5
$\begingroup$

When a senior CV participant comments, “CV is a statistical blog,” what does this mean for the community? How is this to be interpreted? A literal interpretation might suggest that only those questions with an unambiguously theoretical statistical stance are appropriate. Obviously, this isn’t true as there are many queries being accepted with concerns that are applied or business related, with a minimal theoretical component. Not only that, given the apparent level of many OPs' statistical literacy, a purely theoretical response might provide an answer but would it be appropriate if it weren’t understood?

On the face of it, it would appear that many CV participants have a clear idea of the kinds of questions and issues CV should handle. For instance, questions that are mostly concerned with programming and/or specific software packages are usually relegated, e.g., to Stack Overflow -- but by no means all. Many, many questions get through that are mostly about statistical software, its implementation and idiosyncrasies. Responses, too, will include programming tips. Similarly, questions that may have a quantitative component but are more heavily content weighted towards specific academic silos such as economics or finance have a higher probability of getting reposted to one of those SE groups. By the same token, inadequately phrased and/or naive questions have a much greater likelihood of being put on hold regardless of their statistical content or value. And so on. The point is that, while there is an appearance of community clarity in terms of the questions that are accepted (or not), the fact is that there can be considerable ambiguity and leeway in these community decisions.

So, would questions concerned with the history of statistics be appropriate? For instance, what about questions taken from a book like Crosby’s The Measure of Reality: Quantification and Western Society 1250-1600? Is there a clear line between statistics, machine learning, AI or information theory? Are wider controversies such as ongoing debates about the “scientific method” appropriate? How about questions concerned with the overlap and triangulation of qualitative (idiothetic) and quantitative (nomothetic) methodologies? What about issues crossing the blurred line between philosophy and statistics? Deborah Mayo’s blog https://errorstatistics.com/ is an example of this. Would cultural anthropological issues as embodied in a book like Chrisomalis’ Numerical Notation, which reviews the earliest cultural origins and evolution of numeracy, be appropriate?

Just wondering if there is any clarity in the community on these things as they're not clear to me.

$\endgroup$
9
  • 9
    $\begingroup$ Can you provide a link for your quote, so we can get context? $\endgroup$
    – Glen_b Mod
    Commented Apr 9, 2017 at 22:14
  • 15
    $\begingroup$ I'd say in relation to the opening sentence, the statement cannot be factually correct - CV is demonstrably not a blog, it's a question and answer site by design, and blog-like posts are off-topic (such posts are not in general a question followed by some answers that may be posted by anyone). Even more telling, we actually had a blog for our site (if our site was a blog already why would it then have a blog?), but it only got two or three posts in my entire time that I was actively using the site. All on-site SE blogs were recently shut down (one or two SE sites maintain offsite blogs) $\endgroup$
    – Glen_b Mod
    Commented Apr 9, 2017 at 22:19
  • 1
    $\begingroup$ I think a better title would be: "CV is not a statistical blog" and would make more sense in the context of Q's first paragraph. $\endgroup$
    – Andre Silva
    Commented Apr 10, 2017 at 18:35
  • 1
    $\begingroup$ I edited the title, trying to make it more explicit. Feel free to edit it further. (cc to @AndreSilva) $\endgroup$
    – amoeba
    Commented Apr 11, 2017 at 11:37
  • 4
    $\begingroup$ My gripe with the quoted comment (without knowing any other context) is that stackexchanges are not blogs any more than wiki's are blogs, or search engines are blogs. CV is a question and answer site, and specifically a stackexchange which is a very particular kind of internet resource. CV is not a blog, statistical or otherwise. :) $\endgroup$
    – Alexis
    Commented Apr 14, 2017 at 2:20
  • $\begingroup$ @glen_b The "senior participant" in question has participated on this specific thread but has either forgotten making this statement or prefers not to be identified with it. Either eay, this makes me reluctant to reveal who is responsible. Wrt whether or not CV is a "blog," clearly the community does not agree with this choice of words. I want to note that the wording is not mine, it's the anonymous senior participant's. $\endgroup$
    – user78229
    Commented Apr 14, 2017 at 14:32
  • 2
    $\begingroup$ Perhaps it was me before I knew what a blog was. Anyway, it was doubtless a slip - what "statistical" encompasses is the interesting bit. $\endgroup$
    – Scortchi Mod
    Commented Apr 14, 2017 at 14:45
  • 3
    $\begingroup$ @DJohnson without context, you may well be doing whoever it is a disservice, by quoting out of context. I suggest that perhaps your question can manage without a quote you won't provide context for. Just ask whatever you want to know without it. $\endgroup$
    – Glen_b Mod
    Commented Apr 14, 2017 at 14:50
  • $\begingroup$ @Scortchi Nope. It wasn't you. $\endgroup$
    – user78229
    Commented Apr 16, 2017 at 19:50

