2
$\begingroup$

I came across an old question

Computing p-value using bootstrap with R

and I thought that the answers might benefit from a graphical presentation.

I liked the answer by Jan S. which shares my idea about it; The problem with the discrepancy between the p-value and bootstrapped distribution is similar as the recent question Why do the bootstrap calculated p-value and the confidence intervals seem to contradict each other [R code, specific example]? .

I was thinking about adding some graphics and

  • I could do this in a new answer

    but that thread is already more than 10 years old and adding a new answer to answers that have already been upvoted several times might be an answer that gets less attention. Also one might worry about getting too many answers on a single question, possibly it could be better to just upgrade another answer instead of adding a completely new answer.

  • I could also add my idea to the already existing answer.

    But that answer is already old and the user that created it is not active, and also all the votes that it gathered in the past might not reflect my additions. In addition, people that initially created answers might like to have some personal control over curating how the questions develop (not much the issue here with an inactive user, but possibly problematic in other cases).

I perceive the Q&A format on stackexchange as very individualistic. Sometimes I have the urge to change a current answer rather than adding a new answer that is 'completely my own'. But as explained in the example case above I feel there are several considerations with changing an already existing answer (and I have the feeling that big changes are not very common). For that particular case, how should I handle?

$\endgroup$
21
  • 3
    $\begingroup$ Unless the post is community wiki, I would feel particularly uneasy in adding anything substantial, albeit it could have provided much insight to the existing post. This deserves a separate answer altogether, imo. If the post is useful, the community would definitely get sight of it. $\endgroup$
    – User1865345
    Apr 28 at 8:04
  • $\begingroup$ @User1865345 I doubt your final sentence. Would the community get sight of it? And would it be useful for the community if an idea that builds upon a previous answer is posted as a seperate answer? The post might be useful but is in a thread that has already 3 other highly upvoted answers. Is it user-friendly if a person that lands on that question has to dig through several questions to gather multiple ideas? It seems a bit weird to add a completely new answer when one only has an idea to change/improve an already existing answer. $\endgroup$
    – Sextus Empiricus
    Apr 28 at 8:17
  • $\begingroup$ I often have these considerations when I come across an old question and I get the idea that an image might help to explain the idea more clearly. We do often edit other questions here for spelling or improving language, but that seems to be the limit. $\endgroup$
    – Sextus Empiricus
    Apr 28 at 8:21
  • $\begingroup$ If this helps further address the issue of the question that somehow didn't get articulated in the already positively received posts, you should compose another post with clear disclaimer that this is in relation to another post, but since it has not been highlighted, this warrants an additional post that provides the relevant discussion. $\endgroup$
    – User1865345
    Apr 28 at 8:22
  • $\begingroup$ An example of a major edit that I once made is here: stats.stackexchange.com/revisions/354698/4 $\endgroup$
    – Sextus Empiricus
    Apr 28 at 8:25
  • $\begingroup$ If the user were active, I could have thought about asking them about the addition provided the addition is a graph, a snippet, an extract or info along the same line. However, if the addition is longer than that, this should be a seperate post along the explicit statement of why it is connected to the original answer in concern. $\endgroup$
