According to my profile page, I've voted on 1100 questions and voted on 800 answers. If Q is the number of question votes I've given and A is the number of answer votes, then $Q-A=300$. What is the distribution of $Q-A$ across all users? What about looking at just upvotes?

  • 2
    $\begingroup$ My intuition was more upvotes for questions than for answers. It's easier to tell a good question from a bad question than it is to tell a good answer from a bad answer. But Woodface's data shows my intuition was wrong. $\endgroup$
    – user153918
    Mar 19, 2015 at 19:01
  • $\begingroup$ @Alonso: There are many easy to judge answers to not very good questions. $\endgroup$ Mar 27, 2015 at 20:17
  • $\begingroup$ I am fairly new to stack exchange and usually upvote answers that I have learned something from. That is far more than the questions which intrigue me, but I will perhaps consider upvoting more questions now. $\endgroup$
    – Amy B
    Jul 10, 2015 at 19:43

1 Answer 1


As of March 15, 2015, Math.SE users cast

With about 410k questions and 599k answers, the averages are:

  • $2.2$ upvotes per question
  • $2.5$ upvotes per answer
  • $0.13$ downvotes per question
  • $0.05$ downvotes per answer

Only the votes on non-deleted posts are counted above.

The averages given above mask large differences between individual voters. Back in December 2014, I compiled some statistics for the 17 users who cast at least 10000 votes by that time. The chart distinguishes question votes (blue) from answer votes (green); it includes both up and down votes.


  • $\begingroup$ So this is interesting to me because it means I am somewhat atypical in pivoting questions more than answers. $\endgroup$
    – abnry
    Mar 17, 2015 at 20:34
  • 1
    $\begingroup$ I added a chart that shows there is a lot of variability there. $\endgroup$
    – user147263
    Mar 17, 2015 at 20:51
  • 1
    $\begingroup$ what do you use to generate the chart? $\endgroup$ Mar 17, 2015 at 21:01
  • $\begingroup$ @frogeyedpeas This one was 100% manual. I looked at the list of top voters, navigated to each profile and recorded the votes in a spreadsheet (OpenOffice or Excel, don't remember). The Q vs A split of votes is not available in Data Explorer. This is why I'm showing the data from December: don't feel like doing this exercise again just now. $\endgroup$
    – user147263
    Mar 17, 2015 at 21:08
  • $\begingroup$ Is there any scope of say hooking into the stackexchange API and generating these sorts of tables in an automated fashion? $\endgroup$ Mar 17, 2015 at 21:14
  • 2
    $\begingroup$ @frogeyedpeas Not the Q/A split. Up/Down is available: both counts are returned by /users API method. It still takes work to handle the data, in any case. $\endgroup$
    – user147263
    Mar 17, 2015 at 21:17
  • $\begingroup$ If you're interested here's an updated version of the chart (I didn't know about that API thing, that's great). $\endgroup$ Mar 27, 2015 at 8:05
  • $\begingroup$ Here's also a chart of the up/down votes from the top voters (I'm too lazy to sort by number of votes...) The results aren't very surprising. $\endgroup$ Mar 27, 2015 at 8:39
  • $\begingroup$ Woah, @Woodface's downvote's. $\endgroup$
    – abnry
    Mar 29, 2015 at 18:22
  • $\begingroup$ As of September 18, 2017: $1734358$ upvotes on questions, $2768815$ upvotes on answers, $143827$ downvotes on questions and $68408$ downvotes on answers. Seems like we have become more strict with the quality of questions recently. $\endgroup$ Sep 18, 2017 at 20:07

You must log in to answer this question.

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