# Statistics: What is the average response time on StackOverflow?

What is the average time that passes between when the questions are posted and the answers which get accepted are posted? (not until the answer is accepted, but until the answer that got accepted was posted)

## 1 Answer

http://odata.stackexchange.com/stackoverflow/q/9449/

The average is 3.71 days... but the standard deviation is 25.2 days.

It number is so high because there are some questions like Date arithmetic in Unix shell scripts that take over 2 years to get the answer, creating a strong bias towards a long time. See this query for the top 100.

The median is more sensible — 0.0147 days, which is 21 minutes.

• Seems like a lot of time, but I got the same number: odata.stackexchange.com/stackoverflow/s/453/… – Kobi Aug 17 '10 at 10:28
• The median sounds like the more sensible number. The standard deviation is going to be affected by the questions that don't get any answers at all. – ChrisF Aug 17 '10 at 10:38
• @Chris: Questions that don't get any answers won't be selected by the query. – kennytm Aug 17 '10 at 10:39
• It's interesting that in odata.stackexchange.com/stackoverflow/q/9462/…, if you edit the query's order to be by days descending, you get 9 questions answered within a negative time lapse. This question (stackoverflow.com/questions/1215557) was answered long before it was asked. Among the questions with a positive time lapse, this (stackoverflow.com/questions/179791) was answered in a record time of 4.71 seconds. – Daniel Daranas Aug 17 '10 at 10:53
• @Kenny - I'm clearly not thinking straight this morning. However, truly unanswered questions will affect it as some of these do eventually get answers. – ChrisF Aug 17 '10 at 10:54
• Actually, it is possible there was another accepted answer before the current one, so the real average is lower. We can't tell how common that is, since I don't think that data is resent in the data dump. – Kobi Aug 17 '10 at 11:00
• @Kobi However, if the user first accepted answer A and then switched to accepting answer B, it makes sense to use the time until B in the calculation - we will measure not how long it took to get an answer which was first accepted, but how long it took to get an ultimately accepted answer, which was presumably "the right one" according to the original poster's opinion. – Daniel Daranas Aug 17 '10 at 11:04
• @DanielDaranas: how does that happen? merged questions? – Kip Aug 17 '10 at 12:54
• @Kip: I don't have the slightest idea. What you suggest sounds like a plausible explanation, though. – Daniel Daranas Aug 17 '10 at 14:09
• For completeness, I give the links to the two complementary queries mentioned here. 1) Longest time when an accepted answer is posted: data.stackexchange.com/stackoverflow/query/9452/… 2) Shortest time when an accepted answer is posted: data.stackexchange.com/stackoverflow/query/71842/…. – Daniel Daranas Jun 6 '12 at 12:05