2 Answers 2

Reset to default
11
$\begingroup$

We've had statistical questions concerning applications from Astrology to Zoology—these kind of "practical, answerable questions based on actual problems" are our site's bread & butter (to my mind "questions with an unambiguously theoretical statistical stance" are a garnish). It's not usually difficult to apply @gung's criterion—"What does the OP need explained?"—to decide if a question's really about the economics or whatever rather than about the statistics, & we migrate relatively few questions on these grounds. Some examples follow:

I think most users will agree that the facts and concepts needed to answer these belong to Economics, Physics, & Finance.

History, philosophy, & cultural anthropology of statistics are on-topic, as is the role of statistical analysis in the scientific method. That is, questions in these areas have been treated as on-topic, though there's not been much explicit discussion. I'd certainly not want to rule them out, but I would remind askers of the SE dictum that "real questions have answers": CV isn't a blog; nor is it a discussion forum. (I'd also note that some of the examples of topics you give seem to fall into more general philosophy &c. of Science or Mathematics rather than specifically of Statistics, but if you swapped e.g. Crosby’s The Measure of Reality for Hacking's The Emergence of Probability I'd answer "on-topic".)

Most migrated (not "relegated"!) questions go to Stack Overflow, & many closed questions are closed because they "focus on programming, debugging, or performing routine operations within a statistical computing platform". A tricky issue here is whether a latent statistical question can be discerned in a post that prima facie doesn't ask one. More inconsistency than we'd like in their treatment seems inevitable; but a hard-line policy of insisting on statistical questions asked in English (together with mathematical notation) would put CV's help beyond the reach of people having great need of it.

$\endgroup$
4
  • 5
    $\begingroup$ +1. I'd underscore that part of the rationale for software questions belonging elsewhere is historical, namely that SO existed before CV and was handling software questions already. It's a moot point whether -- were we to start again from fresh -- the same decisions would be made. It's not surprising that many new members expect or would prefer a one-stop shop covering the range from statistics to statistical software. A big negative is that few things are so uninteresting as questions on using software you don't use yourself. Until the whole world uses R, or whatever, that will remain true. $\endgroup$
    – Nick Cox
    Commented Apr 10, 2017 at 12:52
  • 2
    $\begingroup$ @NickCox: I think that historical accident is a happy one: it'd be harder to maintain the distinction between statistical & software questions if both could be asked on the same site; & it's worth maintaining. $\endgroup$
    – Scortchi Mod
    Commented Apr 10, 2017 at 15:05
  • 1
    $\begingroup$ I tend to agree personally but every day we close questions based on not understanding this. $\endgroup$
    – Nick Cox
    Commented Apr 10, 2017 at 15:06
  • 2
    $\begingroup$ @NickCox: True, but mitigated by the fact that the obvious software ones are a quick decision, & the ambiguous ones would need clarifying anyway. $\endgroup$
    – Scortchi Mod
    Commented Apr 10, 2017 at 15:28
9
$\begingroup$

I've never seen anyone put "CV is a statistical blog" exactly, although I may have missed it. I often put something like "CV is a Q&A site for questions about statistics (machine learning, etc)" in a comment to a thread that is about something like 'how to convert data from wide to long in R', which is about software / coding rather than statistics. I may sometimes shorten that to "... about statistics" (without the parenthetical expansion) when the context suggests to me that it may be acceptable shorthand. I certainly do not mean that machine learning, data mining, or data visualization are off topic here; I simply take them to be assumed or that it is unnecessary to state them in those cases.

