24
$\begingroup$

Whoever maximizes the ratio $$ \frac{\text{People Reached}}{\text{Questions Answered}} $$ must have an uncanny sense for which questions are going to be heavily viewed and worth answering. In some sense, they use the site most efficiently (if your goal is to reach people). Who is this person? What is their "impact factor"?

I wonder if one of the database wizards would be interested enough in this silly query to code it up. Perhaps the answer turns out to be uninteresting because someone created an account just to answer a single popular question, in which case maybe there should be a requirement that the winner has answered at least five questions or something. A related question might be to replace "People Reached" with "Reputation".

$\endgroup$
10
  • 23
    $\begingroup$ this sort of thing tends to have a maximum for some user you won't recognize who answered exactly one question that happened to reach many people. The usual practice (with sports, for example) is to demand a substantial minimum number of occurrences of the denominator. $\endgroup$
    – Will Jagy
    Apr 6, 2017 at 21:03
  • 1
    $\begingroup$ This post contains SEDE query which should return people reached for an individual users: Include the “People Reached” and “Helpful Flags (count)” data in SEDE. $\endgroup$
    – Martin Sleziak
    Apr 6, 2017 at 23:15
  • 5
    $\begingroup$ @JonasMeyer deleted answers do not count towards people reached. see e.g. this thread $\endgroup$
    – quid Mod
    Apr 6, 2017 at 23:19
  • 1
    $\begingroup$ Following up on @WillJagy's remark a way to find somebody like this could be to search among the highest view threads for users with unique answer there of score at least five. $\endgroup$
    – quid Mod
    Apr 6, 2017 at 23:24
  • 10
    $\begingroup$ It's going to be massively skewed towards people having answers to questions that went viral. A question like the Batman equation has 421528 views so if you have an answer here then you will for sure be in the top. ...and so it is: ShreevatsaR (5130), who has the accepted answer, has the greatest ratio among all users with more than 250 answers. If we raise the bar to 500 answer it's J.M. (3530) who also has an answer there. If we only consider people with more than 1000 answers then it's Arturo (2420). $\endgroup$
    – Winther
    Apr 7, 2017 at 12:29
  • 2
    $\begingroup$ If you replace people reached by reputation, it is much easier and there probably already are many variations of such query. Just try to search a bit, for example, results of this search seem reasonable: average score top site:data.stackexchange.com. From the ones I've tried, Top 500 answerers on the site seems quite good. $\endgroup$
    – Martin Sleziak
    Apr 7, 2017 at 16:35
  • 3
    $\begingroup$ @JonasMeyer Questions also count towards the number of people reached, so it is possible to have reached a positive number of people reached with 0 answers (but a positive number of questions). $\endgroup$
    – wythagoras
    Apr 13, 2017 at 9:45
  • 2
    $\begingroup$ @wythagoras: That should answering the question easy: Just seek the people who have zero answers but more than zero questions. Their ratio then is infinite, which is as large as it can get. ;-) $\endgroup$
    – celtschk
    Apr 16, 2017 at 10:25
  • 1
    $\begingroup$ @celtschk Dividing by zero is heresy, not infinity. $\endgroup$
    – wythagoras
    Apr 16, 2017 at 10:52
  • $\begingroup$ @WillJagy I wonder if this can be mapped to something like Wilson score confidence evanmiller.org/how-not-to-sort-by-average-rating.html which "balance[s] the proportion of positive ratings with the uncertainty of a small number of observations". I've also asked at: stats.stackexchange.com/questions/376361/… in the hopes that someone there will know. $\endgroup$
    – Ciro Santilli OurBigBook.com
    Nov 10, 2018 at 20:54

1 Answer 1

Reset to default
6
$\begingroup$

I amalgamated some queries in order to get what OP wants:

This query does what was requested.

Some observations:

  • It only counts undeleted answers;
  • As was expected, without some lower bound on the denominator the results are pretty skewed to singularities: on the top 100 with the requirement of at least 1 answer, the highest number of answers is 3;
  • There is some error in the calculation of "people reached". The original query which I borrowed this calculation from is here;
  • Requiring a minimum of 100 seems to start giving more significant results.
$\endgroup$
0

You must log in to answer this question.

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