2
$\begingroup$

I recently answered a multiple choice question on CV and although the OP did not label it homework it obviously was. Should we try to give hints? I had a hard time finding a way to suggest an answer without providing my rationale for picking a specific choice. A moderator thought that my answer violated policy. Here is the post. Community memebr gung edited the question and presumably added the homework tag that was missing.

What should be done? Should we just not respond to multiple choice questions? Is there a way to be helpful with questions like this one without answering it?

| |
$\endgroup$
  • $\begingroup$ I think that some questions that are multiple choice can be easily handled without violating policy, but I had a particularly hard time with this one and am really asking about this one and questions similar to it. $\endgroup$ – Michael R. Chernick Sep 25 '12 at 12:38
  • 3
  • $\begingroup$ While there have been discussions here about answering homework questions and the SE policy, this question is specifically addressing only multiple choice answers where I think the issue of giving a helpful answer could conflict with staying within the borders of the guideline. In the referenced question even Zen's short comment seems to me to be one that violates the guideline. $\endgroup$ – Michael R. Chernick Sep 25 '12 at 15:05
  • 1
    $\begingroup$ On the main SO site, the homework has now been officially depreciated. Aside, I think this is a good thing. $\endgroup$ – csgillespie Sep 26 '12 at 10:51
  • $\begingroup$ @csgillespie It may still be relevant on sites like Cross Validated or math, IMO. $\endgroup$ – chl Sep 26 '12 at 12:39
  • $\begingroup$ @chl It may, but why is learning statistics any different from learning Java (say)? Looking at the reasoning on Meta-SO, I tend to agree that Homework questions tend to be poorly written or localised. In that case, they should be closed. $\endgroup$ – csgillespie Sep 26 '12 at 14:30
  • 1
    $\begingroup$ @cs I agree with you. (In fact, I voted to close the question that inspired this thread.) But to address your (almost rhetorical) question: a huge difference between learning stats and learning Java is that in the latter case, most problems--even those often faced by programmers in the trenches--are well defined and clear. In stats, half the battle lies in recognizing what the problem is and being able to characterize it. $\endgroup$ – whuber Sep 26 '12 at 20:47
  • $\begingroup$ @whuber However, with stats homework the problem is (usually) well characterized. $\endgroup$ – csgillespie Sep 27 '12 at 13:43
4
$\begingroup$

How about "provid[ing] helpful hints" as suggested by @csgillespie and follow @whuber's suggestion on one of the thread @Procrastinator referenced in his comment: "How can we use homework questions to the benefit of this community?"

For this question, in particular, I would have sketched briefly the distinction between continuous and discrete outcomes, with a particular emphasis on the Binomial and Poisson law and, maybe, how they relate one to the other. The OP also favored one of the option, and ruled out option D; we could rely on this and propose further discussion on probability distributions. This is what you did here and there in your post, but your third sentence already suggested a solution which leaves little room to arouse curiosity and thinking on the side of the OP.

In any case, I believe we should not feel embarrassed by how a question is formatted (many HW questions could easily be recasted as MCQ, IMO), and follow usual advice to answer homework questions, as suggested in the comments.

| |
$\endgroup$
  • $\begingroup$ I understand your approach to the question and it seems that it would work to give the hint that the answer must be binomial or Poisson without pointing to the Poisson. But at that point I am not sure that he has enough information to figure out Poisson over binomial. I found myself that it is a little difficult to decide which model to apply. Maybe there are rational arguments that can be made for either case. In fact, if you take the parameters n and p and make p decrease as n increases for the binomial such that np stays fixed at a rate lambda the binomial approaches the Poisson. $\endgroup$ – Michael R. Chernick Sep 25 '12 at 20:16
  • $\begingroup$ So maybe an argument could be made that both can approximate the solution. Perhaps it is most helpful in this case to provide all this information and it would not settle the issue of Poisson over binomial. So this question has added complexity in that although I think Poisson is a better answer and a more appropriate model the binomial cannot be ruled out. $\endgroup$ – Michael R. Chernick Sep 25 '12 at 20:21
  • $\begingroup$ But I do think in general it is harder to give good hints in a multiple choice question without pointing to the correct answer or at least explaining why certain choices can be eliminated. How far is going too far according to the guidelines? $\endgroup$ – Michael R. Chernick Sep 25 '12 at 20:23
  • 2
    $\begingroup$ @Michael Guidelines are simple: don't provide full answer, and try to provide as much useful hints as possible so that it is beneficial for the community in general, and the user, especially. I must admit this is not always an easy task. Comments may also be used to gauge user's level of knowledge. That may help going beyond simple MCQ. $\endgroup$ – chl Sep 26 '12 at 12:45
  • $\begingroup$ Agreed @chl. It takes a little more effort but I am learning thatin the grand scheme of things it is best to follow the guidelines. $\endgroup$ – Michael R. Chernick Sep 27 '12 at 23:17

You must log in to answer this question.

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