[EM] proportional vote - proportional term

[Jonathan Lundell]

At 4:08 PM +0000 4/3/07, mcgratsc at comcast.net wrote:
>This is a how-to question, less of a philosophical or policy 
>question, i hope I've found the right group, apologies if not.

Maybe, maybe not. I don't think it hurts to discuss the 
practicalities of using our methods, though.

>I run a small non-profit org whose Board has extolled the virtue of 
>STV and related proportional voting processes as superior.  I must 
>follow that guidance in the upcoming election of a small steering 
>committee, as a test of using this for the larger Board elections 
>next year.
>
>My goal is to elect a 5 person committee, the three with the most 
>support will win a 2 year term, the balance (2 ) would win a 1 year 
>seat.  This split will allow future years to re-elect a portion of 
>the panel.
>
>I have OpenSTV as my (current) desktop ballot counter but am open to 
>other solutions. 
>
>I expect 20-30 votes to pick 5 of the 7 candidates running.  While 
>this small number may make me lean towards a human-powered 
>calculation, I'd prefer something more mechanical, especially 
>because this new process will likely roll out to the bigger Board 
>elections with hundreds of ballots cast.

There are STV methods that are amenable to hand counting, but the 
distortions they introduce are especially intrusive in smaller 
elections. I'd use OpenSTV (but then I work on it, so...).

Speaking of methods, my #1 piece of advice is to be sure you've 
specified your STV method in advance. Different STV methods can yield 
different results, especially in close elections, and especially in 
small elections. The exact method you choose is less important than 
choosing *something*--before the election.

I agree with Chris that STV is of limited value in a two-seat 
election. I understand the motivation for staggered terms, but it 
does tend to work against the interests of PR.

I'd also argue that STV methods are of little or no use in 
identifying "the tree with the most support", as STV methods don't 
really rank their winners ("most support" is a fuzzy concept at best 
in many situations, and to equate "elected earlier" with "more 
support" is in general a mistake).

If I had to do what you're proposing, I think I'd count the election 
once for three seats, and then count the same ballots a second time 
for two seats, after withdrawing the three winners from the first 
round.

Beyond that, I'd either abandon staggered terms, or expand the 
committee to six members.
-- 
/Jonathan Lundell.