9
$\begingroup$

As adequately addressed in R-specific StackExchange site or greater integration of R community within CV?, there has been a discussion on whether or not to start an R-specific StackExchange site (an idea I support, since my not-strictly-stats-R-questions aren't R-programming questions per se either, and don't really belong in SO).

The discussion in the linked question took place a year and a half ago, but I cannot see any conclusion. So I am curious whether the community decided on not pursuing the idea of starting an R Stack Exchange site, or if the initiative simply withered away with time.

$\endgroup$
  • 4
    $\begingroup$ Could you provide an example of an R-specific question that would fit neither here nor on SO? $\endgroup$ – whuber May 30 '13 at 19:46
  • 1
    $\begingroup$ Questions on data visualization, for instance (bar plots, histograms, density plots)? That's not a statistics question if you're only wondering about some details on how to implement that in R, but I (maybe wrongly) don't consider that a programming question either. $\endgroup$ – Sverre May 30 '13 at 19:55
  • 1
    $\begingroup$ But more to the point, I believe the mere confusion as to where an R question best belongs is reason enough to consider a separate R site. (But the intent of my question here is only to probe the outcome of the previous debate). $\endgroup$ – Sverre May 30 '13 at 19:56
  • 9
    $\begingroup$ Before going any further, please review our site's faq. In particular, data visualization questions are on topic here--unless they are purely about the details of creating a particular visualization on a particular platform: that's a programming question. There was a lot of confusion in the first year or so, but hundreds of comments and helpful guidance provided by high-rep community members in the last year or so indicate there is a pretty clear consensus on what belongs here and what does not. $\endgroup$ – whuber May 30 '13 at 20:02
  • 2
    $\begingroup$ The related thread at meta.stats.stackexchange.com/questions/1/… also provides helpful guidance concerning this community's thoughts about R-related questions. I do not follow the logic to your recent comment: if there is confusion about where to direct R questions when there are two obvious places for them, how would having three sites that field such questions reduce that confusion? I think it would only increase it (exponentially). I do see some merit in having a site dedicated to R programming, basically ripping it out of SO. $\endgroup$ – whuber May 30 '13 at 21:31
  • 3
    $\begingroup$ I was worried that asking this question would start another debate on R questions. But my question only addresses one thing: what was the outcome of the discussion that the comments and answers to the question I linked to said was being had offline? I couldn't find any information about that. I tried to restate that in my comment above ("The intent of my question here is only ... "). Could we please stay on-topic? $\endgroup$ – Sverre May 31 '13 at 11:44
  • 1
    $\begingroup$ No clear consensus has been reached. The FAQ was created by fiat and was not subject to community consensus. Previously arrived at consensus provides pretty clear guidance *(meta.stats.stackexchange.com/questions/1/). A new consensus clearinghouse question was set up here: meta.stats.stackexchange.com/questions/1577/…. However, in reality, long term, the answer is - whatever active users want and do. $\endgroup$ – russellpierce Jun 14 '13 at 3:20
  • 1
    $\begingroup$ @whuber Although I do not like the idea of fracturing R support across even more SE sites, I think this might be the kind of example you asked for: meta.stats.stackexchange.com/questions/1640/… $\endgroup$ – A.M. Jul 3 '13 at 3:19
9
$\begingroup$

The short answer:

Reality has a status quo bias; the initiative seems to have simply withered away with time. I doubt anyone (or the community as a whole) explicitly decided not to do it. Either way, nothing along those lines came of that discussion, and there isn't an R.SE site listed in Area 51.

Peripheral discussion:

  • I think you're right that there can be R questions that aren't a great fit for either SO or here. One such question might be, 'is there an R package for _____?' (Note this question comes up regularly and seems to be tolerated.) In addition there are questions that are off-topic on all SE sites by design (e.g., open ended discussion questions). However, there are other sites on the internet--the r help listserv and forums, etc. I don't believe that SE should try to expand to the point where it encompasses any question that could be formed in English.

  • I agree with @whuber that having an R specific site in addition to CV and SO would likely lead to more confusion, not less.

  • I personally have never felt any confusion about where a given question belonged unless the question itself was confusing (confusingly worded, for instance).

    For what it's worth, my criterion is simple: what does the OP need explained to them? If someone were to ask 'how do you calculate the average in R?', that doesn't seem like my prototypical conception of programming, but they still need to know how to use R to get something done; an answer that explained what an average is would not help them.

