[EM] Set proportionality criterion for divisor methods?

Kristofer Munsterhjelm km-elmet at broadpark.no
Fri Aug 28 23:09:06 PDT 2009


Would this work as an analog of Droop proportionality, but for voting 
methods based on divisor list methods?

With a given divisor p, if X voters vote Y candidates above all others, 
then at least min(Y, f(X/p)) of the candidates in this set should be 
elected, where f is a rounding function (simple rounding off in Webster).

Then, when electing v seats, set p to the lowest possible value so that 
there is at least one set of cardinality v that can be elected. All sets 
that can be elected without contradicting the first paragraph in that 
instance (there may be one, or there may be more) are considered "eligible".

To pass the criterion, the method must elect one of the eligible sets.

//

If everyone votes party line, this should reduce the method to an 
ordinary divisor method. For instance,

100 voters: A1 A2 A3 A4
100 voters: A1 A2 A3
100 voters: A1 A2
100 voters: A1
  50 voters: B1 B2 B3 B4
  50 voters: B1 B2 B3
  50 voters: B1 B2
  50 voters: B1

3 to elect, then the lowest p for Webster satisfies round(100/p) + 
round(50/p) = 3, i.e. p = 40.1 (actually, ever so slightly more than 40, 
but this doesn't matter at this point because the constraints don't 
change until p = 40 on one hand and p slightly lower than 200/3 on the 
other).

p = 40.1 gives A 2 seats and B 1, as expected. For this example, there 
is only one eligible set. If voters had been indifferent as to the 
ordering of their own party's candidates, there would have been more, 
such as {A1, A3, B1}, {A2, A4, B2}, etc.



More information about the Election-Methods mailing list