We are almost half-way through our beta. (As of today, there are 48 days left out of 90 total days.) This seems like a good time to take a step back and see how we're doing overall.

  1. How is the site developing? Are people happy with the community and the questions that are being asked?
  2. Are there any particular things that we should focus on for the remainder of the beta period?
  3. How has the moderation been? Any feedback is welcome.

One thing that I've been a little disappointed by is the relative scarcity of real-data questions. There are lots of questions about methods/tools for X and many others asking for clarification of some statistical concept; but not so many where we actually get to look at some data as a group. If there's interest in that beyond me, maybe we could facilitate that by having a way to host data alongside a question (yeah, there are a million file hosts, but do people know how to use them?). Maybe this isn't the right venue for that kind of question, though.

Other than this particular little desire, I agree with Srikant's post.

  • $\begingroup$ I think that's a great idea. I would love to see more real problem solving on the site. $\endgroup$ – Shane Aug 31 '10 at 21:09
  • $\begingroup$ the secretive nature of my work prevents me from sharing any real data, although I could come up with synthetic data possibly... $\endgroup$ – shabbychef Sep 1 '10 at 17:40
  • $\begingroup$ I've got a similar problem, chef - which is why we've got to prod everybody else into doing it :p $\endgroup$ – Matt Parker Sep 1 '10 at 18:15

I could be wrong but the 'rate of question flow' seems to have slowed down off late. I think the quality of questions and answers is fine although voting levels still seems a bit low.

I think for the remainder of the beta period we really should figure out how to increase traffic. Of the promotion ideas that we came up with on "How to promote the site?", Tal's idea on "Stat flashmob week" seems the easiest one to implement in the short term.

Could we compile a list of bloggers who may be receptive to the idea? (Andrew Gelman has already blogged about the site here.)

Reg moderation- I do not think we have had any serious issues so far. So, your mettle as a mod has not yet been tested! :-)

  • $\begingroup$ A "flashmob" is a great idea. I have noticed the same rate decrease. At some point, we will his a "natural" level, and I don't think it should necessarily be more than 10-20 questions a day, but during the beta we should make an effort to push the limit. $\endgroup$ – Shane Aug 31 '10 at 15:50
  • $\begingroup$ I agree, the flow of questions is a bit slow. Perhaps we should all start to submit some questions (even when we know the answer). I would also suggest trying to avoid "wiki" type questions. $\endgroup$ – csgillespie Aug 31 '10 at 20:35
  • $\begingroup$ I would rather that we focus our energy and time on getting new users to the site. A steady stream of new users is what we need. If we continue to post questions/answers then the site will look like a community of 10 users who talk to each other. I think we have enough seed questions as of today. $\endgroup$ – svadali Aug 31 '10 at 23:51
  • $\begingroup$ @Srikant: I actually think that these things go hand-in-hand. Having more good questions will be one of the best marketing tools that we can have. And a flashmob does more than bring questions: it brings people and excitement. I had never used stackoverflow before the R flashmob, and now I'm the second biggest contributor under the R tag. $\endgroup$ – Shane Sep 1 '10 at 13:45
  • $\begingroup$ AS to the 'list of bloggers', J.D. Long is a good candidate, I think. cerebralmastication.com $\endgroup$ – shabbychef Sep 1 '10 at 17:45
  • $\begingroup$ @srikant we should be able to measure the 'rate of question flow'. with statistics! $\endgroup$ – shabbychef Sep 1 '10 at 17:54
  • $\begingroup$ @shabbychef yes there is opportunity for us to analyze our own performance! But, I think we do not have access the raw data. We would need the raw data for not only our site but comparable sites such as math.stackexchange.com, mathoverflow, cstheory etc. $\endgroup$ – svadali Sep 1 '10 at 18:28
  • 1
    $\begingroup$ That's currently possible with the overflowr package (see related question). stackoverflow.com/questions/3533334/… And I reiterate my point from here: please feel free to contribute to this package! meta.stats.stackexchange.com/questions/328/… $\endgroup$ – Shane Sep 1 '10 at 18:54
  • $\begingroup$ @shane I do not know enough about R to contribute which may also explain why I am having issues compiling the package on my windows platform! $\endgroup$ – svadali Sep 2 '10 at 13:58
  • $\begingroup$ Got it; sorry! I'll post instructions over the weekend. BTW. I think that it's time for you to learn R! :) $\endgroup$ – Shane Sep 2 '10 at 14:40
  • $\begingroup$ @shane Please drop in a comment as and when you get a chance to post some instructions. Learning R is on my agenda and I may use this package as an excuse to move that agenda forward! $\endgroup$ – svadali Sep 6 '10 at 23:35

One thing I noticed is at least some kind of overlap between the stats and the stackoverflow for programming. There are some questions that appear on both sites in different forms; on stats the method is asked, and then on stackoverflow a correction of the code is asked.

Example : on stats and on stackoverflow

No clue as to how to solve this, as I believe it's not possible to merge questions from different sites. One solution is to direct all R questions to the stats, but that solution seems not satisfactory to me. For one, many questions on R are about data manipulation, and far from every question on statistics is asked by somebody who uses R.

For the rest I am a happy user of stats. I didn't even realize it was a beta version :-)

  • 1
    $\begingroup$ I was happy to have been chosen as the overlapping example :-) But R is struggling to overcome matlab on stack and needs more questions :-P $\endgroup$ – George Dontas Sep 1 '10 at 15:16
  • 1
    $\begingroup$ I (for one) wouldn't mind if more of the stackoverflow questions came over here. Especially if they aren't purely programming questions. $\endgroup$ – Shane Sep 1 '10 at 15:31

Site content, moderation, as well as aims and scope seem ok with me.

I am just wondering if it is/was planned to automatically tag the software refered to in each Q&A. From what I already saw, the top used softwares are R (hopefully!), SPSS, Matlab; I don't consider MS Excel as a statistical software, though. Stata users seem not be very represented there; they have their own mailing-list, but R users too. It may be, however, interesting to see if solutions are available with different software, since it may increase the popularity of the site.

  • $\begingroup$ You mean, to read the answer/question and automatically tag it based on the code? I don't think that's planned; it would be fairly difficult to implement. But having people provide multiple different language solutions is a nice idea. $\endgroup$ – Shane Aug 31 '10 at 15:53
  • $\begingroup$ Yes; this is proposed for syntax highlighting in some blogging system, via Pygments, Coderay or things like that. $\endgroup$ – chl Aug 31 '10 at 15:58
  • $\begingroup$ Wow. That would be pretty neat. Feel free to make a feature request either here or on meta.stackoverflow.com. $\endgroup$ – Shane Aug 31 '10 at 16:03
  • $\begingroup$ Actually, we have to add manually a language type to the code chunk in markdown, e.g. {% highlight python %}; so this would rely on users goodwill. However, it could provide both syntax highlighting and tagging. I'll ask for the feature and see how users react to it. $\endgroup$ – chl Aug 31 '10 at 16:21
  • $\begingroup$ then a moderator could edit questions to add the highlighting, assuming the rate of new questions is not so high. $\endgroup$ – shabbychef Sep 1 '10 at 17:41

You must log in to answer this question.

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