First of all and just to clarify, the goal of this site is not to collect personal opinions, but facts and best approaches to certain problems.
Given the nature of our subject (academia), this is only an ideal and often the best we can offer are educated opinions.
However, you will not find questions like “Should pay-to-view journals be outlawed?” or similar on this site (if you do, please flag to close as primarily opinion-based).
do the answers here necessarily reflect the accepted answers to questions or problems of whole academic community?
Of course there is a certain bias to the answers and votes given by this community due to its tendency towards computer-heavy fields and people who become active on such websites in the first place.
Here is an example, where a custom or rule completely differs between fields and this wasn’t reflected in the answers for about four years (still, all the previous answers did make appropriate disclaimers, so nobody can complain that we spread false information).
However, quantifying this bias would be largely infeasible:
You would have to make a large-scale survey soliciting the stance of a representative selection of persons from the academic community on a representative selection of questions on Stack Exchange – and ensure that you do not get a bias due to who will participate in such a survey.
That being said, we do not exist in a vacuum.
If there is a huge amount of people with differing opinions, some of them are bound to eventually stumble upon our site, and some of them in turn are bound to tell us that we are wrong on the Internet.
(And just in case we are censoring any dissent, somebody is bound to start a website informing the Internet about this – which hasn’t happened yet as far as I know.)
Now, for most of our questions, the general directions of answers align and votes merely indicate which answer presents the best reasoning or simply was first.
Cases with strongly dissenting answers are rare, and it rarely happens that somebody joins our site just to tell us how wrong we are.
I would consider this a good indicator that the aforementioned bias is not a huge issue when it comes to the correctness of answers.