When I was lamenting the (relative) lack of fluid dynamics questions in the chat yesterday, it was mentioned that I may not be the only one to feel that way given the number of people who answer those types of questions seems to greatly exceed the number who ask.

I was also looking at Map of all Stack Exchange sites (except the three biggest) because I am really interested in data analysis and visualization.

Anyway, that got me to thinking -- has anybody ever looked at the distribution of users and the number of questions asked vs. answered? For instance, I've answered 22 and asked 0. On SO, it's 88 to 30. anna-v is 641-4, Qmechanic is 360-0. Obviously they are the big dogs and may not be representative of the community.

So has anybody looked at these kinds of trends? Are we "top-heavy" in the sense that we have many recurring experts ready to answer but (relatively) few recurring users ready to ask? Is it possible for anybody to get that kind of data (questions asked/answered, number of answers accepted, etc), and if so how? (I answered my own question on this part with the Data Explorer)

I also wonder how physics would compare to other SE 2.0 sites.


I created a simple scatter plot to see what data we would be dealing with, the X axis is questions asked and Y is answers given. I'm looking for techniques on how to show all the different scales of data: https://stats.stackexchange.com/questions/45466/techniques-to-show-data-spanning-multiple-decades but it will give you an idea.

Also, should anybody want to play around with the data and see what they can come up with, the query is http://data.stackexchange.com/stackoverflow/query/88269/question-answer-and-comment-counts-for-each-user It pulls out comments too, maybe there's a cool way to incorporate that information. Cooler than just doing a 3D scatter anyway.

  • $\begingroup$ meta.physics.stackexchange.com/questions/1581/… $\endgroup$ Dec 8, 2012 at 13:16
  • 1
    $\begingroup$ And of course, Piotr links to a github where his code can be found--though you'll need to speak python. $\endgroup$ Dec 8, 2012 at 18:14
  • 3
    $\begingroup$ It's a common phenomenon on many of the SE sites that high rep users don't ask questions. Remember, you're not limited to one or the other. Usually there's a shame/embarrassment factor involved that is strongest in the expert's own head. That said, there's nothing wrong with experts asking and answering their own questions (or asking just for the sake of putting up a question) as long as the question is of quality. I think that most everyone here is capable of that, people just have to step up and ask the questions. $\endgroup$
    – casperOne
    Dec 9, 2012 at 5:52
  • $\begingroup$ @casperOne I agree, the only time it is "problematic" is when the site is top-heavy or bottom-heavy. So looking at data like this could show whether users retained or what the ratio of "experts" to "non-experts" is, both of which are indicative of the health of the site. It may end up not being useful, but it could still be interesting :) I'm certainly not saying that experts who don't ask questions is bad. $\endgroup$
    – tpg2114
    Dec 9, 2012 at 6:00
  • $\begingroup$ Have you looked at the Stack Exchange Data Explorer? It's updated weekly and you can get most of the information you'd need to make your determination there. If you want help writing some of the queries, let me know, I'd be glad to help. $\endgroup$
    – casperOne
    Dec 9, 2012 at 6:03
  • $\begingroup$ @casperOne Thanks, I did play around with it earlier and was able to extract the number of questions, answers, and comments per user. Now I need to figure out the best way to display it, and determine if there is any other information that would be cool to pull out and correlate. There's actually an odd symmetry to the data -- there's a nice balance between people who only ask questions and people who only answer them. $\endgroup$
    – tpg2114
    Dec 9, 2012 at 6:06
  • 2
    $\begingroup$ That's common on most off the sites, you'll find. People who do move along that path move towards the "expert" level but not the other way (they stop asking questions). It's a social problem really, because you can't depend on an influx of new people to ask questions. Best advice I can give is to be encouraging and lead by example. $\endgroup$
    – casperOne
    Dec 9, 2012 at 6:11
  • 1
    $\begingroup$ An example of the data is stats.stackexchange.com/questions/45466/… where I just threw up a scatter plot. The X is questions asked, the Y is answers given. The outliers are particularly interesting and it's pretty easy to guess who some of the data points are. $\endgroup$
    – tpg2114
    Dec 9, 2012 at 6:12
  • $\begingroup$ @tpg2114 I lurk (and occasionally contribute) to Cross Validated as well. I actually saw your question before your comment and knew it was about this case =) $\endgroup$
    – casperOne
    Dec 11, 2012 at 13:05


You must log in to answer this question.

Browse other questions tagged .