    – User1865345
    Apr 28 at 8:25
  • $\begingroup$ For your example, you clearly stated your edit in a comment to AdamO, who is a regular user in the community. However, we are dealing here with a user who, for all practical purposes, won't return here. If you feel the addition would be small but important in addressing and highlighting the solution in a better way, please edit it. But if it is larger, I opine that this should be in a seperate post. $\endgroup$
    – User1865345
    Apr 28 at 8:28
  • $\begingroup$ @User1865345 my worries about adding a new post is that in the long term reading cross validated becomes like reading a history page of Wikipedia or a list of revisions. Instead of a single concise answer that gathers multiple contributions into one post, the users of cross validated have to dig through several posts. A post that explains something and another post that explains the other post, that's monstrous. $\endgroup$
    – Sextus Empiricus
    Apr 28 at 8:34
  • 1
    $\begingroup$ I believe that the addition will be big but important. What I worry about with adding a new post every time in that case, is the unnecessary growth of posts. At what time will a post turn into compost? Or is it everlasting plastic? $\endgroup$
    – Sextus Empiricus
    Apr 28 at 8:35
  • $\begingroup$ I can sense your worry which is somewhat legit. But if I were to embrace it fully, then all queries should have to have at best one or two answers. I can personally tell, I love when a question gets quite a few answers: I take time to read them, try to comprehend the different approaches and even if two posts resemble same, chances are I would still learn a varied take. $\endgroup$
    – User1865345
    Apr 28 at 8:42
  • 1
    $\begingroup$ I have edited the title to articulate your main concern. Please feel free to edit or undo that, if you wish. $\endgroup$
    – User1865345
    Apr 28 at 8:53
  • $\begingroup$ This does seem like a small (but probably growing) structural problem with SE - new developments will mean that highly-upvoted answers are no longer accurate. If your planned addition is based on such a new development, it might justify a carefully-worded new thread (and perhaps even closure of the existing thread as a dupe, to point people towards the accurate new answer). If it's something that could have been written when the question was posted, I would probably add a new answer to the existing thread and just accept that it will get less attention. $\endgroup$
    – mkt
    Apr 28 at 8:57
  • $\begingroup$ I would definitely not edit an existing answer to make substantive changes. $\endgroup$
    – mkt
    Apr 28 at 8:58
  • 1
    $\begingroup$ This comes down to what makes sense in individual threads. In practice, the person who answered has long since stopped contributing, so to me that encourages editing their answer so long as edits are clear to all. I recommend something like EDIT 28 April 2023 in the answer. But if that person expressed themselves uncomfortable with the edit, then I would back off and create a new answer with a cross-reference. It's a clash between high-level principles (a) the site being ideally a repository of high-quality answered questions (b) a courteous need to respect others' styles and wishes. $\endgroup$
    – Nick Cox
    Apr 28 at 12:33
  • 2
    $\begingroup$ So, experuenced users can disagree on this. I too am a little divided on the point. I find myself a little irritated if people edit my posts, but almost always the intent is benign and the change makes sense, so I tell myself to get over it, and there is no sense in which other people need my permission. And I edit others' posts mightily, usually as a matter of language or presentation, so a taste of my own medicine should be acceptable! $\endgroup$
    – Nick Cox
    Apr 28 at 13:26

