15
$\begingroup$

Consider the reputation graphs of stats.stackexchange in a log-log scale:

enter image description here

This seems to mean that very few users answer most of the questions, at least compared to stackoverflow:

enter image description here

See this other query of quality of answers, you can see that there are a few users with more than 1200 answers, and one with more than 3k answers.

Is there any other data confirming that few users answer most of the questions, or am I completely wrong?

You can look at the data of the last year answers per user and reputation. 4492 users answered questions, many just 1 or 2 and 1 up to 700 answers!
enter image description here

$\endgroup$
7
  • 11
    $\begingroup$ It may be worth pointing out that these curves are not "histograms." They are log-log plots (base 10) of reputation bins (on the horizontal axis) and bin frequencies (on the vertical axis). Also of note is that the most extreme four points for CV (at the lower right) represent just four people--implying they are highly uncertain indicators of the population characteristics--while the extreme four points for SO represent 95 people. It is also worth observing that these plots do not directly represent answer rates: they are total reputation. $\endgroup$
    – whuber Mod
    Commented Nov 24, 2016 at 14:44
  • $\begingroup$ I would like to know it too. Ask the editor $\endgroup$
    – llrs
    Commented Nov 24, 2016 at 20:08
  • $\begingroup$ @gung The table that you copied here looks very strange. What are these numbers? $\endgroup$
    – amoeba
    Commented Nov 25, 2016 at 8:07
  • $\begingroup$ I didn't put that pictures, and are created using the open data of stackexchange, which don't include legends and titles in the graphs it automatically creates. $\endgroup$
    – llrs
    Commented Nov 25, 2016 at 11:13
  • 1
    $\begingroup$ @AndreSilva, if you look at the revision history, you'll see that I added the figures. The original post relied on a lot of information posted elsewhere (SE data pulls) that required readers to navigate elsewhere & read other materials to understand the question fully. I copied the figures to try to make the question self-contained. $\endgroup$
    – gung - Reinstate Monica
    Commented Nov 25, 2016 at 13:39
  • $\begingroup$ @amoeba, I don't know what happened with the table. I must have messed something up & not noticed. $\endgroup$
    – gung - Reinstate Monica
    Commented Nov 25, 2016 at 13:40
  • 4
    $\begingroup$ For future reference, @Llopis, any question posted on the SE system (eg, on CV or meta.CV) is supposed to be self-contained. It is OK to link to other sources for context, but whatever information is necessary to understand & answer the question is supposed to copied into the body of the question. We want these questions to remain valid & intelligible even after the link goes dead, or the data from which the pull is drawing changes. $\endgroup$
    – gung - Reinstate Monica
    Commented Nov 25, 2016 at 13:45

2 Answers 2

Reset to default
16
$\begingroup$

One problem with looking at the total number of answers to look at whether a few users answer most questions is it doesn't tell you whether someone is answering questions now. Someone may have a thousand answers but not have answered a question in years.

(Yes, this may mean that the "few users" issue is even fewer than you suppose)

So I'd look at something like answers in the last year, perhaps.

But yes, a lot of answers on site are concentrated in just a few hundred users, and somewhat more concentrated than on stackoverflow.

[However, that doesn't indicate that the users who answer most here are heavier answerers than on stackoverflow. I answer a fair few questions here, but it's still only an average of about 14 a week, or about two a day -- more on the weekend, somewhat fewer during the week -- but heavy answerers on SO answer more in a day than I do in a week (don't ask me how, though average answer length may be a factor).]

I think at least part of the reason for that is it takes considerably less learning time to get to the point of giving reasonable answers to typical SO questions than it does here; the entry barrier is lower there. As a result, while there's something like a hundred times as many questions posted per day, there's quite a lot more than a hundred times as many regular answerers.

[I'd very much like to expand our base of users who answer questions regularly, but I'm really not sure of a good way to do that.]

$\endgroup$
2
14
$\begingroup$

I've generated graphs that more directly answer your question here.

That was over two years ago, but you could update with newer data if you want. Two years ago we were in the middle of the pack. The Math and Mathematica site were more dominated by a smaller number of users, and theoretical computer science is the most even.

Also see this question about generating Gini scores for sites. For that metric the cut-off you use to count users matters a lot.

$\endgroup$

You must log in to answer this question.

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