Talk:List of Answer Skills/@comment-24850617-20140521030222/@comment-24924403-20140521062820

You need a large sample pool to get the correct result. It took a while before Ai's multiplier was confirmed because some of the data got skewed due to abnormally high results (the average was like 34% or something). Eventually we got lucky and managed to find both the extreme maxima and extreme minima based on our hypothesis. So that's why the intro on testing asks for at least 30 samples. If you want to ensure good results, I suggest learning some kind of spreadsheet to help with the calculation. It really helps.