11
$\begingroup$

If one purpose of the site is to promote good statistical practice, why are tutorials not part of our offerings?

It seems many users could benefit from well-written tutorials on subjects such as regression, expectation maximization, maximum likelihood, the delta method, survival analysis, etc., just to name a few.

I'm wondering why this functionality hasn't been proposed. Perhaps the idea has been discussed before and discarded - I searched a bit, but didn't find anything.

I think it is worth considering as it could attract many more users if we provide excellent tutorials in areas that are commonly used by those needing certain types of technical analysis.

Why not become the go-to site for best practices by posting tutorials? We could vote up and keep the best ones and then reap the benefits as users share with others where they learned the methodology.

$\endgroup$
  • 3
    $\begingroup$ Could you tell us how your vision of a tutorial differs from existing mechanisms, especially self-answered questions and our blog? $\endgroup$ – whuber Mar 6 '14 at 4:32
  • 2
    $\begingroup$ @Glen_b spells out that this is possible in principle and occasionally done in practice. Why not done more? A good answer might take somewhere between 2 minutes and 2 hours to write, and that's to many people acceptable, especially if one wants to see the answer oneself! Some of the more spectacular answers probably take much more. But to me a tutorial, even on territory on which I am comfortable, would take more like 2 days or 2 weeks, and I won't give the time, or more precisely I would give the time but it will end up as something I should be doing any way (e.g. a paper or lecture). $\endgroup$ – Nick Cox Mar 6 '14 at 11:08
  • 1
    $\begingroup$ For whuber's question, I envision a separate area of CV specifically geared towards topical instruction. I view it as a reference, and the difference would be that the approach is not specific to a particular user's question/problem. For me, if I need knowledge of a statistical area, I usually go to a text or a reference like $\it{Kendall's \ Advanced \ Theory \ of \ Statistics}.$ Online, I try Wikipedia first. Both of these are valued sources, but I think we could provide another one that gives general instruction (and worked examples, too) of great value to the general users. $\endgroup$ – soakley Mar 6 '14 at 14:22
12
$\begingroup$

There are many ways to promote good statistical practice. For example, one-on-one tutoring by experts would likely be invaluable to those receiving the tutoring, if a somewhat inefficient use of expert time. We may not be able to pursue every form of encouraging good practice within the goals of the SE network. CV's central purpose is a statistical question-and-answer site, so we need to keep that in mind.

Within that central purpose, I see two obvious ways to have some form of tutorial using the features of SE that are already available:

  1. The blog. While the blog hasn't updated in a long time, several of the blog posts have elements of tutorial in them.

    Tutorial posts aren't explicitly listed on the about page, I think it wouldn't take much to convince people to add it - currently the blog isn't being used at all, so why not?

    An explanation about the blog and how you get to post to the blog is here. You might have to get the mods to revive the chat mentioned there.

  2. Questions and answers. If you see a burning need for a tutorial, you can frame it in the form of a question and answer (it's completely legitimate to answer your own questions).

    Such a situation came up for me a while ago, when in the course of answering a question I needed to use interpolation, and realized the OP wouldn't know how to do that. (I'd previously answered several other questions with the same problem -- but to explain interpolation in a post which was primarily about something else would be getting too far off track.)

    So I decided to put up a tutorial on interpolation in tables in the form of a question and answer - one that I (or anyone else) could link to whenever needed, rather than giving an overly-brief half-baked explanation every time or pointers to less directly relevant material off site.

    There are a few such questions-with-answers written by other people on the site (and I think we could stand to have more).

    [If you're looking for tutorials with broader participation on a single good answer rather than a potential plethora of answers, you might even seek to set up some community wiki questions for the purpose.]

Shorter tutorials would be well placed for Q&A, longer tutorials would perhaps be better as blog posts. (But the 'post of the week' option on the blog allows for highlighting of questions/answers on the blog as well.)

If you do either of those things (organize blog post-tutorials or Q&A tutorials), or manage to get others to do some (I'd love to see a few tutorials from some of the experts), I would be cheering.

Indeed, if there's a topic that there's an obvious need for a tutorial on, I could perhaps contribute one, or contribute to one, as long as it's something I know a little about.

$\endgroup$
  • 2
    $\begingroup$ @gung thanks for the edit. Your expression was better-chosen than mine. $\endgroup$ – Glen_b Mar 7 '14 at 1:12
5
$\begingroup$

Doubtless many users could benefit from following well-written tutorials, but I'm not sure it's often owing to the paucity of these that some (seemingly) haven't yet. Your examples are covered in numerous blogs, on-line lecture notes, textbooks, &c. On the other hand resources to answer specific questions are a lot scarcer, & we all must know how frustrating it can be to look something up in half a dozen different places & still not understand it. So perhaps people feel they're spending their time more usefully by answering ten unique-ish questions on Kaplan–Meier, hazard ratios, &c., than by writing one more "how to" for survival analysis.

$\endgroup$

You must log in to answer this question.

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