As I said, I propose 4 methods:
Cycle-Webster (CW)
Adjusted-Rounding (AR)
Weighted-Webster (WW)
Webster (W)
Advantages of CW & AR over W & WW:
The equal s/q for all cycles, and the resulting low bias correlation in
tests, of Webster and Weighted Webster is of a statistical nature. But
Cycle-Webster and Adjusted-Rounding deliver that advantage in every
apportionment.
CW & AR are mathematically simpler than WW. Their completely transparent
achievement of the goal of equal s/q for all cycles, resulting in equal
representation expectation for all, can easily be explained and demonstrated
to anyone.
Advantages of W & WW over AR & CW:
W & WW are true divisor methods, making them more traditional, and maybe
more acceptable to people. It gives them the properties proven for divisor
methods, if thats considered important.
Advantages of Webster:
Webster is the simplest of those 4 methods, and it has precedent in use.
Its one of the traditional divisor methods.
Though, without uniform frequency distribution, Webster will allow some bias
correlation in tests, I claim that such a test result isnt so unfair when
its caused entirely by the distribution, as opposed to being caused by the
method itself. Webster has no intrinsic bias.
Mike Ossipoff
----
election-methods mailing list - see http://electorama.com/em for list info
