9
$\begingroup$

I haven't been reviewing long, but it seems like almost daily I see questions closed due to programming or data requests. But to me the majority of these are more often statistical questions rather than programming questions (which statistics often is going to involve I'm afraid). A good example is this question:

enter image description here

To me this is not about programming. The OP is clearly asking about whether or not the structure of their clusters in the mixed model is correctly defined and how to program this. This to me doesn't scream "I need somebody to debug code" but rather "I need to understand how hierarchical structures work in these models." Oftentimes there are questions I pose here about GAMs that almost inevitably require some programming feedback on because that is the nature of the beast.

This is only one example, but can someone here clearly define what is considered programming-only and what is considered CV-worthy? I am more often leaned towards keeping these kinds of questions open, but not sure if that is correct.

Edit 1:

Another example recently that gets to the heart of my issue.

https://stats.stackexchange.com/review/close/344885

Edit 2:

One more that doesn't seem to fit this specific close vote:

https://stats.stackexchange.com/review/close/344919

And another that I would consider borderline given it's not clear if it's a programming question or stats question (as it's currently phrased, this could be a contender for a close given they really only highlight the error code).

https://stats.stackexchange.com/review/close/344946

$\endgroup$
28
  • 2
    $\begingroup$ Please vote to keep open on such questions! $\endgroup$ Commented Apr 12, 2023 at 13:08
  • 5
    $\begingroup$ I agree with you as I have also seen an increasing number of these recently. I have a feeling I voted to close your example as needing details or clarity. In general if a programmer with minimal knowledge of statistics could answer it then close it but if there is a statistical issue then leave it open. You can always add a comment if you think it is in danger of being closed erroneously. $\endgroup$
    – mdewey
    Commented Apr 12, 2023 at 13:37
  • 9
    $\begingroup$ One useful test that is often used on this site is "Does answering the question require statistical knowledge?" If it does, then the question is on-topic. (This test isn't perfect, but gets to the core of the on-topic/off-topic distinction.) $\endgroup$
    – Sycorax Mod
    Commented Apr 12, 2023 at 13:42
  • 2
    $\begingroup$ That heuristic is pretty close to the mental operation I go through to decide on keeping this open or not @Sycorax. There will always be borderline cases, but I find examples like these are more clear in my opinion. $\endgroup$ Commented Apr 12, 2023 at 14:06
  • 2
    $\begingroup$ I vet properly on questions like these which are based on programs and even then I err quite sometimes but AFAIR I voted to close it on grounds of clarity and not due to the question being program oriented. $\endgroup$ Commented Apr 12, 2023 at 14:20
  • 1
    $\begingroup$ How would a (new) user know what kind of details they would need to add to improve clarity? $\endgroup$
    – dipetkov
    Commented Apr 12, 2023 at 15:18
  • 4
    $\begingroup$ I agree that we seem to be a little too trigger-happy about questions like this one, and posted a similar question a while back, which I now link to in a boilerplate comment aimed at close-voters: "Close-voters: I believe this question is primarily about statistics, not programming, and that it should therefore stay open/be reopened. See Closing "software questions"." $\endgroup$ Commented Apr 12, 2023 at 15:27
  • 8
    $\begingroup$ The principle is clear -- is statistical knowledge needed? -- but the fact that experienced members here can disagree in good faith shows that there are always cases in which different people jump different ways. Perhaps unfairly I sense that questions phrased in terms of R or Python fare better than some others. True or not, a question to the voters to keep might be: Would you vote to keep even if the language or more generally software concerned is not at all one you use yourself, and even one you personally deprecate? $\endgroup$
    – Nick Cox
    Commented Apr 12, 2023 at 23:11
  • 2
    $\begingroup$ That Q seems like a little bit of a gray zone to me. I would not have voted to close as a programming question, but I can imagine someone might based on the way it's phrased. The way I decide if a Q is off topic is I ask myself, 'what needs to be explained here'. I imagine the answer, & if it's 'in lavaan use function( , argument=TRUE), then it's off topic. In this case, the explicit question is whether the data have 3 levels and what analysis can address mediation in that case. Those are statistical questions. $\endgroup$ Commented Apr 13, 2023 at 11:48
  • 1
    $\begingroup$ I voted to close it because it asks too many things. $\endgroup$
    – utobi
    Commented Apr 15, 2023 at 5:18
  • 2
    $\begingroup$ I think the tone of some exchanges is a little sharp. Please remember: 1. No one can vote to close without experience and especially enough reputation. 2. When three non-moderators agree on voting to close, that's explicitly not an isolated opinion. 3. If a moderator is voting to close that is #1 powered up. Let's phrase disagreements as "I respectfully disagree with this decision because..." not "This decision is quite wrong. Don't vote this way". (These aren't intended as quotations, just stylised examples of different tones.) $\endgroup$
    – Nick Cox
    Commented Apr 15, 2023 at 8:40
  • 2
    $\begingroup$ (Also, please remember that there can be different grounds for closure, and being a programming question might be one of several: cf. @utobi's last point.) $\endgroup$
    – Nick Cox
    Commented Apr 15, 2023 at 8:40
  • 1
    $\begingroup$ @utobi yes this is why i explicitly say "for programming", but the other reasons are still valid. I think your close vote makes more sense given how much is being asked. $\endgroup$ Commented Apr 15, 2023 at 13:22
  • 2
    $\begingroup$ @Nick reminds me of an excellent criterion I use when making VTC decisions related to programming: "if I were to change the reference in this question from software platform X to platform Y, would that change my opinion?" If not, this is a statistical question. If yes, I have to investigate further, because there (IMHO) is the true gray area. It covers questions concerning discrepancies between platforms, about the uses and meanings of arguments to statistical functions, and related issues. I think this community tends to close such questions more often than we should. $\endgroup$
    – whuber Mod
    Commented Apr 19, 2023 at 13:32
  • 1
    $\begingroup$ As for the "data request" aspect, there are probably less grey area situations compared to programming. The obvious response is generally to redirect the question to opendata.stackexchange.com. However, my experience has been that compared to CV, opendata.SE suffers from a lack of active users. So just an aside to say that if you have some domain knowledge relative to datasets and whatnot, having a look once in a while at opendata.SE would probably be helpful, and might encourage people to ask their data request questions there rather than on CV. $\endgroup$
    – J-J-J
    Commented May 6, 2023 at 11:49

0

You must log in to answer this question.

Browse other questions tagged .