Many people have explained that they are reluctant to answer questions with a low accept rate. Some are happy sharing their minimum accept rate, others go by feel on a case by case basis, and some choose not to indicate their minimum.

However, the data dump might have this information.

It would take me an hour or two to do what I'm sure someone here can do in minutes:

  • Generate a scatter plot graphing accept rate vs the number of answers a question received

Alternately, probably a little easier:

  • For each number of answers received, get the average accept rate (ie, take all the questions with 4 answers, and get the average accept rate, then take all the questions with 5 answers and get the average accept rate, etc)
  • Or the converse: for each accept rate (say 0%, 10%, 20%, etc) find the average number of answers per question.

If there is a trend we should be able to see it. It won't be conclusive, because there are a lot of variables (perhaps those with low accept rates simply don't post very good questions as well). Further, correlation does not imply causation.

But the data is there, hopefully it's easy to generate, and if so perhaps we can have a discussion using hard numbers regarding what people are doing right now.

It might be worthwhile to split the question data set into two - questions posted prior to the accept rate being shown, and those posted after showing the accept rate. If there's a shift then we can, to some degree, measure how much showing the accept rate has affected answer rate.

If it's not easy data to get, please discuss how to extract this kind of information from the data set, and perhaps some queries that get close - I can do it with a head start, but again it will take me a lot longer than you SQL experts out there.



Browse other questions tagged .