(Fwd) Re: Single-winner ranked ballot voting methods

[Steve Eppley]

I asked Gary Cox of UCSD for info related to spatial (or whatever)
analysis of ranked ballot methods.  Here's his reply.

--Steve

------- Forwarded Message Follows -------
From:          Gary Cox <gcox at weber.ucsd.edu>
Subject:       Re: Single-winner ranked ballot voting methods
To:            seppley at alumni.caltech.edu (Steve Eppley)
Date:          Tue, 29 Oct 1996 08:46:00 -0800 (PST)

> In my opinion MPV wouldn't be a significant reform, since potential
> candidates still have the "spoiler" dilemma when deciding whether or
> not to run, and the voters still have the "lesser of evils" dilemma
> when ranking the candidates.  These dilemmas would act to deter 
> candidates from running, inflate the perceived support of the big 
> two, and marginalize the other candidates and their policies.  
> 
> For example, in a 3-candidate race between Clinton, Dole, and Nader,
> given these sincere voter preferences
>    46%:  Dole > Clinton > Nader
>    20%:  Clinton > Dole=Nader
>    34%:  Nader > Clinton > Dole
> if the voters all vote sincerely then Condorcet winner Clinton would 
> lose, Dole would win, and Nader's supporters would regret voting 
> sincerely instead of for "lesser evil" Clinton.  CV&D considers this
> kind of scenario to be unlikely, but I'm skeptical of their opinion
> and I'm looking for some solid research on this question in hopes of 
> changing some minds at CV&D.  (A very simple model shows that wing
> candidates would tend to espouse policies which would squeeze away
> voters' highest preferences from the center candidate, thereby
> defeating the center if MPV is used.)
> 
> While hunting for a basic analysis of how voters, candidates, and
> potential candidates would behave given various ranked ballot
> methods and various distributions of voters' preferences, and how
> likely or unlikely those distributions are, I noticed in _Game Theory
> and Political Theory_ that Ordeshook references your analysis of
> multicandidate single-issue plurality rule elections.  Are you also
> familiar with similar work for elections which use ranked ballots? 
> If so, perhaps you can help me find the info I'm hunting for. 
> Thanks in advance for any tips you can offer.

My work in the Amer Jour of Pol Sci covers so-called 'scoring rules'
-- systems that use ranked ballots -- as well as Condorcet completion
procedures, approval voting and other systems.  It is confined to
analyzing candidate strategies.  There are two pieces, one in 1987
and one in 1990.  I think Ordeshook cites both.

Michael Dummett has a 1984 book on Voting Procedures that I think
handles some strategic aspects of STV from the voter's perspective.  
He also has a book ms (still unpublished or maybe just out) that 
deals with STV.

There are also various simulation studies of Condorcet efficiency
(how frequently will a given procedure select a Condorcet winner, if
one exists) by e.g. Samuel Merrill III.

Cheers,
 
Gary W. Cox				          Dylan's Dad Can Be Reached At:	
Professor, Political Science		     PHONE:     619-534-1428	
University of California, San Diego	FAX:       619-534-7130
La Jolla, CA 92093-0521                 INTERNET:  gcox at weber.ucsd.edu 

Director, Lijphart Elections Archive 
The LEA is a web-accessible archive covering elections worldwide which 
you can find at http://dodgson.ucsd.edu/lij/.