We get quite a number of self-study questions, and one common comment says that

We'll provide hints to help you get unstuck.

So... how specifically should we provide these hints? Should we write comments to the question? If the OP answers in comments, we can comment again, and finally post a full answer as a summary. This has the advantage that the train of thought is there for later generations to follow. The disadvantage is that comments are ephemeral and may be deleted sooner or later.

(The other disadvantage is that someone may barge in with an answer in the middle of our pedagogical back-and-forth, but than can also happen with the alternative below.)

Or should we post hints as an answer, wait for the OP to comment, then update the answer and iterate? Or post hints as an answer, then do a socratic Q&A in comments, finally update the answer? Or even only leave the original answer-hint and the comment conversation below, so later generations don't get an immediate answer as a spoiler?

Thoughts, anyone?

  • 4
    $\begingroup$ Long discussions in comments are not pretty solution, but sometimes they are necessity. $\endgroup$ – Tim Nov 19 '17 at 20:44
  • 4
    $\begingroup$ I usually at least try to do the former (discussion in comments & then convert to an answer). Here is an example when I did so. Alternatively, here is a case where I posted an initial hint as an answer & then had a discussion w/ the OP in comments underneath my answer. $\endgroup$ – gung - Reinstate Monica Nov 19 '17 at 21:05
  • 4
    $\begingroup$ Stern strategy is to comment if and only if the OP indicates some thinking towards a solution. Sweet strategy is to give a broad push regardless. ("Consider calculating the probability that the condition is not satisfied." "Did you consider using Cauchy-Schwarz?" "Are you getting confused between addition and multiplication?") $\endgroup$ – Nick Cox Nov 20 '17 at 9:27
  • 3
    $\begingroup$ I think that, in general, the strategy of giving hints in comments and then perhaps making then into an answer is good. But if the solution is something very straightforward (such as missing a - sign somewhere or using addition instead of multiplication) that might go to an answer right away. $\endgroup$ – Peter Flom Nov 20 '17 at 12:04
  • 1
    $\begingroup$ NB There's a format for spoilers that can come in handy - see stats.meta.stackexchange.com/a/1815/17230. $\endgroup$ – Scortchi - Reinstate Monica Nov 20 '17 at 15:06
  • 3
    $\begingroup$ I'd rather we did not have this tag, and treat self-study questions just with our custom close reason for it, and corresponding guidance through our meta faq. Also, we would relax answering in a full way or not, leaving the decision to post writers. Have expressed my opinion here. $\endgroup$ – Andre Silva Nov 20 '17 at 16:23
  • 9
    $\begingroup$ Every technical forum I know is in tension on homework and assignments (I don't like the term self-study either but I doubt the term is the main issue). The tension is between a general goal of wanting to help those able and willing to help themselves and specific goals of (a) not wanting to support the idle and clueless (arguably in their own best interests too) (b) not wanting a forum to be swamped by elementary or trivial questions (c) not wanting to collude in or condone, however unwittingly, plagiarism or collusion, which are offences (NB) in any respectable university or equivalent. $\endgroup$ – Nick Cox Nov 21 '17 at 2:44
  • 6
    $\begingroup$ (a) (b) (c) are all in practice serious matters and, although they can also arise in some questions from researchers, they seem to imply the necessity for an explicit policy. In short, I can't agree that the tag should be abolished. $\endgroup$ – Nick Cox Nov 21 '17 at 2:46

The purpose of the tag

Let me begin by explaining what I see as the primary value of the self-study tag.

Let us imagine that we were in an ideal world where OP's always read the help, and the tag wiki and used the tag as we would wish. Then:

When first answering a question, the tag acts a signal to answerers that the question is actually asking "Can you offer me guidance on how to approach a problem like this" -- so the correct answer is one that helps guide OP's to answering such questions themselves.

However, this is not really the purpose of a tag; that aspect could be handled instead by asking OP's to signal at the start of their question is of this kind.

The purpose of a tag is for later readers, searching for particular classes of answer. In this case the tag serves the next honest OP who also has a self-study question and wants to get hints, not complete answers.

By adding the self-study tag to a search, they should not see any complete answers in their search, even if they exist on site. This means that the honest student gets what they want, assistance, without missing out on the value of solving their own problem.

That is*, I think self-study posts should generally avoid giving fully-complete answers (except perhaps where posted by the OP) -- the answers serve best that give clear, solid guidance on how to approach them.

* (and my thoughts on this have progressed from when I first started answering them, as I came to understand the purpose of tags better)

The ongoing value of a self-study question

This way, self-study questions have ongoing value beyond the immediate question, and are able to serve a series of more-or-less puzzled students.

