Eye color, unlike Race or Gender, is pretty evenly distributed over the obvious confounding variables like "Age" or "Preference of staying home with children".

Eye color, unlike Race or Gender, is pretty evenly distributed over the obvious confounding variables like "Age" or "Preference of staying home with children". I'd expect it to be +/- 10%, though probably not "equal" enough to keep "disparate impact" folks from calling it out.