1 Answer 1

Reset to default
3
$\begingroup$

You should not make substantive changes to someone else's answer. This has been discussed before on Meta.CV, e.g., in Are you allowed to edit an answer that you disagree with?. That does not mean that no edits are possible. You can see here that over the years I have found nearly 10k places where an edit would be beneficial (I believe edits to my own posts are not included in that count). Many of those were answers, but none of them were substantive (excepting the instance mentioned below). Edits are best for things like spelling, punctuation, grammar, formatting and clarification that preserves the authors intent. The post that is listed as by a user, even if they've deleted their account, should reflect what they wanted to say. If you think an answer is incorrect, or incomplete, you can vote down and/or leave a comment. You are free to add your own answer and reference theirs. But you should not make a substantive change without their permission. The idea that a figure would be helpful is, at least in the abstract, quite reasonable. I can relate an occasion in which that happened with me. I made some plots to illustrate a point someone else was making, and I thought they would benefit the answer. So, I asked them in comments if it would be OK. If they had said no, or not responded because they were no longer around, I could have put them in another answer and referred back to the answer at issue, and I could have put another comment under that answer pointing to mine. But I would not have changed someone else's answer in a meaningful way without their permission.

$\endgroup$
11
  • 1
    $\begingroup$ (+1). This is absolutely what I was saying and advocate. I do understand the concern that Sextus is trying to articulate. However, at the very least, I am not bothered at the sheer number of answers a query receives over the years: if the post is useful, it deserves to be added as an answer even if it is in reference to other one (in light of the fact of the scenario discussed here). $\endgroup$
    – User1865345
    Apr 29 at 13:09
  • 1
    $\begingroup$ "You are free to add your own answer" In my question I noted that this is demotivating for old questions where answers already have gathered several tens of votes. "If you think an answer is incorrect, or incomplete, you can vote down and/or leave a comment." votes have little effect on old questions and comments don't work if the original poster is not active. The activity in old question and answer threads has died out. What do we do with that? $\endgroup$
    – Sextus Empiricus
    Apr 29 at 14:37
  • 1
    $\begingroup$ "But you should not make a substantive change without their permission." This seems to be the current etiquette. I like to challenge that this is a strict rule. People add their answers with an open license. They do not own the answers. It is just a custom that we don't do this, but it can be changed. $\endgroup$
    – Sextus Empiricus
    Apr 29 at 14:39
  • 1
    $\begingroup$ @SextusEmpiricus, the design of the SE system is that information accumulates over time by people answering and those answers being evaluated by the voting mechanism. Eventually, in theory, consensus emerges from the community's votes. Taking an answer with "several tens of" votes and making substantive changes to it so that it says what you want it to say instead of what the author said breaks that mechanism. You might consider that some of your posts have garnered disagreements in comments. Is it your preference that those people change what you put? After all, you "do not own the answers". $\endgroup$
    – gung - Reinstate Monica
    Apr 29 at 16:51
  • 3
    $\begingroup$ More to the point, let's say I like Jan S's answer better without the figure you add. Should I just edit it back out? What happens then? Do we get into an edit war until the mods lock the thread? How is this supposed to be adjudicated? I suppose it is more realistic that no one much cares enough to bother with it and the figure just stays by default. But that just sidesteps the issue--it doesn't resolve it. $\endgroup$
    – gung - Reinstate Monica
    Apr 29 at 16:55
  • 1
    $\begingroup$ I get your ideas edit-wars. I am not sure how to solve it but possibly a Wikipedia alike system might work (I have no personal experience with it, but I do know that it is much more collaborative and evolving/changing). $\endgroup$
    – Sextus Empiricus
    Apr 29 at 17:32
  • 1
    $\begingroup$ The points system, I agree that this doesn't work well if a post with a lot of votes gets changed by a lot. But the same is true that the system doesn't work well for new answers, which did not get the same exposure. In this sense my question can be seen as part of a larger issue about how to improve the q&a database and what to do with old parts that had a different voting history. (An example if that sometimes duplicates can replace older posts when they are better, but there is no standard path to improvement of old parts of the website). $\endgroup$
    – Sextus Empiricus
    Apr 29 at 17:34
  • 2
    $\begingroup$ @SextusEmpiricus I appreciate your concern with incorrect answers that have lots of upvotes. Unfortunately, there's little we can do about that within the SE framework. I've been trying to get the author of this answer on SO stackoverflow.com/a/49444783/2482661 to either correct or remove it for a year. It's the top result for a number of common Google queries but it the code do what it says it does... and the question isn't even about code! The only thing to be done is leave a comment or offer a better answer (or even downvote if you feel the answer warrants it). $\endgroup$
    – Sycorax Mod
    Apr 29 at 18:06
  • 1
    $\begingroup$ @Sycorax that's another good example case. But for the record, I am not saying that the answer of the case that I described here is incorrect. I actually found it the best answer and I already upvoted it. It is just that I was thinking that the question could be improved by adding an extra example and/or an image. I am thinking about old questions that can be interesting candidates for a facelift rather than adding a new question that's basically the same idea but formatted differently (the situation where I think I have something to add that's a lot for an edit but too little for a new post). $\endgroup$
    – Sextus Empiricus
    Apr 29 at 19:35
  • $\begingroup$ It is a difficult situation. A similar situation is here: stats.stackexchange.com/a/592453 where I added a new answer rather than editing the existing answer. In that case the original author was still around and actually suggested edits to their own post but I argued that the minor edits were sufficient and "A heavier edit might break down the storyline that you had before.". But, being 10 years younger, the answer that I added is sitting a bit lonely at the bottom of list of answers. In this case, no problems, as it is referenced in the other answer, but that is not standard. $\endgroup$
    – Sextus Empiricus
    Apr 29 at 19:53
  • 2
    $\begingroup$ @SextusEmpiricus, your points are well-motivated. The SE system is imperfect. But in the end, I think Sycorax is right that "there's little we can do about that within the SE framework". In some sense Wikipedia has similar goals, but it is ultimately a different project. It has a different design, with different pros and cons. The designers of SE did not aspire to be Wikipedia. This system has its pros and cons; I'm much more active here than on W, but others may choose a different balance. $\endgroup$
    – gung - Reinstate Monica
    Apr 29 at 21:20

You must log in to answer this question.

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