7

I have invested 150 rep into a question but got no answer. Lately, I invested another 250 rep, but no answer yet. Now I'm curious if my questions are too bad (or too specialized), or if the investment was too low.

How high is the probability per bounty-size to get at least one answer?

3
  • 2
    IMO, it really all depends on the question. I dont think it can really be generalized. – Josh Mein Oct 11 '13 at 17:07
  • 7
    The first comment tells you why you are not getting answers. It is not a programming question (or if it is, it is entirely unclear what it asking). This is not a question that belongs on Stack Overflow and should be closed. – Oded Oct 11 '13 at 17:09
  • 1
    And now, a better question would be; do you get that bounty back now that you drew attention to a question and got it closed? – Johnny Bones Oct 11 '13 at 18:01
8

To try and answer your question about the probability (using obtainable statistics)...

This query gets all bounty answers posted on any Stack Exchange site, in a neat format (quarterly).

Assuming you can get the number of answers existent in the site, you can then get the probability by summing all values:

With this query (for Stack Overflow):

Year Quarter 100  200 300 400 500 500+ Total     

---- ------- ---- --- --- --- --- ---- ---- 

2009 1       856  349 107 28  6   23   1369 
2009 2       1033 447 115 28  4   41   1668 
2009 3       1380 579 149 39  3   39   2189 
2009 4       1678 555 141 45  19  40   2478 
2010 1       2150 772 243 37  7   81   3290 
2010 2       2169 636 183 49  20  44   3101 
2010 3       3473 328 116 29  71  0    4017 
2010 4       3602 285 105 20  75  0    4087 
2011 1       4432 337 133 15  89  0    5006 
2011 2       5019 384 142 25  60  0    5630 
2011 3       5434 486 145 28  130 0    6223 
2011 4       5998 491 215 31  174 0    6909 
2012 1       6736 585 239 39  124 0    7723 
2012 2       6969 528 226 31  123 0    7877 
2012 3       7694 718 267 50  176 0    8905 
2012 4       7699 675 266 37  196 0    8873 
2013 1       8626 683 265 48  160 0    9782 
2013 2       8617 660 271 78  224 0    9850 
2013 3       8469 734 264 57  212 0    9736 
2013 4       480  42  11  3   14  0    550  

The probabilities are the following (assuming there exists 5,867,631 questions and 11 million answers at the time):

Prob Unbountied 100   200   300   400   500   500+  Total    

---- ---------- ----- ----- ----- ----- ----- ----- ----- 

     96,28%     1,58% 0,18% 0,06% 0,01% 0,03% 0,00% 1,86% (per question)
     98,01%     0,84% 0,09% 0,03% 0,01% 0,02% 0,00% 0,99% (per answer)
               84,67% 9,40% 3,30% 0,66% 1,73% 0,25% 100,00% (109263 bounty questions)

So, you can get some conclusions from this:

  • Bounty questions accounts for 1% of total answers given.
  • Most bounties (84,67%) are only +50 or +100 rep, sliding down (with an exception to 500)
  • One can think and assume that most questions are already answered before the need arises to add more bounties to the same questions.

So, with the last reasoning, the best course of action (statistically speaking) is to wait, and then give a +50 or +100 if the views weren't satisfactory.

I assume someone with SQL skills might code this right off the Data Explorer (as for me, my SQL is rusted).

9

There really is no way to say, it's just going to vary too much.

What a bounty gets you is attention. Whether that attention is going to translate into answers, or answers that actually help you, will vary greatly. How clear/understandable is the question? How hard is it to answer? How long would writing an answer take? How many people are qualified to answer it? Is an interested user who comes across it going to be able to figure out the solution through time/research or is it only really possible for a few experts to answer?

Then of course there's just luck/chance as well.

So while the bounty size likely does matter to some degree, there are just so many variables, and these variables all have such a large effect, that trying to look at the statistics of answers during bounty periods of various questions with different bounty sizes is unlikely to be too meaningful, as the effect of the bounty size is likely to be skewed by these other factors.

4

This meta question is perhaps well intentioned, but ultimately not asking a question which really makes sense.

Asking, "how likely am I to get an answer about how to train dogs on StackOverflow?" is just as silly - your question isn't really a programming question.

See this comment.

5
  • 1
    You make an important about this not making a lot of sense for the OP's particular question, but it would be interesting to see some statistics like "average time to receiving an answer after a posting a bounty, as a function of bounty size". – Joshua Taylor Oct 11 '13 at 17:18
  • 1
    @JoshuaTaylor I don't think I've ever had a successful bounty :( – enderland Oct 11 '13 at 17:19
  • OK, we've got one data point. That's a start. :) – Joshua Taylor Oct 11 '13 at 17:19
  • 2
    Another interesting question might be whether bounties are more valuable/effective on some sites than another. E.g., can you spend less on a bounty on math.SE than on SO, and still have the same probability of getting an answer with X minutes. If so, could we establish some reputation exchange where we can trade points from one site for points on another? ;) (That last suggestion is in jest, but there may be something to it. Obviously reputation as an indicator of community trust shouldn't transfer between sites, but perhaps for bounties it could be useful.) – Joshua Taylor Oct 11 '13 at 17:25
  • Aww, looks like people have already discussed that. – Joshua Taylor Oct 11 '13 at 17:28

You must log in to answer this question.

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