Regarding what is on topic here, that is an issue that has been discussed extensively several times on meta.CV (you can read through some of the threads tagged under on-topic, e.g.). For what it's worth, my own criterion has always been straightforward: What does the OP need explained? If it is a statistical (machine learning, etc.) concept, it belongs here, if not, it belongs elsewhere. A statistical concept need not be theoretical (depending on how strictly one wants to define that term).

The biggest issue concerns questions about software use, particularly coding. That is already on topic on Stack Overflow, where questions get good answers very fast. From the point of view of the SE system, it isn't best to have questions equally on topic on multiple sites. That leads to cross-posting, among other problems, and is analogous to the problem with duplicate threads (cf., here). To some extent, that will always exist, as the sites are living things, but where it can be avoided, it probably should.

As far as your other topics, the history of statistics has long been considered on topic here, and philosophy questions are often fine. It is certainly true that what is on topic or not will never be perfectly consistently enforced because the system is managed by the collective efforts of many different people. It is also true that in the past the boundaries were less clear and somewhat in flux. On occasion, old threads that should be off topic are closed, or sometimes are 'locked' to preserve their historical value.

$\endgroup$
8
  • 4
    $\begingroup$ +1 I fully agree. I would also note that in last year or two we have done a much better work of moving "How to do this in R" questions in SE. Especially some earlier (<2012) questions (eg. here and here) where almost R exclusive while now, if any questions remain usually they have clear statistical context. This is important because it helps shape the profile of the site as "Statistics" rather than a "Coding" help community. $\endgroup$
    – usεr11852
    Commented Apr 9, 2017 at 20:30
  • 2
    $\begingroup$ (+1) Not sure the philosophy of statistics is such a grey area: See Add Philosophy as a possible SE network to migrate off-topic questions, & @whuber's answer:-"The meaning and interpretation of the terms and concepts we use are of interest to this community and are on topic here". $\endgroup$
    – Scortchi Mod
    Commented Apr 10, 2017 at 8:13
  • $\begingroup$ @usεr11852 This is a digression, but I'd say if one comes across an old thread about coding [in R] that would be closed as off-topic (or migrated) according to our today's standards, one should vote to close as off-topic. These old threads, when they are open, convey a wrong message about what questions are appropriate here. And there is little harm in closing them now. $\endgroup$
    – amoeba
    Commented Apr 10, 2017 at 8:47
  • $\begingroup$ @Scortchi, perhaps that came off too strong. I tweaked the phrasing. $\endgroup$
    – gung - Reinstate Monica
    Commented Apr 10, 2017 at 12:15
  • 1
    $\begingroup$ @amoeba Instead of voting to close old off-topic questions, please consider flagging them for moderator intervention: we can "lock" them, which has the desired effect but also includes a prominent explanation why. $\endgroup$
    – whuber Mod
    Commented Apr 17, 2017 at 16:11
  • $\begingroup$ @whuber Are you talking about "historical lock" (meta.stackexchange.com/questions/126587)? So the message in this case will be "This question exists because it has historical significance, but it is not considered a good, on-topic question for this site, so please do not use it as evidence that you can ask similar questions here." (I copied if from the linked meta.SE thread)? I was sometimes hesitating to go ahead with this option when I thought that a Q does not have "historical significance". This sounds to me as a high bar. Am I misinterpreting it? $\endgroup$
    – amoeba
    Commented Apr 17, 2017 at 16:37
  • $\begingroup$ @amoeba Yes, that was what I was referring to. "Historical significance" may sound glorious, but all it really means is that the question was once considered on-topic (as evidenced by the fact it has stood open for a relatively long time) but we want to make it clear that it no longer is. This message doesn't carry implications about how we value the thread, except insofar that the fact we are willing not to delete it altogether implies it bears some relevance to the site. $\endgroup$
    – whuber Mod
    Commented Apr 17, 2017 at 16:43
  • $\begingroup$ @whuber All right. To be honest, I don't necessarily see advantages of "historical lock" over good old closing as off-topic, but if you say that it's preferable, then okay. $\endgroup$
    – amoeba
    Commented Apr 17, 2017 at 19:25

You must log in to answer this question.

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