2
$\begingroup$

I have already flagged this question on StackOverflow, but I am not sure if a flag on that site gets anyone's attention on this site.

I can't get the question's original (CrossValidated) link now, but this is the question's new (StackOverflow) link: https://stackoverflow.com/questions/17373838/how-can-rcmdr-r-commander-be-used-for-within-subject-aka-repeated-measures-e

The question is about use of an R GUI, not about programming at all (even in R). It is clearly unfit for StackOverflow, and I think it should be moved back to CrossValidated.

It is in the same vein as these CrossValidated questions: https://stats.stackexchange.com/questions/24304/r-versus-r-studio-output-differences Good GUI for R suitable for a beginner wanting to learn programming in R? Suggested R packages for frontier estimation or segmentation of hyperspectral images ...which deal with the use of statistical analysis software.

$\endgroup$
16
  • 6
    $\begingroup$ The CV question is here. If a question is of the type "which buttons do I press to achieve X" rather than "what are the statistical issues with X", then it's arguably programming-related even if it doesn't generate code as such. Sometimes the boundary is very hard to discern and on occasion posts I think would be okay to stay are migrated. On the whole I think it's mostly working fairly well. $\endgroup$
    – Glen_b
    Jun 29, 2013 at 3:40
  • $\begingroup$ I understand to a point. By that logic I could almost understand a migration...but to SuperUser, definitely not StackOverflow (again...programming not involved at all). $\endgroup$
    – A.M.
    Jun 29, 2013 at 5:07
  • $\begingroup$ Note, for comparison, this similar CrossValidated question: stats.stackexchange.com/questions/13784/… It is essentially the same as my question, except that it does involve programming! $\endgroup$
    – A.M.
    Jun 30, 2013 at 19:55
  • $\begingroup$ Note, for comparison, this similar StackOverflow question where the users there recommended the questioner post it on CrossValidated instead: stackoverflow.com/questions/5694664/… $\endgroup$
    – A.M.
    Jun 30, 2013 at 19:56
  • 1
    $\begingroup$ Except yet again here it clearly is not working well. As the question remains fallow on Stackoverflow. The correct close reason if any was of a "poor quality"/"insufficient research" type - not an off-topic type. $\endgroup$ Jul 1, 2013 at 3:24
  • $\begingroup$ Yes, I think the lack of any attention to my question on StackOverflow in days should be taken as a clue that it is in the wrong place. ...and I hate to hammer on this point again, but maybe it's because it's a non-programming question forced over to a programming site! $\endgroup$
    – A.M.
    Jul 1, 2013 at 18:06
  • $\begingroup$ @Glen_b At first I did not want to argue too forcefully that my question is unfit for SO (since then it might be moved to SU instead of CV!), but I have to point out that I find the viewpoint "If a question is of the type 'which buttons do I press to achieve X' rather than 'what are the statistical issues with X', then it's arguably programming-related" impossible to understand. I don't want to be too harsh, but by that logic ringing a doorbell would also be "programming-related". $\endgroup$
    – A.M.
    Jul 1, 2013 at 20:34
  • $\begingroup$ @Glen_b There seems to be a false dichotomy entrenched here: that a question must be fit for either CV or SO. Even so, wouldn't you agree that the argument for the question not being fit for SO is stronger than the argument for it not being fit for CV. $\endgroup$
    – A.M.
    Jul 1, 2013 at 21:27