It also suggests that stronger hints should appear later in the answer (so that a reader need only read as far as required to get over their hurdle)

Indeed, if you wish, you can add spoiler markup (start each line with >!) so that

readers must mouse over
to see the strongest hints


[If we do answer self-study questions fully we'll eventually end up with a repository of complete, detailed answers to almost any reasonable question that might be set for a statistics student. We'd have questions set by experts (mostly) and solved by experts (mostly), without any of the practice passing through the minds of the students the work was intended to benefit. Imagine the state of music performance of students who -- instead of actually practicing playing -- could send in videos of someone else performing the pieces they were supposed to learn!]

  • 3
    $\begingroup$ This is an interesting perspective. However, (1) we have so many complete answers (instead of "hints") in the [self-study] tag, including from our most respected users, that I am not sure it can serve this purpose in reality. (2) I strongly doubt that people use our site like you suggest: restricting the search to [self-study] in order not to see full solutions. Somehow I cannot imagine anybody doing that. I might be wrong, but I think that here you are trying to provide a service that [almost] nobody needs. $\endgroup$ – amoeba Nov 25 '17 at 11:38
  • 1
    $\begingroup$ I share amoeba’s opinion. Besides, following this approach we give priority to help an individual user rather than attempting to build the best repository of questions/answers that we can. $\endgroup$ – Andre Silva Nov 25 '17 at 12:37
  • 1
    $\begingroup$ @AndreSilva My answer is arguing for the opposite -- making self-study questions useful beyond a single user. The fact that some questions have already been answered fully (i.e. the policy has not been sufficiently adhered in every case) doesn't imply that it's best to continue that way. Amoeba seems here to be arguing that we simply abandon existing policy, while I am trying to make the existing policy work better with the SE framework. While we could alter the policy (to one where self-study is essentially superfluous since every question would in effect be a free-for all), ... ctd $\endgroup$ – Glen_b Nov 25 '17 at 17:15
  • 1
    $\begingroup$ ctd... that would be something to address in a new question. We would need to revisit the arguments for why we don't want complete answers on self study questions (and we'd need to revisit the question of what the purpose of the tag is, independent of the policy -- what are tags actually for?) $\endgroup$ – Glen_b Nov 26 '17 at 0:20
  • 2
    $\begingroup$ I agree with this perspective, even though I doubt later readers approach such questions in the same spirit as the OP. You can usually feel some degree of desperation in the OP's tone. But keeping from entirely solving the problem should be the rule, including spoilers. $\endgroup$ – Xi'an Nov 26 '17 at 16:14
  • 1
    $\begingroup$ @XI'an I agree -- there should still be work left for the reader; that's the point of assigning such material; a spoiler is simply one way of helping people that choose to avoid bigger hints to do so - even though indeed most readers would not care to avoid any hints that are going. $\endgroup$ – Glen_b Nov 26 '17 at 21:22
  • 2
    $\begingroup$ Glen, I think there are some false premises here: (i) self-study questions are already being answered fully, not being properly tagged, and what you fear is far away from happening; (ii) if what you feared actually happened (almost all reasonable statistics questions with complete answers here) this would be probably the best thing that could ever happen to this site, being great not only for students but for society in general --- there would be nothing like it anywhere else. Students who really want to learn must have self control to attempt first, it's not this site's job to enforce that. $\endgroup$ – Carlos Cinelli Nov 27 '17 at 4:02
  • $\begingroup$ By the way I'm a self taught guitar player, and watching fully explanatory instructional videos was one of the most effective ways to speed up learning. $\endgroup$ – Carlos Cinelli Nov 27 '17 at 4:02
  • 1
    $\begingroup$ PS: I don't mean to favor answering poorly written questions demanding answers such as "hw, ans pls!!!" --- these questions should be simply downvoted and closed --- my point is just that I don't see any special reason to give a question special treatment, other than the question's content itself. $\endgroup$ – Carlos Cinelli Nov 27 '17 at 4:12
  • $\begingroup$ "Imagine the state of music performance of students who -- instead of actually practicing playing -- could send in videos of someone else performing the pieces they were supposed to learn!" $\leftarrow$ I think you just described the musical practice of "sampling". $\endgroup$ – Sycorax Jul 26 '18 at 4:45

It seems the community has unnecessarily struggled with "homework" questions for a while, mainly due to a worry of having students getting easy answers around here. I see the good intentions of this policy, but it seems it should be clear by now this is something out of our control and there's no reason to try to address it here.

If you take a careful look at the current questions on the "self-study" tag, many of them are not even homework questions with very good complete answers. The idea of hints would be completely inappropriate, leading to wrong/incomplete answers that would not be useful to anyone.

This idea of giving hints also seems to bias many users to question the OP's intents, leading to unnecessary noise. Instead of asking for clarification or answering the question, the person takes the time to ask "is this homework" or "this smells like homework", which serves no purpose. I have also seen very poor answers which should be comments, induced by this policy of "giving hints".

