Where a & b are 2 consecutive ingegers, Jefferson, having b as its rounding point is very large-biased, and Adams, with a as its rounding point, is very small-biased.
So it stands to reason that there's some roundng p;oint inbetween that is unbiased. But how could it be? Bias-Free's roiunding points show that, to make a cycle's s/q = 1, each cycle neds a different rounding point. How to reconcile those two arguments? I suggest that there is a point inbetween that will be less biased than the traditional methods. It will be crudely unbiased, with, in some sense the larger states being balanced with the smaller ones, in s/q. But, on a finer scale, it will be biased, and there will be plenty of measured correlation between q and s/q. At least 3 of my 4 methods, and maybe all of them, will be more unbiased than your method. Mike Ossipoff ---- election-methods mailing list - see http://electorama.com/em for list info