  • $\begingroup$ @A.M. Compare these two: Let's say I am using Excel. I record a macro and use it regularly. By recording and using a sequences of steps, I have quite literally written a program - I can examine and manipulate its code! ... Your argument seems to be that if I instead write that macro down on paper and follow its steps on the computer, it is off topic on SO, but if I type it in, it would be on topic on SO. And if that isn't where you draw the line, how does "what sequence of buttons do I press to do X" differ from "what macro do I write down to follow?" $\endgroup$
    – Glen_b
    Jul 2, 2013 at 0:13
  • $\begingroup$ @A.M. In particular, the definition (and I agree it's not the only one possible, but bear with me a moment) given in the first sentence at Wikipedia Computer program is: A computer program, or just a program, is a sequence of instructions, written to perform a specified task with a computer ... seems to quite directly include answers to 'What buttons do I press?' but excludes 'ringing a doorbell'. I find that quite a reasonable definition and indeed it corresponds directly to my understanding of the term. $\endgroup$
    – Glen_b
    Jul 2, 2013 at 0:43
  • 1
    $\begingroup$ @A.M. I don't think anyone here actually thinks those (CV and SO) are the only two possibilities; I know I have recommended movement to other places than SO many times (most frequently, math.SE, occasionally to one of a couple of others). $\endgroup$
    – Glen_b
    Jul 2, 2013 at 0:45
  • $\begingroup$ @Glen_b This really seems to be a game of "How low can you go?" with respect to calling different things 'programming'. I suspect that more than half of SuperUser (and a large chunk of AskUbuntu etc.) would fall under the definition you suggest. I think you have my position right on with "write that macro down on paper and follow its steps on the computer, it is off topic on SO". I think that is a great place to draw the line (and I still don't quite understand where you would draw it). Typing values into a few cells in Excel is not programming just because it could be done with a macro. $\endgroup$
    – A.M.
    Jul 2, 2013 at 2:00
  • $\begingroup$ As already explained, the set of instructions about exactly which things you do in which order to achieve an outcome on the computer both fits with the definition I pointed to and with the intuition I hold. But as to precisely where the line is drawn, what matters for the current question is where they draw that line on SO, and that's up to the community there. They don't usually seem to draw it in the same place you do. What matters for us is where we draw the line on what's a stats question, and if we think it fits better elsewhere, we move it rather than just close it. ... (ctd) $\endgroup$
    – Glen_b
    Jul 2, 2013 at 2:35
  • $\begingroup$ (ctd) ... it's then up to the community there to decide if it really does fit with their own standard about where to draw the line and then close it or do whatever else if it's not really a programming question. The mods of SO and CV can communicate if they feel we're often getting it wrong. I don't get the impression there's a strong feeling we're foisting what they think are obviously not-programming questions on them as a matter of course. $\endgroup$
    – Glen_b
    Jul 2, 2013 at 2:36
  • $\begingroup$ That's reasonable, and I am not saying there is necessarily anything is wrong "often" or "as a matter of course". ...but unlike most questions that are thrown over the wall to SO, this one specifically I would say does not meet their about-programming test (and they just have not noticed it yet). I would also echo the comment by @RussellS.Pierce below about this side of that wall: "Neither of the standards applied in this case have been vetted by the community (meta.stats.stackexchange.com/questions/1514/…)"...especially with regard to what kinds of questions are about statistical issues. $\endgroup$
    – A.M.
    Jul 2, 2013 at 2:59

