Adjusted Majority Preference Voting for PR?

[Gary Swing]

In a sketchy three page article published in the book Choosing An
Electoral System (Arend Lijphart & Bernard Grofman, eds.), Steven Brams
and Peter Fishburn suggest a PR system based on the use of single-member
districts with additional members added on from among the "best losers" to
achieve proportionality by party. They call this proposal "adjusted
district voting." They also refer to The Report of the Hansard Society
Commission on Electoral Reform (London: Hansard Society for Parliamentary
Government, 1976). I haven't seen the Hansard Report, but they indicate
that it recommended a similar system. 

Brams and Fishburn suggested that under adjusted district voting, the
first place candidate in each district would win that district seat, but
more members would be added by tallying up all of the votes cast for
candidates of each party, and setting the overall composition of the house
proportionate to each party's share of all votes cast. This would require
using a variable sized legislature. I'm not sure what their seat
allocation formula would be or exactly how it would work.

While I feel that this system would be inferior to other forms of PR, I
would like to hash out some details for "Adjusted Majority Preference
Voting" (AMPV) as a "conservative" PR option requiring only limited
changes to existing election systems.

Here's my take, inspired by Brams/Fishburn and DEMOREP's suggestions
below, from a previous thread:

1) Assume a variable-sized legislative body with 50 single-member
districts.

2) Each district member is elected by majority preference (ranked choice)
voting. 

3) Additional members are added on, based on first preference votes, to
achieve proportionality by party (or at least to reduce distortions). 

4) Each party would ideally be allotted a number of seats equal to Total
First Preference Party Votes x Total Seats (or Total District
Seats?)/Total Votes (TFPV X TS/TV).

5) Additional members allotted to each party would be seated in descending
order of first preference votes cast for that party's district candidates
until all the party's seats are filled.

6) An independent could hypothetically win an additional member seat by
receiving a number of votes equal to at least Total Votes/Total Seats
(which seems nearly impossible to me if the independent didn't win the
district vote). Alternatively, independent candidates could combine their
votes into an independent slate.

7) No candidate who wins the majority preference vote in their district
shall be denied that district seat.

Is this workable? Assuming that the two major parties win all district
seats, it seems to me that the largest party would always have overhang
seats if the seat allocation formula were based on the number of district
seats. If the seat allocation formula were to be based on the total number
of seats (districts + add-ons), how do you determine the total number of
seats prior to allotting the additional seats?

If this scheme is workable, it seems to me that majority preference voting
would be preferable to single member plurality voting because it
reduces the incentive for strategic voting and encourages sincere
voting. This facilitates a proportional distribution based on the voter's
true first choice preference, rather than a forced choice between two
"viable" parties within districts.

For third parties, I think this would still be a difficult system to use,
because it would require a smaller party to run a very large number of
district candidates to win a very small number of seats. It can also
benefit third party candidates running in districts with few candidates
over those running in districts with many candidates. For example, if a
Libertarian candidate in a two way race gets 20% of the vote, that
candidate would be selected for an additional seat before another
Libertarian candidate receiving 10% of the vote from a district with five
candidates. This may not reflect true support for said candidates.

 On Sat, 18 Apr 1998, DEMOREP1 wrote:

> Another alternative is just to have single member districts (SMD).
> 
> An independent would win by getting a number of votes equal to at least Total
> Votes/Total Seats (TV/TS) (somewhat unlikely)
> 
> A party would get seats equal to Total Party Votes x Total Seats/ Total Votes
> (TPV x TS/TV).
> 
> The top vote getters of a party in all districts would get the party's seats. 
> 
> Some districts might not elect any legislator.  

I think this is a serious problem with DEMOREP's design. If the voters in
a district see that the candidate with the most votes in their district
did not get elected, it could undermine voter confidence in the system. I
believe most voters would expect a district vote to elect a district
candidate.

> Other districts might elect 2
> or more legislators (which is irrelevant since the voters in all districts
> would get representation).  Each party would have an incentive to have
> candidates in each district (unlike the current situation with plurality
> winners in SMD).
 
 
-----------------------------------------------------------------------
Gary Swing for US Senate '98  (US Pacifist Party, Colorado)
Campaign Web Site:  http://www.geocities.com/CapitolHill/4415
US Pacifist Party: http://www.geocities.com/CapitolHill/Lobby/4826

"Today the choice is no longer between violence and nonviolence. It is
either nonviolence or nonexistence."
                                               Martin Luther King, Jr.
-----------------------------------------------------------------------