I made a few statistics using DataExplorer on the main SE sites, and a few others sites taken randomly (from the ones I visit from time to time):
QuestionCount AnsweredCount AcceptedCount Accepted/Answered
tex 123789 104753 75154 71.7%
mathematics 691306 582320 365594 62.8%
stackoverflow 12804240 11169780 6976659 62.1%
programmers 42679 40457 25164 62.2%
unix 103348 86958 50792 58.4%
serverfault 232025 208532 117444 56.3%
electronics 70754 66819 37490 56.1% ***
english 77106 72596 40655 56.0%
itsecurity 34928 31910 17177 53.8%
superuser 332158 271327 142278 52.4%
workplace 12807 12582 6429 51.1%
arduino 9063 7528 3639 48.3%
parenting 4449 4287 2065 48.2%
raspberry pi 15810 12206 5717 46.8%
askubuntu 246617 195549 83889 42.9%
We're actually on the average.
We're certainly not on the top, maybe for the following reasons (but that is very subjective):
- People coming here to ask questions are often unaware of how this site works. They come because they want to know "how to wire two led strips on the same power supply" (we get a lot of questions like that), and they don't really bother looking how the community works. Once they have their answer, they never come back. I think it is less common for stackoverflow, because newbies starting to get into development know that they'll probably have more questions in the future. And I'm not really familiar with askubuntu, but I think they certainly get a lot of people asking a single question, as well, and they have a very bad acceptance rate.
- There is another category of questions from newbies in the lines of "is it possible to do this [whatever impossible thing to achieve, or extremely difficult system to design]". In which case the typical answer is negative, and explains why. Although the answer is correct, it will certainly be more difficult for the OP to accept it (because it also means accepting giving up on his idea). I'm not sure it occurs as often in other fields.
- There is also a strange thing happening regularily on electronics that I noticed, is that when we answer, we often provide lenghty information about very technical details that are insignificant to the OP (and may actually confuse him). As if we wanted at all cost avoiding making "approximative enough" answers or omitting some specific cases (probably because we don't like somebody else making a comment telling that we forgot a point). The problem is that OP, if he gets confused by the answers, will then have hard time telling which one helped him the most. It doesn't happen every time, but I think it is more the case on this site than on others.
Now, going back to your question: "How could users be more encouraged to hit the button and therefore help others know which answer was really helpful?". Actually, I think there is no real problem. The system, as it is, works well enough. The "which answer is the most accurate?" feedback is, anyway, mostly given by the upvotes, not the accept flag.