10
$\begingroup$

Anecdotally I've noticed that there seems to be periods of time (during the day) when we get lots of questions asked, and then periods where lots of answering is done. Before making up lots of ideas why this might be the case (Students/Professionals?), it would be interesting to find out if it actually is true.

I understand that there is a SQL database type way of looking at activity on a stack exchange site - so if someone in the know could explain how to answer this that would be great. Anybody got the SQL skills to figure out if questions, comments and answers have different intraday profiles?

$\endgroup$
  • 2
    $\begingroup$ Mods have easy access to daily activity measurements. They exhibit strong weekly fluctuations, as you might expect, as well as large seasonal fluctuations clearly tied to major holidays and school schedules. Without digging into the API, though, I cannot readily tell you anything about diurnal fluctuations. In looking over these charts, one thing is striking: our site has experienced an unprecedented, sudden fourfold leap in "feedback by low rep/anon users" starting February 12. There has been no concomitant change in other metrics. $\endgroup$ – whuber♦ Mar 1 '13 at 19:48
  • $\begingroup$ What counts as "feedback by low rep/anon users"? $\endgroup$ – Corone Mar 1 '13 at 19:49
  • 1
    $\begingroup$ I don't know precisely. I have noticed that when I log in anonymously from another computer, I have no privileges to vote, comment, etc., but there are tools beneath questions and answers inviting me to indicate whether or not I found them helpful. I presume this metric is tallying such responses. $\endgroup$ – whuber♦ Mar 1 '13 at 19:53
  • 1
    $\begingroup$ Hmm.. interesting - I wonder what effect those tools have? Strange that it started now - part way through academic year. Maybe Stats courses get hard at this point :-) $\endgroup$ – Corone Mar 1 '13 at 20:03
  • 1
    $\begingroup$ There was no similar jump at this time last year. (Smaller jumps in overall activity when semesters begin and exams begin can indeed be detected year after year.) The sudden increase in low rep feedback--it's around an extra 60 per day--could be accounted for by a single large class somewhere or a single good Web link, I suppose. $\endgroup$ – whuber♦ Mar 1 '13 at 21:25
  • 1
    $\begingroup$ That in intriguing, @whuber. What happens w/ that information? Does it get incorporated into votes somehow? Do mods or the SE team do something w/ it? $\endgroup$ – gung - Reinstate Monica♦ Mar 1 '13 at 22:07
  • $\begingroup$ @Gung I understand the SE team looks at these low-rep feedback stats, with the eventual aim of perhaps using them to improve the site. The low-rep feedback is relatively new (about a year old or so). The only indication mods have gotten so far is this ability to chart these statistics; they don't seem to show up anywhere else. $\endgroup$ – whuber♦ Mar 1 '13 at 22:11
  • 1
    $\begingroup$ @whuber "Feedback from anonymous users" is collected when an anonymous user votes on a question either up or down. Moderators and top users have access to that information here. This was NOT how this information used to be collected -- we changed it from the "Did you find this useful YES NO" collection style (see Google Support pages for an example) that we used to use. By changing the collection style, we found we were getting MORE feedback. $\endgroup$ – Aarthi Mar 11 '13 at 21:59
  • $\begingroup$ Corone: It might not just be students/pros -- there's a likely change it's also linked to geography. Perhaps European students are asking quesions mostly in the US "morning" while East Asian and/or Russian respondents are providing answers during their morning, which looks like "evening" from a US standpoint. As a global site, there are other potential causalities. (This is the part where I jokingly tell you to start adding dummy vars to your regression. :P ) $\endgroup$ – Aarthi Mar 11 '13 at 22:01
  • $\begingroup$ Thanks, @Aarthi. I still don't see any links there from the moderator tools page. Am I just overlooking them? BTW, did you change your collection style about a month ago? That's when the sites I moderate experienced a 3-4 fold jump in responses. $\endgroup$ – whuber♦ Mar 11 '13 at 22:26
  • 1
    $\begingroup$ @whuber Yes to the second question; to the first, it's not a mod tool link -- it's actually linked from /review! Click on review in the topnav, then click tools underneath the CV logo, then scroll to the bottom and you'll see a anon feedback link. $\endgroup$ – Aarthi Mar 11 '13 at 22:29
  • $\begingroup$ @Aarthi: How come anon users get to vote up and down? I had to wait ages to get that previlage! :) $\endgroup$ – curious_cat Mar 15 '13 at 18:09
  • $\begingroup$ @curious_cat Hahaha, it's just how we collect anon feedback! No scores affected, just to be clear. :D $\endgroup$ – Aarthi Mar 15 '13 at 18:19
  • $\begingroup$ @Aarthi: Whew! Thanks. :) $\endgroup$ – curious_cat Mar 15 '13 at 18:22
5
$\begingroup$

The following SEDE query shows that most activity is between 15:00 and 15:59 UTC.

Here is the graph:

time of creation of posts

Asking and answering follow the same trend but in the evening a bit more answers are posted.

$\endgroup$
  • 1
    $\begingroup$ The time of day effect is surely confounded with whatever numbers of questions arise from different longitudinal zones. For example, a peak around 15:00 UTC catches Europeans in mid-afternoon but also those 6 hours earlier and 6 hours later in time zones 6 hours away. $\endgroup$ – Nick Cox Jul 8 at 8:40
  • $\begingroup$ @NickCox that is true. We don't have the (exact) location of users in SEDE so it is impossible for users of the public SEDE instance to analyze further and I doubt the private SEDE instance for SE staff has better data on that. One thing I considered adding was to see what the distribution (not sure if that is the right term) was for time to first answer when asked at a certain time. But I realize I'm on Cross Validated and wandering off in an area I've no background on might cause a few laughs. So I leave further analyzes to the crowd here. You're capable of doing that. $\endgroup$ – rene Jul 8 at 9:02
  • $\begingroup$ Also of interest stats.meta.stackexchange.com/questions/3106/… and stats.meta.stackexchange.com/questions/3046/… $\endgroup$ – mdewey Jul 8 at 17:47
  • $\begingroup$ @mdewey those seem practical duplicates, better close this one as such. $\endgroup$ – rene Jul 8 at 19:00

You must log in to answer this question.

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