26
$\begingroup$

When this site takes off, we will get a number of homework questions that can be spotted a mile off. Should we

  1. Just down vote them and not answer them.
  2. Provide helpful hints
  3. Make a note in the FAQ and point them to it?

I think that we will get more homework questions than SO (say), just because statistics is taught to a wide variety of people - biologists, medics, engineers, computer scientists, mathematicians. However, we don't want to appear rude and put off future contributors.

Personally, I would go for a combination of all three options.

$\endgroup$
  • 2
    $\begingroup$ I don't yet have the reputation to post an answer, so I'll just comment here that this is one of the major issues MathOverflow has dealt with. The consensus (such as it is) which has emerged is more or less 3. But there have been some cases in which it's not clear what is homework and what is an honest question from an expert in a different (sub)field. For precisely the reason you point out, that may be a bigger issue here than on MathOverflow. $\endgroup$ – Mark Meckes Jul 20 '10 at 17:18
19
$\begingroup$

The tried and true method of contempt, sadism and downvotes?

On a serious note, I think the best way is to encourage someone to put the effort into turning their homework question into an acceptable Statistical Analysis question that can stand on it's own. This involves turning questions like:

  • [copy and paste directly from homework]

  • How do I do X?

Into questions that are more like:

I'm working on problem X and I think that I am supposed to do A, B and C. However, I am starting to think A wasn't such a good idea because it lead me to P.

Can anyone suggest an alternate approach to A?

Or, I tried B, but got Q and I think Q is wrong or cannot understand how to interpret Q so that I can do C.

Here is the work I did on step B to get Q. Can anyone help me understand Q or suggest how I should go about using it to get to C?

Of course, for that special case of homework question:

  • Help! Due in 30 minutes!

Contempt, sadism and downvotes are probably an appropriate response.

$\endgroup$
  • 1
    $\begingroup$ Excellent response, but I have more than a few words to say about that last suggestion concerning "contempt, sadism and downvotes," so I have posted them as a separate answer. $\endgroup$ – whuber Oct 12 '10 at 13:09
34
$\begingroup$

This issue came up at Finding marginal densities of $f (x,y) = c \sqrt{1 - x^2 - y^2}, x^2 + y^2 \leq 1$ , for instance. Let's look at the question constructively: how can we use homework questions to the benefit of this community? These include:

  • A good homework question can be a good question, period. What does it matter if it is homework?

  • Often, the elementary nature of homework questions draws multiple different answers, which illustrates relationships among fundamental techniques and exposes the foundations of statistical theory.

  • A homework question is usually fairly specific (at least it can be), whereas questions arising in other circumstances frequently need considerable clarification before they can effectively be answered.

The possibility, then, that homework questions can contribute to the site suggests they be allowed. Furthermore,

  • If homework questions are not explicitly allowed, people will just be forced to disguise their homework questions, which typically makes them vaguer and requires more work to answer. (We now have a "homework" tag. Let's encourage its use.)

Concerning how to answer homework questions: I would beg this community, both collectively and individually, not to use "contempt" and "sadism" as a tool for anything. That only leads to ill feelings, flame wars, and flagged posts. We should be especially alert for the possibility that a badly constructed question (that perhaps looks like homework) is due to the questioner's unfamiliarity with English. We should respect such questions more than most due to the effort required to formulate them. (If you're not convinced of this, sit down for five minutes and write out a question about statistics in any language you are not fluent in. Now think about posting this where the world can see it and potentially having to engage in a follow-on dialog in that language. How does it feel?) Accordingly, it is foolhardy and counterproductive to rush in with contemptuous or thinly veiled suggestions. It's easy to get people to visit a site once but hard to get them to keep visiting it; a policy of gracious and courteous answers from the community will IMO promote return visits.

