That's interesting. I wish we had more of these kinds of experiments with
real live people to help educate our intuition.
On Fri, 2 Feb 2001, LAYTON Craig wrote:
> Yes, the first one.
>
> In relation to Borda, I remember an experiment we did in 1st year political
> science, where all the students voted on the same thing (I think it was
> chocolate bars). We used a number of voting systems, and had to vote
> sincerely and consistently across all the systems. The voting systems were;
> first past the post, IRV, approval, borda and a rating system (out of 100).
> The voting pattern was really interesting (there were two dominant factions,
> one larger but more ambivalent, and the other smaller and more committed to
> their candidate, and the members of a faction voted almost exactly the same
> way, without prior communication on how to vote), but, anyway, all the
> systems produced nearly the same result, except for borda, which produced a
> wildly different result (out of six candidates, the candidate who won the
> borda count came third in the rating system, and did even worse in some of
> the others).
>
> I thought it was quite interesting.
>
> -----Original Message-----
> From: Forest Simmons [mailto:[EMAIL PROTECTED]]
> Sent: Thursday, 1 February 2001 14:58
> To: [EMAIL PROTECTED]
> Subject: Re:Borda Count
>
> Suppose you have 16 candidates to rank. You know how each of them stands
> on the four issues that you consider vital. No two have the same profile
> on these issues, so if we represent "agrees with you" and "disagrees with
> you" by the letters a and d respectively, the 16 candidates can be
> identified by their profiles: aaaa, aaad, aada, aadd, adaa, adad, adda,
> addd, daaa, daad, dada, dadd, ddaa, ddad, ddda, dddd
>
> In an informal non-binding poll you are asked to rate them on a scale of
> zero to 100%, so naturally you rate them in proportion to the number of
> issues on which they agree with you (assuming all of the issues are
> equally important to you).
>
> aaaa gets 100%
> addd, dadd, ddad, ddda get identical ratings of 75%
> aadd, adad, adda, daad, dada, ddaa get identical ratings of 50%
> daaa, adaa, aada, aaad get identiacl ratings of 25%
> dddd gets 0% .
>
> Next, in another informal non-binding poll you are asked to rank the
> candidates.
>
> Since you cannot distinguish all of them on the issues, you use looks and
> personality to break up the groups with identical ratings:
>
> aaaa > aaad > aada > ... > dddd
>
> The second pollster immediately converts your rankings to a rating via the
> Borda Count with rates between 0/15 and 15/15.
>
> Which would you consider to be a more accurate representation of your
> estimation of the candidates' abilities to represent your viewpoint in
> the legislature?
>
> Forest
>
>