[Election-Methods] voting research

[Kathy Dopp]

> ------------------------------
>
> Message: 3
> Date: Sat, 02 Aug 2008 20:35:49 -0700
> From: Bob Richard <lists001 at robertjrichard.com>
> Subject: Re: [Election-Methods] voting research
> To: Election Methods <election-methods at electorama.com>
>

> I suggested that since the simulations showed that IRV was hard to
> manipulate, the usual cases were close to the phase transition where
> things get hard in the average case.

Bob,

The use of simulations to "show" anything is usually looked at with
skepticism among degreed statisticians or mathematicians - perhaps
since each simulation can depend on assumptions which may not be
stated explicitly.  I would think that your paper would be taken more
seriously if it did not entirely depend on simulations and used proofs
or mathematically-derived formulas instead.  To disprove a hypothesis
is of course the easiest, since it merely requires citing one
counter-example.

That said, I have used simulations myself to show that the patterns
produced by the 2004 presidential exit polls in Ohio were not
consistent with the "exit poll response bias" hypothesis that
pollsters claimed produced the discrepancies; and also I have used
simulations to generate counter-examples  (there were countably
infinite counterexamples) to the hypothesis of The Election Science
Institute and several statisticians, including a former President of
the ASA, used to incorrectly claim that the pattern of Ohio's exit
poll discrepancies were inconsistent with vote fraud.  (Vote fraud in
Ohio's 2004 presidential election has been virtually proven now with
concrete ballot evidence, suspicious destruction of evidence that a
court required to be preserved, and other concrete evidence, and is
back in court again.)


>
> Under certainty, with individual voters, manipulation is easy (because
> with the number of candidates given, the number of possible ranked
> ballots turns into a constant).

Again, the number of possible ranked ballots is a *huge* number as the
number of candidates increases.  Please cite my paper which provides
the  exact formula for the possible number of ranked ballots on its
page 6:

where N= the number of candidates, and R= the number of ballot choices
voters are permitted to make (R would be = to N if voters are allowed
to rank all candidates) then

the number of possible unique ranked ballots (assuming partial
rankings are also allowed) is:

the sum from i=0 to i=R-1  N!/(N-R+i)!

Realities Mar Instant Runoff Voting- 17 Flaws and 3 Benefits
June 10, 2008, Version #2– updated June 25, 2008 and Friday, August 01, 2008
By Kathy Dopp, MS Mathematics
http://electionarchive.org/ucvAnalysis/US/RCV-IRV/InstantRunoffVotingFlaws.pdf

It might be interesting for persons reading your dissertation to have
my research report cited as a resource.

Cheers,

Kathy