[EM] Report on the 2006 Burlington Mayoral election.
Abd ul-Rahman Lomax
abd at lomaxdesign.com
Wed Jan 7 18:15:53 PST 2009
I should say at the outset that I will email a spreadsheet with the
vote data, on request. It's on the web site I pointed to, but it's a
pile of individual files and a little nuisance to download and
convert and combine.
At 08:56 AM 1/5/2009, Kristofer Munsterhjelm wrote:
>To my knowledge, Loyal Ploof is (Green).
I should verify that. Ploof claimsed to be a Green, but there are
signs that the party doesn't recognize him. If this is actually an
official Vermont Green web site:
On the home page, we find:
>This Burlington Green Party is a legally established political party
>that is filed with the Vermont Secretary of State. You may confirm
>this fact at 802-828-0771. Please regard all other "official"
>Burlington Green sites by Owen Mulligan, Rene & Jeanne
>Kaczka-Valliere, Annette Urbschat, James Vos, Loyal Ploof, Frank
>Haddleton, Sean Starfighter, Hillary Weeks and Patrick Stanton to be
>a fraudulent and illegal representation of their Lyndon LaRouche
>cult leader Craig Hill from Montpelier.
Wow! Is the "Burlington Green Party" a different party from the
Vermont Greens? Could be! My recollection was that party affiliations
weren't listed for Ploof or the other candidate, they were considered
There is a post made by Ploof to
http://www.iburlington.com/article.php/ploof4mayor where he gives a URL:
It's a dead link, the whole domain bgreens.org is gone.
Whom did the 3809 Progressive Party voters (C03, Bob Kiss) choose in
C01 197 Beaudin
C02 708 Curley, Republican
(1 vote was C02=C05 C01 C04)
C04 2111 Miller, Democrat
C05 252 Ploof
C06 86 Write-in
From this, Ploof doesn't look like a Green, if I assume that the
Vermont Progressive Party is a progressive party! Unless I my memory
is off, the ballot didn't show a party affiliation. It's important,
many voters will vote based on party affiliation. The small vote for
Ploof could easily be explained by voters who saw him call himself a
Green, or who simply liked the crazy name!
>Not all five mayoral candidates were invited to participate --
>Independent Louie "The Cowman" Beaudin claims he had to "twist some
>arms" for a seat at the table. But once the debate begins, its
>inclusiveness works to Miller's advantage. The candidates speak in
>the order in which they're seated, and Miller is sandwiched between
>Beaudin and Independent candidate Loyal Ploof.
Not to be considered a Green, I'd say. The Vermont Green Party
website, http://www.vermontgreenparty.org/, has been hacked,
apparently, currently it is a page in Cyrillic. It is, apparently,
Russian, translating to an article on some kind of garden....
>Fearless Pete's crystal ball has Mrs. Miller finishing first, Kiss
>second and Curley third. But Hinda doesn't break the 50 percent
>threshold on the first round. IRV kicks in, eliminating the two
>"fringe" candidates. No change in the order.
>The Curley supporters, we predict, will ignore their candidate's
>advice and make Hinda, instead of Bob the Prog, their second choice.
>Many, after all, think of her as a Republican deep down inside her
>stock portfolio. That should end it all.
"Fearless Pete" was wrong, Bob won the election. He was right that no
change in the order took place from vote transfers.
Before transfers: Kiss:Miller, 3809:3106, 55.1% Kiss (relative to sum
of Kiss and Miller)
After transfers: 4761:3896 55.0%
This constancy of vote relationships even with transfers is a
phenomenon that I see commonly with nonpartisan elections. It's not
just that plurality would have produced the same result, it is that
the ratio of the frontrunners to each other remains constant. It's
looking to me like the Burlington election was more about the
candidates than about parties, and, given that Burlington is a
relatively small town, that's not surprising at all.
Was "Fearless Frank" right about the Republican voters?