1 Answer 1

7
$\begingroup$

Despite the close vote reason ('programming issue', which is the default message actually, see below), there's nothing in your question that would look like a statistical issue. As such it is off-topic and it might be better served on another site (certainly not Super User where there seems to be limited interest for R compared to, say, Stack Overflow where there's a vibrant community of R users).

enter image description here

It also worth noting that our site policy has evolved since its inception, and we now try to focus on questions that require statistical expertise rather than questions asking about tools, or polls involving shopping-list or extended discussion.

$\endgroup$
12
  • 6
    $\begingroup$ +1 One test I apply to such questions is "would this question, or at least its answers, be of interest to anyone who is using a different computing platform?" If the answer is no, the question is likely off topic. The next test is "might there be a community on SO that is interested in this particular platform and would likely give a good, quick answer?" If yes, I vote to migrate. If no, I write a comment referring to our resource links. In the present case, the answers were clearly "no" and "yes." $\endgroup$
    – whuber Mod
    Jun 30, 2013 at 13:06
  • 1
    $\begingroup$ On CrossValidated there are thousands of questions tagged 'r', 'spss', 'sas', etc., so clearly questions are not independent of computing-platform. $\endgroup$
    – A.M.
    Jun 30, 2013 at 14:26
  • $\begingroup$ There are also not very many statistics programs, and people working on statistics (besides people doing homework) generally do use one of these programs rather than implementing the theoretical solutions on this site from scratch (which actually would require programming, or superuse of Excel, or much pencil and paper). I don't think CrossValidated would suffer from having similar questions about implementation on in different programs. We just need continue to use the tags already there. ...and if some people really want to see theory-only questions, there can be a tag for that. $\endgroup$
    – A.M.
    Jun 30, 2013 at 14:35
  • $\begingroup$ I actually appreciate you trying to put this question in front of more people who might be able to answer it, but I really think CrossValidated members are more likely to (A) see it, (B) be able to answer it, and (C) benefit from seeing the answer in the future. ...the point of StackExchange sites. Also, what is wrong with this reasoning: meta.stats.stackexchange.com/questions/1/… $\endgroup$
    – A.M.
    Jun 30, 2013 at 14:42
  • $\begingroup$ One more point: If the question asker assumes there is an answer that does not involve programming, even if you assume it does should it really go to the programming SE site (should it be migrated anywhere based only on what the answers are likely to look like)? This is a tough case, and similar to the Meta-SO discussion of whether add tags questions when they relate to good answers but not the question itself. I think the concensus there was that classification of the question should be based on the question alone. $\endgroup$
    – A.M.
    Jun 30, 2013 at 14:49
  • $\begingroup$ OK...one more: You say "there's nothing in your question that would look like a statistical issue". That is just not the case. The statistical issue is that I have a within-subject design and am trying to use a program which defaults to assuming between-subject design. The solution could be program-specific, such as finding a menu selection or restructuring the data, or it could be pure theory, such as calculating the appropriate sums of squares etc. and having any program meant for between-subjects data yield the right output for a within-subjects experiment design. $\endgroup$
    – A.M.
    Jun 30, 2013 at 14:56
  • 2
    $\begingroup$ @A.M. Re: "there's nothing in your question that would look like a statistical issue". Where do you see any statistical issue in your question? Obviously you know what you are after (within- vs. between-subject design). However, you are asking about a graphical interface for R (emphasis is mine), where several plugins are available to extend its basic functionalities (there doesn't seem to exist one for RM ANOVA, though). Also, you might be interested in this Meta thread which is somewhat related to the older thread you cited. $\endgroup$
    – chl
    Jun 30, 2013 at 16:49
  • 1
    $\begingroup$ The accepted community standard is "with any statistical content". Neither of the standards applied in this case have been vetted by the community (meta.stats.stackexchange.com/questions/1514/…). If you want to keep adding those standards, add them in and let the vote take place. You'll likely win as you have elsewhere, but at least this time it won't be for lack of prominence. $\endgroup$ Jul 1, 2013 at 3:22
  • $\begingroup$ @chl I just plain fail to see how you fail to see any statistical content in my question. I have provided links to CV questions with similar levels of attachment to a specific platform but similar levels of statistical content, and I have even mentioned how an answer to my question could be entirely theoretic (i.e. What calculations, if any, could I make manually to a within-subject data set that would make injecting it into a between-subject tool yield the proper within-subject output. If the answer is "None. It's impossible.", fine, but it would still be a (platform-independent) answer. $\endgroup$
    – A.M.
    Jul 1, 2013 at 20:16
  • $\begingroup$ @chi I think the discussion of the suitability of my question for CV is important, but it (and your answer) misses the point of my question. I argue as to why my question is fit for CV as background, but the main question is: Is the question fit for SO? ...and that has a clearer answer IMHO, since it does not involve programming at all (despite the comment by @Glen_b). $\endgroup$
    – A.M.
    Jul 1, 2013 at 20:27
  • $\begingroup$ @RussellS.Pierce On top of your (very good) point, there is also obviously a lack of clarity on what constitutes "any statistical content". $\endgroup$
    – A.M.
    Jul 1, 2013 at 20:28
  • $\begingroup$ @chl Regardless of everything else here, thank you for your comment on Rcmdr plug-ins (even though you are saying there seem to be none for this use :( ). That is exactly the type of non-programming (non-SO, and actually very SU) answer I would have voted up. ;) $\endgroup$
    – A.M.
    Jul 1, 2013 at 21:31

You must log in to answer this question.

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