In sum, I think we should assess a question just by its merits. If you think it deserves your time for a full answer, because it's an interesting question, or because it's a widespread misconception that needs to be addressed, just do it and answer it properly. If you think it's a so simple question that you don't even want to bother answering, post a hint in the comments. If it's a question with no effort at all, downvote it and ask the OP to make it better.

PS: I'm aware my views in this topic might considerably differ from current practices, but I thought this question would be a good opportunity to put these ideas into debate.

  • 4
    $\begingroup$ Personally I agree, but this has been discussed on Meta many times and the strong consensus among the most active and respected members of our forum is that our current policy regarding [self-study] tag works reasonably well and should be maintained. $\endgroup$ – amoeba Nov 24 '17 at 22:43

The answer by Glen_B regarding the optimal use of the self-study tag is helpful to me, but I would like to offer another perspective involving some practical difficulties I have encountered when approaching these types of questions. (I am not a very experienced user compared to some others, so maybe I am still finding my footing with these questions.) Certain questions are easily identified as textbook questions, but sometimes it is hard to gauge how "routine" a question actually is. There are also some other practical problems I encounter when I approach these questions:

  • Practical Problem 1: The goal of the site is to give answers that are elucidating to a broad readership, and not just the specific questioner. However, in the self-study questions the analysis of the question is limited by asking the OP where he/she got stuck, and addressing a hint to assist progress with this part of the solution. You could have different questioners all asking the same statistical question, but they are getting stuck on different parts. This makes the answers very specific to the OP and not necessarily very useful for other readers who are also interested in how to solve that type of question. Thus, there is a tension between giving an answer of broad interest to readers, versus giving a hint to a specific OP that is stuck on one specific part of a question.

  • Practical Problem 2: Often the best kind of "hint" for a problem is to show how one or two missing mathematical steps can be done to get further in the problem. Sometimes this can be alluded to by telling the OP to apply "so-and-sos Law", but sometimes this is insufficient and you need to show how it is applied in context, which means you are actually doing one or two mathematical steps for them. However, writing out mathematical steps properly requires preliminary context and definitions, which means you may have to write out all preliminary definitions and steps up to that point to make sense. If the "road-block" in the question comes near this end, this is pretty close to giving a full answer already. So in these cases you are trying to figure out how to show some steps (e.g., applying some law) without giving something that is pretty close to a full solution.

  • Practical Problem 3: Questions with the self-study tag vary enormously in their level of difficulty. Some are routine undergraduate textbook questions and some are difficult theory questions. One recent self-study question was a functional optimisation problem that is at least at graduate level, and is even a whole new area of study for most graduate students. (Arguably this was mistagged, since the problem is hardly routine.) Such readers mostly have not seen functional optimisation at all, and the only way they will understand it is if you step them through the whole process. It is useless to give a "hint" in this case, because the problem is at a high level and the OP does not know how to start, or how to do any of the later steps. They don't even know how to frame the question accurately.

  • Practical Problem 4: Many questions with the self-study tag are old enough that the OP has finished whatever course raised the question. (Some are five years old; the OP has probably graduated and entered the workforce!) There is a hint given, but the OP does not come back and show the solution obtained with the hint. In this case a full answer does not advantage the OP, but it remains in the repository of answers that other students can look up. As noted above, for the broader readership, they are interested in the overall question and its solution, and not necessarily the exact hint relevant to that OP. So in this case there is no advantage to the OP in giving a full solution, which militates further in favour of an answer that is useful to a broad readership. In this case, what is a reasonable "hint" for an OP who asked a question five years ago?

  • Practical Problem 5: The OP is allowed to bypass the restriction on giving a full answer, but other users are not. This seems quite arbitrary. If full answers are a problem then they are a problem; why would they not be a problem if they come from the same user that posts the question? (To my mind, this illustrates that the purpose of this tag is horribly confusing.)

I am still fairly new to answering these questions. As with Glen_b, it is possible I will come to understand them better with experience, so I apologise in advance if any of what I am saying is silly. Still, I find it a bit confusing deciding how to approach some of these questions.

An aside: In regard to these questions, I also tend to agree with Carlos that a repository like this would actually be really great, and it should not be the responsibility of CV users to enforce self-control on students. I realise this issue has been discussed, so I will just leave this remark here as dissent from the majority view. The practical difficulties above emerge regardless, but I don't actually agree that it is a problem if CV has a repository of solved textbook questions.

It is certainly true that students can skip-out on learning problems by looking up answers on CV.SE, but universities have methods for testing student knowledge independently of this; if a student gets their answers from CV and does not learn the material then he/she will not be able to pass an exam. Is it really up to us to enforce a barrier to knowledge to deal with a student who tries to avoid learning - a fortiori when the student still needs to be able to pass examinations without access to CV experts?


You must log in to answer this question.

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