Quote Originally Posted by Shougun View Post
Definitely still easily flawed data but I just I want add in statistics you can very successfully extrapolate info from very small subsets. Of course the more data the better, especially if there was an unforeseen flaw, but you can see something like 1% accurately predict 99%. Surprisingly small samples can be used for exceptionally accurate results. Then again given the nature of the -insert anywhere- places of feedback... it does obviously move about the data. People for example are more likely to come here if they have something they want to change rather than something they want to praise. This is where someone would make a mistake if they just blindly followed feedback, but if the players can accurately articulate why they want given feedback I believe some of the consequences of listening to a small portion of the community can be alleviated.
The problem with this is that it is an opt-in sample, which fundamentally biases the data.

You need a truly random sample to do any sort of meaningful extrapolation from small sample sizes - which sadly is not the case for the forums.