oh also I should have mentioned, you add up that chi-squared value per side, to a total deviation for the whole die. It kinda makes sense, imagine a perfectly random set of rolls, and see how close to that you get (or don't)