1st 2nd 3rd 4th 5th
2609 votes for C02 (Curley), eliminated, transfers:
exh 768 inc one overvote
C01 286 eliminated, transfers:
C05 88 eliminated, transfers:
C06 10 eliminated, transfers:
C05 103 eliminated, transfers:
C01 50 eliminated, transfers:
C06 1 eliminated, transfers:
C06 57 eliminated, transfers:
C01 12 eliminated, transfers:
C05 7 eliminated, transfers:
C05 3 eliminated, transfers:
C01 3 eliminated, transfers:
Republican votes for the top two:
Rank: 2nd 3rd 4th 5th TOTAL
Progressive: 675 85 69 4 833
Democrat: 720 61 35 3 819
Exhausted 768 136 50 3 958
Overall, the Curley voters did not follow Curley's advice. If we look
at the split, in votes that went to Kiss vs. Miller, it's Kiss 50.4%.
In a nonpartisan election, we'd have expected a 55% ratio to Kiss,
based on the first preference votes and the theory that the
supporters of one candidate in an election, who don't truncate at
that candidate, are typical of all voters as to their preferences
between the remaining ones. Obviously, this breaks down when partisan
affiliations are involved or if voters follow a recommendation from
their favorite or otherwise are atypical.
In this case, it looks like traditional party closeness (Republicans
being closer to Democrats than Progressives, and Miller being called
a Republican light by the Progressives) was a factor, but not a huge
one. The vote still went to Kiss by a small margin.
The Republican voters (i.e., voters preferring the Republican
candidate, they are not necessarily "Republicans") could have turned
the election to Miller if they had voted consistently.
The final result was Kiss, 4761, Miller, 3986. That's a margin of 775
votes. More Republican votes were exhausted than that.
One lesson: political parties and candidates don't own voters. They
can recommend vote transfers, and it may have some effect, but not
necessarily a large one.
>Some more information on this data set, according to my election program:
>Schulze returns C03 > C04 > C02 > C01 > C05 > C06. So does minmax,
>Borda, Vote For and Against (1, 0, 0..., -1), and Nauru Borda.
>Plurality returns C03 > C04 > C02 > C01 > C06 > C05. So does Hare
>(IRV) and Carey.
>Antiplurality returns C03 > C02 = C01 > C04 > C05 > C06.
The only reason that C05 beats C06, I think, is all those voters who
voted with full ranking *excluding* write-ins. C06, though, would be
a multiplicity of candidates, many unique, possibly many not even
intelligible. Nobody bothers to count these more specifically, since,
even if they were all one candidate, that candidate can't win.
*However*, of course, it does happen that write-ins win elections;
when the write-in vote gets high enough, they have to look more closely.
This was not a difficult election, it's pretty straightforward.
That's why all methods produce almost the same result, differing only
down in the difference between the write-ins, considered as one, and
In the data above, you can see that some voters who voted for Kiss or
Miller in *fifth position* had that vote counted. (4 votes for Kiss,
3 votes for Miller). They didn't rank the other frontrunner (which
follows), hence these were 4 Kiss > Miller votes and 3 Miller > Kiss
votes, but they voted indicating that every other candidate, plus a
write-in, were better than the candidate who got their vote in the
last round. 4 of them voted for the winner. Were these votes sincere?
Did they really think that the Cowboy and Loyal Ploof were better
than Kiss? Did they realize that ranking Kiss last (on the ballot)
was still a vote for him?
I see quite a few signs that voters didn't understand the method. One
of the most remarkable phenomenon to me is the overvotes. They mostly
make sense, they probably were intentional. Did the voter know that
these would void that rank and all ranks below, under some
conditions? (If more than one candidate involved in the overvote is
still active when that rank is reached, the ballot is considered
exhausted, as I recall.) So your overvote might count, or it might
not, you don't necessarily know. Most of the overvotes were buried
under frontrunner votes, so they didn't actually cause exhausted
ballots. Overvotes in first position did.
>All positional methods are whole.
>The Condorcet matrix is
> 0 1289 804 1161 2028 3290
>5165 0 3397 3556 5136 5875
>6961 5730 0 4763 7027 7351
>6747 5545 3991 0 6790 7336
>1869 1318 603 987 0 3094
> 431 311 194 323 510 0
>(row beats column), and the WV basis is thus:
> 0 0 0 0 2028 3290
>5165 0 0 0 5136 5875
>6961 5730 0 4763 7027 7351
>6747 5545 0 0 6790 7336
> 0 0 0 0 0 3094
> 0 0 0 0 0 0
>which means that C03 is the CW.
>(If my program has bugs, there will be errors in the above.)
Now, let's look at San Francisco!
More information about the Election-Methods