Nobody wants to spend their precious spare time just doing someone else's homework for them: that benefits neither party. An approach frequently taken on the math site is to provide hints, a solution strategy, or a sketch of a solution (cgillespie's option 2). Sometimes these get amplified into a full-blown solution in response to comments, but even then some respondents have modified the problem slightly (e.g., to use different constants in an expression) so that some work has to be done to answer the original question. That strikes me as a good balance that addresses the positive aspects of answering homework while retaining the utility of the problem in helping the questioner learn.

Finally, why don't we build consensus on this issue and get it expressed in the FAQ?

$\endgroup$
  • 2
    $\begingroup$ Agree with all these points. I really like the consensus developed on StackOverflow: meta.stackexchange.com/questions/10811/… -- perhaps we can use that as a template for our own FAQ? $\endgroup$ – ars Oct 12 '10 at 16:37
  • $\begingroup$ @ars: Good reference, thanks. All we would have to do is modify explicit references to coding. $\endgroup$ – whuber Oct 12 '10 at 18:26
  • $\begingroup$ @ars I have implemented this idea by inserting a line into the faq and creating a wiki for the "homework" tag. $\endgroup$ – whuber Mar 19 '11 at 13:55
  • $\begingroup$ I truly believe whether the question is a homework question is completely irrelevant. What we should ask is: is this a good, interesting, well formulated question? I've seen so many poorly written answers with the excuse that they are "hints". If someone wants to just give hints, post as a comment, not as an answer. This policy seems to lower the quality of the site overall. $\endgroup$ – Carlos Cinelli Nov 10 '17 at 23:26
  • 2
    $\begingroup$ @Carlos On the contrary, what we currently understand to be a "self-study" question is one that is peremptory, artificial, generally devoid of interest, and intended only as a routine exercise in a textbook or examination. By raising the bar for asking and answering such questions, this site is immeasurably improved: users will not have to wade through hundreds of similar but useless posts when searching for material. $\endgroup$ – whuber Nov 11 '17 at 17:47
  • $\begingroup$ But see how you classified the question by its properties not whether it’s homework or not. I’m saying this because this homework “exceptionality” policy as if it’s something different seems to be unintentionally causing bad/noisy behaviors, such as comments asking whether it is a homework or poor answers. $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:06
  • 1
    $\begingroup$ @Carlos But that's precisely how we do classify the questions: we don't care whether they're homework; we care about the nature of the question. That's one reason the tag name was changed to "self-study" from the original "homework". $\endgroup$ – whuber Nov 11 '17 at 18:33
  • $\begingroup$ But should we still keep the "give hints not answer" policy? Or a special policy for self-study at all? It seems the guidelines are outdated, creating some noise: and the great answers for allegedly self-study questions do not follow the guidelines at all! $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:36
  • $\begingroup$ Just chiming in here that maybe as the "homework" tag died, the self-study should die too, and we have a clear focus on content/clarity. $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:38
  • 2
    $\begingroup$ @Carlos Again, I beg to differ: based on having read almost every question ever posted on this site--but not on any specific study--I believe the great majority of answers to self-study questions do follow the quidelines and that many of the remainder are from newbies who don't yet know the guidelines. I'm not opposed to your objective of focusing on content and clarity. $\endgroup$ – whuber Nov 11 '17 at 18:39
  • $\begingroup$ Could you show an example? I am going over the self-study tag and that's not what I see. $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:41
  • $\begingroup$ @Carlos I'm sorry, it's not my burden to show examples. If you think there's a problem then it's up to you to produce examples. You could do that in a meta post if you like. Cheers. $\endgroup$ – whuber Nov 11 '17 at 18:42
  • $\begingroup$ For example, your answer here is great stats.stackexchange.com/questions/99170/… very different from a poor hint like in this other question stats.stackexchange.com/a/257620/39630 $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:43
  • $\begingroup$ I might bring this up in a meta post later, still pondering about it, that's why I commented here to get your feedback. I will wait a bit to watch more user behavior and go over more examples before doing it though. $\endgroup$ – Carlos Cinelli Nov 11 '17 at 18:52
10
$\begingroup$

Just to reopen this, it seems the current practice is often (2) provide helpful hints, particularly in comments to the question. My view is that after a suitable period of time it is worth giving a fuller answer, as I did with How to find percentiles of a Normal distribution? 10 days after the question was asked. Is that reasonable?

