, or routine use of statistical software are off topic. Programming questions [with a reproducible example](https://stackoverflow.com/help/mcve) should be posted on StackOverflow.com. For help with particular software try the [support links we maintain](https://stats.meta.stackexchange.com/q/793).

Fitting in "because Y" here isn't likely going to happen due to length but I'm not sure it's necessary in this case. You have a good amount of help available to users who have their questions closed with this reason.


So, considering this, you may wonder why I'm declining this request.

It's because of that last line item in the chart above. Half of one percent of closed questions on this site - 12 of them in the last 90 days - use that custom close reason. That's really not enough from our point of view to justify a fourth close reason.

That said, I wonder if you could rework your more generic "this isn't about statistics" close reason to include this somehow. I'm not sure what sorts of questions it tends to catch but it reads like the default "blatantly off topic" close reason

This question does not appear to be about statistics within the scope defined in the help center.

There's another close reason style that may be a better fit here. Here's an example of how it's being used on Life Hacks:

Does not seem to need a life hack — A "life hack" is a seemingly intractable problem that can be solved by thinking outside the box. Unfortunately, everyday "How to…" questions about learning a craft or new skill are outside the scope of this site. See about Lifehacks. If the author can show how this needs an "outside the box" solution, edit and 'flag' to reopen.

Y'all will be better able to judge whether that's possible but I could see something like this working:

Questions on Cross Validated should be about statistics, machine learning, data analysis, data mining, or data visualization but not about datasets. This question does not seem to meet our scope. These questions may be welcome on other sites, see the help center for more information.

This is a much broader close reason than the Lifehacks one, so it's difficult to make it perfectly line up ... but it really seems to be a catch-all. If there's a specific reason most of these questions are closed other than being blatantly off topic, that may be worth adding. There's 67 characters remaining and rewording could increase that count.

One last note, you might want to update your on topic page. It mentions that datasets are welcome on GIS in some cases but says nothing about Open Data! Don't forget to add it and link to their on topic page to make certain that anyone visiting can see what sorts of questions are in scope.

$\endgroup$
1

You must log in to answer this question.

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