[EM] Weighted Bias-Free
Michael Ossipoff
mikeo2106 at msn.com
Thu Jan 18 05:07:26 PST 2007
Weighted Bias-Free allows for a nonuniform state-size probability density
distribution.
Its formula is based on an assumed distribution function of the form:
B/(q+A).
A & B would have to be determined by measuring the aveage density of states
in a number of regions of the population range, and then using least-squares
to fit B/(q+A) to those densities.
When A and B havew been chosen, the Rounding point of Weighted Bias-Free
(BFW) is:
((b/(b+A))^b/(a/(a+A))^a)*((a+A)/(b+A))^A
This problem had many opportuniies for error, and I'm not so arrogant as to
claim that I couldn't make an error. Butthe formula has the look of being
right. It isn't possible to test the reasonableness of its rounding points
till A has been chosen. B doesn't affect the rounding point, but is part of
making the curve-fit.
Mke Ossipoff
More information about the Election-Methods
mailing list