    Here's another way to think about it: R (in part because it's freeware, in part because it's common) is often used to illustrate a point in an answer on CV, but this is simply instrumental to the real answer. If another language could have been used equally well, the question definitely belongs here; if not, it probably belongs on SO.

    Under any rubric there will be some threads that end up on one site that might better fit on the other; so be it. For example there are questions on SO where someone asked why something worked like it did in R, but the problem isn't that they didn't understand R's syntax but that they don't understand the underlying statistical concepts. There are some questions here (especially older questions that are too old to migrate) that really belong on SO.

    The argument is sometimes made that R programming questions with any statistical implications should stay here because a user would need to be statistically savvy to provide an answer that is sensitive to those (perhaps unstated) issues. I don't buy this. If you look at the list of SO users who frequently answer R-taged questions, you will see that these people are quite statistically sophisticated. Moreover, R is hardly used for anything other than data analyses of various sorts; almost anyone who is knowledgeable enough of R to answer SO questions will be familiar with statistics.

    For more information about whether a question should be migrated, you might want to read: What standard should be applied in deciding whether a question requires sufficient 'statistical expertise' to be posted to CV?

$\endgroup$
  • 2
    $\begingroup$ * Please don't take anything in the comments or my answer personally, there is no implied criticism here. $\endgroup$ – gung - Reinstate Monica May 31 '13 at 18:32
6
$\begingroup$

As the original Asker of the META Q you link to, I think "we" (the people on SO who were getting annoyed with questions being closed that seemingly, or at that time, fell between the SO and CV cracks) decided that within the StackExchange ecosystem, what we hoped to achieve with an R-related SE site didn't really fit in with the SE ethos or could be covered within the SO or CV banners (plus some others).

To be honest, I'm now more frustrated with CV and its relationship with R and questions about R than I am with the SO side of things. If you are using R to do stats and have a statistical question about what the code in R is doing (i.e., a function you are using, not something you wrote), then I don't see why that would be OT for CV. Others' opinions do vary.

$\endgroup$
  • 2
    $\begingroup$ I'm sorry to hear you are frustrated. I wonder if you might start a new meta.CV thread to air out the issues you have. Of course, you don't want to get your hopes up & end up being even more upset, b/c there is no guarantee that the consensus will change. Nonetheless, it is possible to revisit the issue, & it might help the site's policies evolve further & in a positive direction (especially if you can clearly outline your vision & link to specific examples, etc.). $\endgroup$ – gung - Reinstate Monica Sep 16 '13 at 23:39
  • 1
    $\begingroup$ Here is one test for whether an R question is on-topic here: if it would interest you at least as much if every occurrence of "R" in the question were replaced by "BMDP," then it's on-topic. $\endgroup$ – whuber Sep 17 '13 at 1:54
  • $\begingroup$ @whuber Yes it would interest me if the statistical aspects were of interest given my background and interests. If the r tag is on a question I will look even if the statistical aspects aren't to my taste because I know that software and may be able to help regardless. I'm not suggesting that all R questions are suitable for Cross Validated, but I don't think questions using R for statistical insight are programming and should be on-topic here. Otherwise, where do you draw the line; do we not mention any stats package here? $\endgroup$ – Reinstate Monica - G. Simpson Sep 17 '13 at 2:45
  • $\begingroup$ @gung I meant "frustrated" more in the sense of not being sure what is and isn't acceptable here than expressing some dissatisfaction with Cross Validated in general. Personally, as a user of statistics I would find a Cross Validated devoid of software-related statistical questions a much less enticing & interesting place. $\endgroup$ – Reinstate Monica - G. Simpson Sep 17 '13 at 2:48
  • 1
    $\begingroup$ Thanks for your response, @Gavin. I certainly don't think that we should "not mention any stats package here". If you were to read through my answers (& stay awake), you'd see that I often illustrate points w/ R code. The point is the R aspect is instrumental. I could have used SAS or Python, etc. If someone understands the statistical issues & only wants to know about how R works or how to get something done in R, then it's off-topic, IMPO. If you want, these issues might be best worked out in their own thread, but I'm also happy to discuss here via comments, whichever works best for you. $\endgroup$ – gung - Reinstate Monica Sep 17 '13 at 3:23

You must log in to answer this question.

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