First-Choice Winner Standard

Bart Ingles bartman at
Thu Oct 8 17:45:46 PDT 1998

New Democracy wrote:
> Dear Bart,
> You wrote: "... the Approval count is ignored (weight factor = 0) whenever
> there is a first-choice majority."
> Donald: You cannot escape that easy. All methods must face the First-Choice
> Winner Standard. The defeat of a first-choice winner is a valid test of any
> election method. You are not allowed to restrict the debate by saying that
> the test does not apply to some method.

We should start by clarifying a couple of terms:
  (a) first-choice majority, and
  (b) first-choice winner (i.e. non-majority, plurality winner)

>       DEMOREP is correct when he said: "Approval could defeat a
> first-choice winner."

While the statement is obviously correct, it was actually my
mis-statement of DEMOREP's objection, and should have been:  "Approval
could defeate a first-choice _majority_ winner."

My method would have preserved a first-choice majority winner, but would
have allowed the Approval component of the count to defeat a
first-choice [non-majority] winner.  This is, after all, the purpose of
using methods other than FPP.

> You also wrote: " I doubt that Approval would fail to yield a majority very
> often..."
> Donald: It will all depend on which defination of majority you will be
> using. If you are using Mike O's defination, your statement will become a
> false reality. In fact Apporval should yield two or three candidates with
> "false majorities".
>      But if the majority is based on the number of choices used, then
> Approval will fail to yield a majority most of the times.

I meant majority as in "I doubt that Approval would often fail to yield
a candidate for whom a majority of voters voted YES".  Although now that
I think about it, I'm not too sure how true even that statement is.  For
example, in a somewhat realistic example, you might have:

(Approval vote -- no preference order implied)

37%  Clinton
6%   Perot, Clinton, H.Browne
5%   Perot, Dole, H.Browne
6%   Perot, H.Browne
5%   H.Browne, Clinton, Perot
3%   H.Browne, Dole, Perot
38%  Dole


48% Clinton
46% Dole
25% Perot
25% Harry Browne

In other words, I would expect to see a fair amount of truncation by
major candidates who have a good chance of winning, producing a result
similar to IRO (or any other method but FPP).

>      DEMOREP is correct a second time. He wrote: "Approval doesn't
> guarantee that the winner will have a majority." [...]

We all agree to that.  Neither does any other method.  Not unless you
force the voters to rank every choice, whether they do it randomly or
-- Bart

More information about the Election-Methods mailing list