Following on my previous question: Should our FAQ include information about answering questions, and the subsequent discussion in the comments there, this question is intended to elicit discussion from community members with the goal being to develop a set of suggestions that can be linked to from the main site's FAQ.

The central question can be conceived of several ways:

  • What advice should we provide a new user who is largely unfamiliar with Cross-Validated, but wants to help out answering questions here?
  • What constitutes a great answer on Cross-Validated?
  • To what ideal should our answers aspire?

etc.

share
    
Note that some suggestions regarding possible answers can be found from links posted at my previous Q. –  gung Sep 19 '12 at 14:09
3  
Very nicely done (both the Q and the A). As you proposed, I have linked to this thread from our FAQ. –  whuber Sep 19 '12 at 17:55
1  
A parallel thread on asking good questions is a worthy complement to this one. –  whuber Jan 10 '13 at 19:39
    
A whole list of threads on meta-SO can be found here that may be relevant for readers. –  gung Feb 19 '13 at 13:43
add comment

1 Answer

up vote 19 down vote accepted

As noted in our about page (our about page has changed since this was written--it's more useful now, but don't be confused if the following description seems out of place), Cross-Validated is understood, by design, to overlap somewhat with other types of sites that already exist on the internet (specifically: Wikipedia; blogs--e.g., statistics blogs like Rob Hyndman's, Jeromy Anglim's, John Cook's, Peter Flom's etc., and possibly R-bloggers, although CV is not to be confused with an R programming site; statistics forums--e.g., statisticsforums.com, or perhaps talk stats; and Digg / Reddit). This fact makes it difficult to write the 'perfect' CV answer--there is no formula or set of boxes that, if checked off, will be guaranteed to produce the best answer.

However, some guidelines can be given:

Answer the original poster's question:

  • This is a question and answer site, we do not just provide information about statistical topics
  • Your answer should be written such that the OP will be able to understand it (given their current level of statistical sophistication, as best you can divine) and utilize that information to accomplish their goals
  • If the question is not clear enough for you to be able to answer in this manner, you can comment on the question to ask the OP for more information
  • Be respectful of the OP. Many people have math anxiety, and many questioners here clearly feel uncomfortable (I can't count how many questions I've seen that start with 'I'm sorry if my question is stupid, I don't get stats...'). Bear in mind that many questioners do not speak English as their native language; think of how difficult it is to learn a foreign language, and then imagine trying to ask, and understand the answers provided to, a very technical question in a language you are not comfortable with. Realize that sooner or later, you will be frustrated with an OP for some reason; always maintain a tone of respect and civility.
  • Make sure your answer is readable: Be sure your answer is edited thoroughly, uses complete sentences, etc. You may also want to use our formatting options such as numbering and/or bullet points in some cases.
  • Check back on your answer from time to time initially. The OP, or other users, may have left a comment requesting clarification etc. Comments that are posted to your answer will show up in your inbox and so you should notice them (assuming you log on), but may also be posted elsewhere by users less familiar with the site, and something worth your attention may occur elsewhere on the thread.

Provide information that is authoritative and comprehensive:

  • We intend for this site to build up a permanent repository of information about statistics
  • Your answer should not just help the OP, but future information seekers who will visit the site (or search Google) with a similar question

Answers that are well connected to other information can be more helpful in both of the above respects:

  • It is ideal to be able to cite scientific references (e.g., papers published in the statistical literature, or books from academic publishers) to back up factual claims that are made
  • Links to other resources on the internet can be very helpful, but due to the mismatch between our goal of permanence and the Internet's dynamic nature (e.g., link rot), you should always include enough information in your answer that it would still be useful if the link went dead.
  • We prefer that you do not simply cut and paste from an authoritative source, however. It is best to provide some additional information to introduce or otherwise connect the quoted material to the current question. Moreover, if you do quote from another source, you need to clearly mark the quoted material as such, and cite the source.

It is often most helpful to illustrate your answer:

  • Once your reputation is above 10, you will be able to upload images into your answer. If you cannot do that now, if you can upload your figure anywhere else on the internet and post a link to it, a higher-rep user can upload your figure for you and post it in your answer.
  • Simulations can often be helpful to illustrate a point, so code (such as R code) can be helpful. However, try to make your code as transparent as possible (e.g., comment more than you think is necessary), bear in mind that not everyone will be knowledgeable regarding your favored language, and make it reproduceable (e.g., be sure to set / fix the random seed).
  • Mathematical proofs are often invaluable (note that CV supports TeXification via mathjax), especially for the subset of questions on mathematical statistics. However, make sure that your proofs are clear enough to be understandable by the OP and likely future readers of the thread. (That is, mathematicians often skip steps that they think are obvious, but what's obvious to you may not be to the OP; also make sure your surrounding discussion adequately motivates and explicates the proof.)

Things to avoid:

  • Be aware the the ordering of the answers will fluctuate over time as they are voted on, so it's better to refer to another answer by that answer's username, not 'the answer above'.
  • Please avoid providing answers that only amount to 'me too!', or that duplicate existing answers on the thread
  • Please don't use the answer field to ask for clarification from the OP, to comment on the question or another answer or respond to others' comments, for help from the community, or to ask an entirely new question

Please bear in mind that, although we believe the quality of an answer will correlate with the number of upvotes it receives, the vote total your answer gets is a random variable. Many of the most highly voted answers on CV are jokes, quotes, cartoons, etc., in part because they attract a lot of page views. It does happen that there are truly superb answers on CV with very few votes because few people end up reading the thread in question, and thus there are few opportunities for votes. In addition, many readers (perhaps unfortunately) will neither upvote nor downvote an answer. Rest assured that we appreciate the help you provide to the OP, and to making CV great. And recognize that even good answers sometimes pick up anonymous downvotes for reasons that are unknowable. Try not to let such things bother you, and maintain civility at all times.

share
1  
+1 @gung Very nicely put. I think we should vote you in as our next moderator when the opportunity comes up. –  Michael Chernick Sep 19 '12 at 15:20
6  
Thanks, @MichaelChernick, but I'm very happy with our current moderators, and have little interest in becoming one. Actually, I have recently taken a new position and, unfortunately, have less time to dedicate to CV. –  gung Sep 19 '12 at 15:24
1  
Although I have proffered an answer to my own question, others should feel free to provide their own answers. Alternatively, if anyone disagrees w/ something here, or thinks something needs to be added, they should feel free to comment & I can make the changes, or even edit this to make the changes themselves. The point here is to get something that reflects the community's consensus regarding good answers, so it shouldn't simply reflect my opinions. –  gung Sep 19 '12 at 17:47
    
I especially like that 2nd bullet under "answer the original poster's question". –  Peter Flom Oct 9 '12 at 0:30
1  
One addition: In some cases, there may be no answer that is within the grasp of the person asking the question. If someone who is having trouble understanding what a t=test is asks a question that demands a multi-level model, what is the best thing to do? –  Peter Flom Oct 9 '12 at 0:32
1  
That's a good question, @PeterFlom. I don't necessarily have a good answer, but I wonder if you want to ask it as a new meta Q, it might make for some interesting & lively conversation. –  gung Oct 9 '12 at 1:16
    
"Be sure your answer is edited thoroughly," ... except of course, if you edit to improve it enough times it becomes community wiki. –  Glen_b Jan 29 '13 at 3:20
add comment

You must log in to answer this question.

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