$\endgroup$
  • 1
    $\begingroup$ (+1) Although I was the one who challenged your reply (on the basis of the apparent consensus in this thread), I personally think your view is a good one: why not eventually provide a worked example? Later, when the same question comes up (the one you answered will recur many times :-), we will only need to link to the example. A potential compromise is to answer a similar question, perhaps by changing the input data. $\endgroup$ – whuber Mar 19 '11 at 19:51
  • 2
    $\begingroup$ Some HW questions have lead to interesting threads, where the OP was clearly interested in going beyond the question asked; most of the time, however, such questions are rarely followed by the (unregistered) OP which prevents from just giving clues and waiting for a feedback. I like the idea of having a bag of typical HW questions that we could refer to, though. For the moment, I'm still acting (I hope so) as recommended: ask for what has been done, provide some hints or illustrative pictures, and then see how it goes. (...) $\endgroup$ – chl Mar 19 '11 at 20:21
  • 3
    $\begingroup$ (...) Why answering to someone who doesn't reply at all? The same policy should apply to HW or general questions in this respect. $\endgroup$ – chl Mar 19 '11 at 20:23
  • 4
    $\begingroup$ The stated goal for SE sites is to have most of their traffic come from search engines - which implies that for the most part, it's people looking for answers. I also support a policy of helpful hints, and after some period of time that renders a fully worked example "useless" showing up on an assignment, someone posting an answer. A homework question might still very well be a question worth answering. On another note entirely, I support self-labeling, not the community going "This feels like homework". Being on the receiving end of that at SO almost killed my interest in the entire network. $\endgroup$ – Fomite Oct 17 '11 at 20:21
9
$\begingroup$

I would say that we should ask them to put more effort into the question, and if the user persists, just close the question. The role of the site is not for doing other people's homework.

$\endgroup$
  • 3
    $\begingroup$ Any professor who is asked a homework question explicitly in office hours would tell the student to do a little work and rephrase the question, which is essentially what I mean. $\endgroup$ – Shane Jul 19 '10 at 20:48
  • 4
    $\begingroup$ Then we need to differentiate between homework questions that someone has put effort into asking and homework questions that have had no effort expended on them. $\endgroup$ – Sharpie Jul 19 '10 at 20:49
1
$\begingroup$

"Obvious" homework?

  1. First, look at the questioner's SE network profile to see if the person is obviously a student or not.
  2. If the person is not obviously a student, consider assuming good faith (as is documented on Wikipedia (and yes, I understand this is not Wikipedia)).
  3. Consider just asking in a comment, "Is this for a course?"
  4. Consider just ignoring the question for a while, because one thing is generally certain about homework: It has a due date. And after ignoring it for a while (a week?), and if nobody else has provided an answer, simply comment and ask, "Are you still looking for an answer?" If the questioner doesn't respond within a few days, then perhaps it was homework, or in any case, the questioner must not care that much about an answer.
$\endgroup$
  • 1
    $\begingroup$ I thought this was very gentle (no objections to gentleness rather than its opposite) until I read #4 which advises us, IIRC, to ignore such questions for a week or so. If that advice was followed, I doubt that any of these questions would get answers. Just guessing, but who sets a personal reminder "Look at this question in a week's time to see if it's still there". $\endgroup$ – Nick Cox Mar 12 '18 at 19:14
  • $\begingroup$ An advantage of our current policy is that to apply it we need only decide whether a question's an exercise or not - much easier than speculating or quizzing people about whether or not it was set as homework. What would we gain from changing it? What (good) reason would anyone doing an exercise have to object to explaining where they were stuck & getting hints for the next step instead of a fully worked-out solution? $\endgroup$ – Scortchi Mar 12 '18 at 20:37

You must log in to answer this question.

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