[EM] Meek and Newland-Britton

James Green-Armytage jarmyta at antioch-college.edu
Sat Sep 11 14:09:18 PDT 2004


Dear David Gamble,

Are you sure that Newland-Britton elects AB? I'm not an expert on STV
formulas, but it seems to me that they both elect AC. Here's how I
calculated it... I'm assuming that we're using the Newland-Britton quota
of votes / (seats + 1). I'm doing a round-by-round count here. "rf" stands
for retention fraction, and "tf" stands for transfer fraction.

A 61
A>B 60
C>D 50
D>C 9

Votes: 180. Seats: 2. Newland Britton quota: 180 / 3 = 60

	State 1	Action 1
A	121		elect, -61. rf = 60/121 = .496. tf = .504
B	0		+(60 x .504)
C	50		
D	9		
exhausted		+(61 x .504)

	State 2	
A	60 (elected)	
B	30.248	
C	50
D	9
exh	30.744

Action 3: recalculate quota based on the number of non-exhausted ballots
149.256 / 3 = 49.752
Action 4: now C has a quota, and is elected. 
The result is {A, C}.

	Does that seem right to everyone? I'm not thinking that the Meek rule
comes into play here. A retains 60 votes as his quota, and the remainder
is multiplied by the transfer fraction and split proportionally (60:61)
between B and exhaustion. In the next round, because some of the vote has
been exhausted, we recalculate the quota using the number of non-exhausted
votes, (180 - 30.744). C now has a quota.
	Even if C had a point or two less, I'm still not sure that Meek and NB
would produce a different result. D would be eliminated and transfer his
votes to C without setting off the Meek rule. The Meek rule would come
into play, for example, if some votes were transferred to the
already-elected candidate A. That doesn't happen in your example at all.

my best,
James

>Hello list members
>Recently I've been having a private discussion regarding the differences
>between the Meek and Newland-Britton methods of counting an election by
>STV.
>With regard to the example:
>A 61
>A>B 60
>C>D 50
>D>C 9
>for 2 seats, I'd be interested in knowing who list members think should
>win and why.
>Newland-Britton elects AB and Meek elects AC.
>David Gamble




More information about the Election-Methods mailing list